Volume 80 July-August 2000 Preface Contents R. Gautham, A new bidirectional topology for electric vehicles Prabindh Sundareson and L. Umanand Debiprasad Panda and V. Rammarayanan Biju S. Nathan and V. Rammarayanan IISc Theses Abstracts Switched junctions in bondgraph for modelling power electronic systems Modelling and simulation of switched reluctance motor drive Designing for zero-voltage switching in phase- modulated series resonant converters Number 4 Book Reviews
Transcript
Volume 80 July-August 2000
Preface
Contents
R. Gautham, A new bidirectional topology for electric vehicles Prabindh Sundareson and L. Umanand
Debiprasad Panda and V. Rammarayanan
Biju S. Nathan and V. Rammarayanan
IISc Theses Abstracts
Switched junctions in bondgraph for modelling power electronic systems
Modelling and simulation of switched reluctance motor drive
Designing for zero-voltage switching in phase- modulated series resonant converters
Number 4
Book Reviews
Power electronics has entered the curriculum of undergraduate electrical engineering as a sub- ject over two decades ago. During the past decade, several universities have introduced post- oraduate degrees with specialisation in power electronics. A number of Indian industries have b
been introducing into the market place an ever-increasing number of power electronics prod- ucts. The growth of academic research and industrial development in this area has been sub- stantial. Two special issues of the Jowml of the ind im imtitute qf Science have been planned to present the diverse development and research work that is going on at the Indian Institute of Science, Bangalore, in the area of power electronics.
Power electronics is becoming an enabling technology in several disciplines. The paper en- titled "A new bidirectional topology for electric vehicles" by R. Gautam et ai. covers the appli- cation of inverters to electric vehicle drives. Another paper by L. Umanand extends the appli- cation of bondgraph theory to power electronics systems.
Switched reluctance motors are attracting more attention recently as an industrial drive, on account of several positive features. The paper entitled "Modelling and simulation of switched reluctance motor drive" by Debiprasad Panda and V. Rammarayanan covers the extensive work done in this area.
As the switching frequencies are pushed to higher and higher limits, lossless resonant switching applications are on the rise. A paper by Biju S. Nathan and V. Ramanarayanan out- lines the design methodology for series resonant converters featuring zero-voltage switching.
The availability of low-cost digital hardware has made possible several efficient modula- tion techniques for inverters employed. The paper entitled "Triangle comparison and space vector niproaches to pulsewidth modulation in inverter fed drives" by G. Narayanan and V. T. Ranganathan covers novel modulation techniques and their effectiveness at high power.
The paper "A modified area-product method for the design of transformer and inductor" by G. S. Ra~nanarnurthy and V. Ramanarayanan covers computer-aided design of electromagnetic elements in power electronics systems. The iterative approach claims better utilisation of the materials employed in the electromagnetic elements.
" Rotor side control of grid-connected wound rotor induction machine" by Rajib Datta and w
V. T. Ranganathan breaks fresh ground for wind-energy applications method leading to better utilisation of the machine, power converter wind potential.
Modem power devices require close protection in order to exploit application. A paper by G. Narayanan et al. covers the short-circuit vices.
through a novel control as well as the available
their full features in any protection of IGBT de-
We thank Prof. M. S. Shaila for giving us an opportunity to guest-edit these issues. Thanks are also due to Mr. N. M. Malwad, Mr. K. Sseenivasa Rao and Ms R. Geetha for their help in processing the manuscripts, coordinating the reviews and the editing and printing process.
Department of Electrical Engineering Indian Institute of Science Bangalore 560 012
PADIYAR, K. R. RAMANARAYANAN, V. RANGANATHAN, V. T. Guest Editors
Volume 80 July-August 2000 Number 4
R. Gautham, A new bidirectional topology for electric vehicles Prabindh Sundareson and L. Urnanand
L. Umanand
Debiprasad Panda and V. Ramanaray anan
Biju S. Nathan and V. Ra~nanaray anan
Volume 80
Switched junctions in bondgraph for modelling power electronic systems
Modelling and simulation of switched reluctance motor drive
Designing for zero-voltage switching in phase- modulated series resonant converters
September-October 2000
G. Narayanan and Triangle comparison and space vector approaches to V. T. Rangmathan pulsewidth modulation in inverter-fed drives
G. S. Ramana Murthy and A modified area-product method for the design of V. Ramanarayanan inductors and transformers
Rajib Datta and Rotor side control of grid-connected wound rotor V. T. Rangmathan induction machine
G. Narayanan, Protection of insulated gate bipolar transistors S. R. Muralidham, against short circuit A. S. Anand and V. Ramanarayanan
J. hdiart I m t . Sci., July-Aug. 2000, 80, 319-326. O lndian Institute of Science
R. GAUTHAM, PRAB~NDH SUNDARESON AND L. UMANAND Centre for Electronics Design and Technology, Indian Institute of Science, Bangalore 560 012. email: [email protected] Phone: 9 1-80-36008 10; Fax: 9 1 -80-3600808
Received on February 2, 2000.
Abstract
This paper describes the development and testing of a novel power converter topology which can he used in systems requiring a high-voltage gain, with features like less stringent requirements on the DC filter capacitor and bidirectional power flow capabilities. This converter is of a modified dual flyback topology and is used as the main converter in ;In
electric two-wheeler (ETW). The advantages, important parameters for the working of the converter and inlpleinenta- tion issues are discussed.
Historically, the evolution of power converter topologies has followed that of the development of new power devices.'" These configurations have tried to utilise improved characteristics of power devices to achieve higher power density, higher efficiencies of operation and robust- ness. One important aspect of these developments is to reduce the filtering requirements by switching at higher frequencies. However, until now, a converter topology that can combine a high-voltage gain with less stringent capacitor requirements has not yet been proven. Research on electric vehicles has broadly taken two paths till now; one along the 'hybrid EV' (HEV)? and the other along the 'electrical only' path. A hybrid EV contains a gasoline engine in addition to the electric drive, whereas in 'electrical only' drives, the drive is provided by electricity alone. Various power electronic configurations exist for these drives and induction motor-powered drives have been the focus o f research recently. This is mainly because of the higher ruggedness of induction motors. In applications like these, the battery and the in- verter need to be interfaced with a high-gain DC-DC converter that provides a high DC link voltage. The size of magnetics can be reduced by switching at high frequencies. However, due to large RMS currents (caused by switching) flowing through the capacitor, the loss due to ESR of the capacitors becomes significant. As a result, the designer needs to add more capaci- tors and the capacitors become bulky and costly. Further, the DC-DC converter silould allow bidirectional power flow to accommodate regenerative braking. This paper addresses these issues.
Section 2 addresses the problems encountered in conventional boost and flyback topologies and discusses the new bidirectional topology, which is well suited for electric vehicle applica- tions.
Switch drive signal
0
FIG. I . Buust converter. FIG. 2. Boost converter cunent ( I , ) .
otivation for the
In the boost configuration, shown in Fig. 1, the energy stored in the inductor is used to charge the capacitor to a voltage higher than the input voltage.
Equation (1) gives the output voltage, where D is the duty cycle of operation:
The wavcform in Fig. 2 shows the steady-state capacitor current wave shape when the tmn- sistor Q is switched on and off.
In our application, the input and ourput voltages are 43 and 380 V, respecdvely . The duty cycle from eqn (1) can be calculated to be D = 0.875. The RMS current through the capacitor can be calculated from eqn (2):'
where
tl = duration o f OFF period of the switch and t 2 = total duration of a period of switching. With the above equations, the RMS current through the output capacitor can be calculated at a switching frequency of 20 kHz. The other parameters are:
With these parameters, the RMS current is found to be 13.76 A.
With this configuration, the RMS current through the capacitor is found to be high. This current causes heating in the capacitor due to ESR of the capacitor. As a consequence, the cost and size of the converter becomes prohibitively high. Further, the boost configuration can pump power in only one direction.
In the isolated nonregenerative flyback converter (Fig. 3), the transformer stores the energy when the switch is turned on. This energy is transferred to the output capacitor when the switch is turned off. The wave shape of the capacitor current is similar to that in the boost converter topology (assuming turns ratio of unity). Here, the transformer actually acts as a 'store house' of energy.
In Fig. 4, which is the regenerative flyback configuration, reverse power flow is possible by turning on the switches Q O and Q 1 alternately, depending on the power flow direction. In this configuration, high voltage gain i s possible by adjusting the turns ratio of the primary and sec- ondary. Also, bidirectional power flow is possible, and there is isolation between the primary and the secondary.
In a manner similar to that in the boost converter, the high RMS current causes heating in the capacitor due to ESR, and the input current drawn is not pure DC.
The new converter is bidirectional. In the forward direction (battery to DC link) of power flow, the converter boosts the DC link voltage. In the reverse direction (DC link to battery) it bucks the DC voltage and charges the battery. Some of the main considerations for choosing the con- figuration are:
Reduced RMS current o f the DC link capacitor Bidirectional power flow Galvanic isolation Transformer design for fixed 50% duty cycle Simplicity of controls for the converter
R + GAUTHAM et 01.
- drive 3 signals
FK;. 5. Principle of the main converter.
If two flyback converters are used in pa~allel with the switches operated in complementary fashion, then the RMS current in DC link capacitor can be rninimised to less than 2 A. The schematic of the main converter is shown in Fig. 5. The hl on the primary side is 6.25 A. The dl on the secondary side is 6.25/8 = 0.78 12 A, where 8 is the voltage gain between primary and secondary. Now the RMS current through the capacitor i s (ref. Appendix 1 ).
which is much less than 2% of the capacitor current in the case of boost converter. This will reduce the size and cost of the DC link capacitor. Besides, the size and weight of the magnetics is a minimum. In this configuration the secondary side also needs a switch like IGBT or MOSFET. This will enable power flow in both directions. If power flow in the reverse direc- tion is desired then the switching signal will have to be steered to the secondary side. Driving the IGBTs/MOSFETs of the converter from the control card will not be a problem as no extra isolation i s required. The size of the main converter can be further reduced by the usage of high flux density (1.5 T) c u m such as METGLAS cores (AlliedSignal). Considerillg the above advantages, this configuration has been chosen for the main converter.
The switches QO and Q1 are operated in complementary fashion with a duty cycle of 50%. When a switch is turned ON, the inductor connected to the respective switch stores the
BIDIRECTIONAL TOPOLOGY FOR ELECTRIC VEHICLES
FIG. 6. Current waveform in the secondary winding for single converter (secolldary load cu~rent of I A).
FIG. 7. The sum of the currents in the secondary wind- ings of both the converters. Scale: 1 AmpfDiv.
energy. When the switch is turned OFF, the stored energy is passed on to the DC link through the diodes in Q2 and 43. It can be seen that the capacitor current waveform has very less rip- ple. When the switches QO, Ql are turned off and converter PWM pulses are steered to switches Q2 and Q3, power flow takes place in the reverse direction. In this mode, QO and QI act as diodes and charge the battery.
For the capacitor current shown in Fig. 4, it can be shown that the RMS current through the capacitor is given by the relation (~1412) (refer to Appendix 1).
The converter designed as per the above concept was tested in the laboratory. Figure 6 shows the current in the secondary winding for a single converter. It can be seen that the RMS current through the capacitor is large.
This will cause heating in the capacitor due to the ESR of the capacitor. Figure 7 shows the sum of currents in the secondary windings of both the converters. It can be seen that the wave- form is a sum of two identical but conlplementary waveforms of Fig, 6 . It can be seen clearly
FIG. 8. Primary winding current for a single converter (load current of I A). Scale: 5 AmpfDiv.
. . . . . . . . . . . . . . .
..... * , . I .... I . .
FIG. 9. Current through the primary snubber diode (for a secondary load current is 1 A). Scale: 5 Amp/Div.
324 R. GAUTHAM et al.
that the RMS current of the waveform is very less compared to that in Fig. 6. This current waveform through the DC link capacitor will cause less heating in the capacitors. The spikes and ringing in the current waveform are again due to the reverse recovery time of the diode and the connecting lead inductance. In the ideal case the spikes will be absent.
Figure 8 shows the primary winding current for a single converter. The current was meas- ured using a Hall effect current probe. The current spike during turn-on is due to the slow turn- off of the secondary diode. When the primary MOSFET is on, the current through the winding should be almost linear. Linearity in this region indicates that the core has not saturated. If the core saturates then the current waveform will have a curved shape with a peak at the turn-off, This will cause high currents through the MOSFETs, which may damage the MOSFET.
Figure 9 shows the current through the primary snubber diode. This diode will conduct only during turn-off of the primary MOSFETs. In this case, the peak current at full load i s 40 A. Figure 10 shows the input current of a single converter with a 100 p,F capacitor con- nected close to the input terminals of the converter. The capacitor should be of low ESR type. Otherwise the capacitor will heat up due to high RMS currents through it. In the absence of the capacitor at the input terminals there i s ringing in the waveforms.
Figure 11 shows the secondary switch wavefom. This was recorded using a 1000 : 1 Tek- tronix probe. The initial spikes, during turn-off, are due to the leakage inductance of the coils. The action of the snubber can be clearly seen during the turn-off as the energy stored in the leakage inductance is dissipated in the snubber in the form of heat. In the absence of the snub- ber the secondary switch is easily damaged. A similar logic can be applied to the primary side. The primary switch waveform i s shown in Fig. 12. The input current with both the converters operating in parallel is shown in Fig, 13. The input current is almost a pure DC but for some small glitches. This i s because of the small dead time between the two converters.
This paper introduces a new type of bidirectional switcl~ing co~~verter topology, which has the feature of reduced ripple current through the outout filter capacitor, and has reverse power flow
FIG. 10. Input current for a single converter with capaci- tor across the input lines (for a secondary load of 2 A). Scale: SArnpDiv.
FIG. 1 1. Secondary MOSFET voltage waveform (at output voltage of 320 V DC).
BIDIRECTIONAL TOPOLOGY FOR ELECTRlC VEHICLES
FIG. 17. Primary MOSFET voltage wavefcwn. FIG. 13. 1. Input current drawn from the battery (both converters in parallel). II. Primary MOSFET voltage ol' one of the converters.
capabilities, along with high voltage gain. The advantages gained by using this configuration are :
Reduced RMS current of the DC link capacitor Bidirectional power flow Galvanic isolation
r Transformer design for fixed 50% duty cycle Simplicity of controls for the converter
It has been shown that if two flyback converters are used in parallel with the switclm operated in complementary fashion, then the RMS current in DC link capacitor can be minimised to less than 2 A. The results have been recorded and discussed in detail and compared with other topologies.
2. NED MOIIAN, UNDELANB, T. M. AND
ROBBINS, W. P.
An overview of power electronics in electric vehicles, IEEE Trmu., 1997,13-44,3-13.
Power ekctranic converters, r~pplicu~ions ond design, Wiley, 1995.
Electrical Power Research Institutes. www.epri.com.
Electric motor bike from EMB, ww w .electric-motorbike.com.
Referring to Fig. 5 , the capacitor current waveform i, is shown in the figure below. The RMS value of this waveform is derived as follows:
ix 2: List of sy
V, : DC input voltage, Vo: DC output voltage, ESR: equivalent series resistance, T,: switching period, D: duty cycle of operation, T: instantaneous time, iloaci: current through load,
NL: ripple current in inductor current.
J. Indim Inst. Sci., July-Aug. 2000,80, 327-332. O Indian Institute of Science
L. UMANAND Centre for EElectronics Design Technology (CEDT), Indian Institute of Science, Bangalore 560 01 2. en1ai1: [email protected]~in
Received on January 20,2000.
Abstract
Modelling of switched-mode power electronic systems using bondgraph is difficult because of lack of switch element in the bondgraph. To overcome this problem, we propose the concept of switched junctions. Were, the dominant junc- tion variable is switched to emulate the practical switch. A few switched-mode power converters are modelled and simulated to demonstrate the efficacy of using the switched junctions.
Keywords: Switched junctions, bondgraphs, modelling, power electronics.
Bondgraph is a system modelling technique. It has been used extensively in modelling inter- connected interacting physical systems.'-3 Bondgraphs not only allow the modelling of systems for analysis and simulation, but also are a powerful tool for automatic computer programming. Bondgraphs were invented by Henry Paynter to overcome the inherent drawbacks of block diagrams for servo controls and simulation problems. They consider both energy and signal exchanges between the components of a system. For systems where power and efficiency play a major role, bondgraph modelling method is very convenient.
Bondgraphs have been used primarily for modelling dynamical systems where there is con- tinuous energy or power flow through all the components of the system. They do not inher- ently support modelling of switched power systems like switched-mode power converters and inverters. To overcome this problem, two approaches have generally been used: (i) macro- modelling the power switches using inductances, capacitances and resistance^;^ and (ii) using modulated transformers and gyrators with modulation index being either 1 or 0, depending on the state of the power switch? The first approach poses no problem in modeIling power elec- tronic systems wherein the components are operating in the linear region. However, for switched power electronic systems, the time constants of the macromodels of the power switch will be in the order of ps and the simulation time will be in the order of seconds. This will make the system a very stiff one and lead to simulations which may take days to finish. In the second approach, wherein modulated transformers or gyrators may be used to represent power switches, the very nature of the effort-flow relationship may get altered. This approach con- trols either the effort or the flow of the power switch. As a consequence, the model is not a correct representation of the physical system in most cases.
In order to overcome the above-stated problems, this paper proposes an extension to the conventional bondgraph junction. In this paper, the concept of a switched junction is proposed which is a generalization of the conventional bondgraph junction. Section 2 lays the foundation for the concept of switched junctions. Section 3 elucidates the application of switched junc- tions to modelling of switched-mode power converters.
The 0- and 1-junctions are the multiports used in bondgraph. They are defined according to ~ h o m a ' , and Karnopp and IZosenberg3. The 0-junction is defined as
f, = 0, where& is the flow in the kth bond of the 0-junction, and k
ek = eoj, 'dk, where eo, is the 0-junction effort. (def. 1)
There is only one flow causal bond at the junction which will determine the junction effort, goy
The l -junction is defined as
e, = 0, where ek is the effort in the lcth bond of the 1 -junction, and F;
fk = j i f;i, Vk, where J i j is the 1 -junction flow. (def. 2)
There is only one effort causal bond at the junction which will determine the junction flow, jh.
The 0- and 1-junctions, as defined by def. 1 and def. 2, respectively, are for continuous power flow through the system. To handle switched power tlow through the system, def. 1 and def. 2 are extended to the switched junctions. The mvirclwd 0-junction is defined as
xfi = O , where ji is the flow in the kth bond of the 0-junction, and k
ex. = eo?j, 'dk, where cob is the 0-junction effort.
If ej and.$ are the effort and tlow, respectively, of the ith flow causal bond of the switched 0- junction and ni is the causal switch of the it11 flow causal bond, then
0 i # n z d i . CI,, = , and
1 i = n
Note that def. 3 reduces to def. 1 when there i s only one flow causal bond at the 0-junction.
The nuitched l -juaction is defined as
e, = 0, where e k i s the effort in the kth bond of the 0-junction, and k
MODELLING OF POWER ELECTRONIC SYSTEMS
FIG. 1 . Brick converter. FIG. 2 . Bondgraph nod el for buck converter.
ji =Aj, V k , he ref!^ is the 1-junction flow.
If ei and j i are the effort and flow, respectively, of the ith effort causal bond of the switched 1- junction and ui i s the causal switch of the ith effort causal bond, then
0 i f n LC, 1.1, = , , and
1 1 = n
(def. 4)
Note that def. 4 reduces to def. 2 when there is only one effort causal bond at the 1-junction.
The concept of switched junctions explicated in Section 2 i s applied for the switched-mode power converters in this section. The bondgraph models for the buck, boost, buck-boost and the full-bridge converter configuration are illustrated in Fig. 1 through 8. Figure 1 shows the buck converter configuration. The corresponding bondgraph model is shown in Fig. 2. The 0'- junctio~l is a switched junction. Its effort is given according to def. 3 as
e,, = ui [ul u = u during DT and u, = ui d ~ n n g (1 - D)T.
where D is the duty cycle and T the switching period of the switch. Therefore, during the period DT, the junction effort is Vin and is 0 during (1 - D)T. The flow in the bonds attached to 0'-junction is determined by the 12-junction. The remaining portion of the bondgraph has conventional meaning as given in ~homa, ' and Karnopp and ~ o s e n b e r g ~
Figure 3 shows the boost converter and the corresponding bondgraph model i s shown in Fig. 4. The 02-junction is a switched junction. Its effort i s given according to def. 3 as
L. UMANAND
FIG. 4. Bondgraph model for boost converter.
% = ui b r , . ] Lii = during DT and ui = i i2 during ( I - D)T
During the period DT, the 0'-junction effort is 0 and during (1 - D)T, it is Vo. The flow in the bonds attached to 0'-junction is determined by the 1'-junction. The remaining portion of the bondgraph has conventional meaning as given in ~horna , ' and Karnopp and ~ o s e n b e r ~ . ~
Figure 5 shows the buck-boost converter and the corresponding bondgraph model is shown in Fig. 6. The 0'-junction is a switched junction. Its effort is given according to def. 3 3s
During the pel-iod DT, the 0'-junction effort is Vin and during (1 - D)T, it is -Vo. The flow in the bonds attached to 0'-juncticn is determined by the inductor L. The remaining portion of the bondgraph has conventional meaning as given in ~ h o m a . ' and Karnopp and ~osenberg-."
Figure 7 shows the full-bridge converter and its corresponding bondgraph model is shown in Fig. 8. The 02- and O ' - ~ U ~ I C ~ ~ O I I S are switched junctions. Their efforts are given according to def. 3 as
ui = ~1 during DT and Hi = 14, during ( I - D)T. ( 5 )
FIG. 5. Buck-boost converier. Fro. 6. Bondgraph model for buck-boost converter.
MODELLING OF POWER ELECTRONIC SYSTEMS
to load
FIG. 7. Full-bridge converter. FIG. 8. Bondgraph model for full-bridge converter.
Equations (4) and ( 5 ) define the efforts at the two switched junctions. The flow in the bonds attached to switched junctions is determined by the 14-junction.
The circuits discussed in the previous sections have been nod el led in bondgraph and simulated in the MATLAB/SIMULINK environment. A bondgraph toolbox has been developed which is used for simulation of all bondgsaph-modelled systems. Figure 9(a) shows the bondgraph model of the buck converter and Fig. 9(b) the waveforms probed at the output which show the output voltage and the capacitor current. In Fig. 9(a), the 0s represents the switched junc- tion. The input voltage applied is 10 V and a duty ratio of 50% is used. The output voltage is
'2-1 Generator
NON-ISOLATED BUCK CONVERTER
FIG. 9(a). Bondgraph model of buck converter and (b) simulation results of output voltage and capacitor current.
seen to settle to 5 V. Both the transient and steady-state responses are as per theoretical calcu- lations.
Modelling switched-mode power electronic systems using bondgraph poses difficulties. This is because there is no switch element in bondgraph. To overcome this problem, either macro- models of power switches or power switches represented by modulated bondgraph elements are used. The -former method leads to very stiff systems and the latter switches either the flow or effort variable losing control on the other variable. To overcome the problem of modelling switched systems, this paper proposes the concept of switched junctions. This proposal is a generalization of the existing definition of bondgraph junctions. With this extension, the bond- graph modelling capabilities are extended even to switched power electronic systems. These extended bondgraph modelling capabilities arc demonstrated on a few switched power con- verter configurations.
References
1. THOMA, J.
Hierarchical non-1ine;ir hondgraphs: a unified ~ncthodology for modelling complex physical systems, Siniirlrrrio~z, 1 992, 58, 230- 247,
Bondgraph nlodelling and sirnulation of a closed loop power smpli- fier, 13th IMACS Cl/or&i Cotigrrss at Dublin, Ireland, July 199 1.
Bondgraph simulation of a current source inverter driven induction motor (CSI-IM j system, Elect. Maclr. Power Systems, 1993, 21, 5 1-67.
J. Indian I m L Sci., July-Aug. 2000, $0, 333-346. O Indian Institute of Science
DEBIPRASAD PANDA AND V. RAMANARAYANAN Department of Electrical Engineering, Indian Institute of Science, Bangalore 560 0 12, India. emails: [email protected] and [email protected]; Fax: 9 1 (0)80 360 0444, 5) 1 (0)80 360 0683.
Received on February 28, 2000.
Abstract
Switched reluctance (SR) motor has a number of' promising features like simplicity, ruggedness, reliability md cost- effectiveness. It offers relatively high torque-on-inertia ratio, is capable of running at high speeds and has high effi- ciency. In all. it has a good potential for motion-contlal applications. However, the need of ii dedicated power con- verter, higher torque ripple imd higher audible noise are i t s major limitation s. In this paper-, the constructional features of SR motor are bricrty discussed. With experimentally obtained flux-linkage and static torque chamteristics, suit- able model of the motor is achieved. Dynamic simulation of the motor using time-step integration method is discussed. Desired control parameters T O N , T O F ~ and I*, hr different speeds and torques are obtained through open loop simula- tion. The torque ripple for different combinations of the above parameters is illustrated. Closed loop dynamic perform- ance is studied in simulation and results of current, torque, and speed waveforms are presented. The torque-speed enve- lope of the SR motor is found through simulation and the results for the same are presented.
Switched reluctance (SR) motor has a number of promising features. Professor Lawret~son in a seminal paper' in 1980 opened up its potential and inspired researchers to exploit the possibili- ties for variable speed applications. Besides its simplicity, it is quite rugged, reliable, and inex- pensive too. It offers relatively high torque-to-inertia ratio, is capable of running at high speeds, and has high efficiency. In d l , it has a good potential for motion control applications.
Though the mechanical construction of the motor is extremely simple, the flux-linkage characteristics and torque production mechanism are highly non~inear ,~ which makes the analysis and controller design of the drive complex. Its major limitations are that line start is not possible, it needs a power converter and rotor position to start and run and generates high torque ripple and audiable noise. Mercifully, several control algorithms are available in the literature for torque ripple minimisation,%ensol-less and acoustic noise reduc- tion.''
In this paper, modelling and simulation o f the SR motor is addressed. The flux and torque are nonlinear functions of phase current and rotor position. The controller design of such nonlinear systems requires computer-based modelling. Besides, the nonlinear equations gov-
334 DEBRIPRASAD PANDA AND V. RAMANARAYANAN
erning its dynamics are difficult to solve analytically. Dynamic simulation is more convenient to study the performance of this drive.
The outline of this paper is as follows. Constructional features of the SR motor are briefly described followed by experimenatlly obtained flux-linkage and static torque characteristics. With the experimental data obtained a suitable model of the motor is achieved. Dynamic simu- lation of the motor using time-step integration i s discussed. The desired control parameters, TONY TOFF and I*, for different speeds and torques are obtained through open loop simulation. The torque ripple for different combinations of the above parameters is illustrated. The torque and speed waveforms of the drive during steady state and transient are shown. The torque- speed envelope is found and its results are presented.
2. Construction
The construction of the SR motor is different from conventional motors. Unlike the conven- tional machines, the number of stator and rotor poles in SR motor is different. The most popu- larly used SR motors have stator : rotor pole numbers of 6 : 4 and 8 : 6. Other possible configu- rations are 6 : 2, 8 : 6, 10 : 4, 12 : 8." Even 2 : 2 single-phase motors are possible, but they
10 require extra parking magnet or some other starting mechanism. Similarly, 3 : 2, 4 : 2 and many other configurations are possible. The motor comprises it single stack. Both the stator and rotor are constructed from laminations. An 8 : 6 pole configuration is taken up here to demonstrate the different constructional features of the SR motor (Fig. l a). This motor has four phases. Each phase winding is split into two diametrically opposite stator poles. To maintain the clarity of the figure, only phase1 (Phl -Phlf) is shown with winding. In SR motor, the phases are excited through a power converter. For illustration, one phase winding with switch- ing devices and supply voltage is shown in Fig. Ib. Semiconductor switches (S1, S2) are on while the phase is excited. Diodes D 1 , D2 are required for dernagnetisation. The other phases follow similar kind of switching. alraneernent. In relation to the SR motor the following u W
observations can be made (Fig. 1):
Ph3' Ph3
Phlr
FIG. 1 a. Cross-section of an 8 : 6 pole four-phase SR motor.
FIG. I b. Phase winding with switches.
MODELLING AND SIMULATION OF SWITCHED RELUCTANCE MOTOR DRIVE 335
SR motor is doubly salient.
e Only stator is energised; the excitation voltage is purely DC.
Stator winding is essentially of concentrated type.
Rotor does not have any winding or permanent magnet.
No carbon brush, commutator or slip ring is required, hence the motor is brushless.
The observations lead to the conclusion that the const~vction of the SR motor is simple. he absence of permanent rnqnets, carbon brushes, slip rings, commutators and cage bars makes this motor rugged and reliable. Such a rugged motor is quite suitable for very high- speed applications. Obviously, the manufacturing cost of this motor will be less.
3.1. Electrical subsystem
The electromagnetic characteristics of the SR motor can be represented by the following equa- tions.
dv . . (~ . i . ) vj = R x i j -I- 1.1 . .I ' J 4-
dyf,k (q , ik ) dt d t
duj; T, = - li = constant, dB
where vj is the phase voltage; ij, f i j . N, 4, W; and Tej, respectively, are the current, self flux, net flux, position, coenergy and developed torque of jth phase, fik is the mutual f lux linkage of jth phase due to other conducting phase (kth phase); ik and Ok are the current and position of kt11 phase. All the equations shown above are nonlinear and the solution of these equations is tedi- ous, Using finite element method,'' it i s possible to solve these equations; however, the accu- racy of these methods depends on the number of elements used for computation and physical understanding of the system. Besides, it is cornputationally complex. In this paper, an aItesna- tive modelling approach from test results is followed. The experimentally obtained data of static flux-linkage and static torque characteristics are used in modelling. Mutual flux and eddy current effects are neglected.
The experimental methods for measuring the flux-linkage and static torque characteristics of the motor are explained elsewhere (Figs 2a and b)? It may be seen that both the characteris- tics are nonlinear functions of position and current.
The static flux-linkage characteristics are approximated by the following mathematical ex- pression.
DEBRIPRASAD PANDA AND V, RAMANARAYANAN
Current in Amp
FIG. 2a. Experimentally obtained flux-linkage character- istics for different positions (0" : 1" : 30") and currents (0: 1 : f S A j .
where i and yare the current and the flux linkage of a particular phase; K2 and Kj are constants and &(@, ( 0 , iy2( 8) are functions of position. For the test motor, the constants K2 and K3 are chosen as I 1 and 185, respectively. The magnetic characteristics of a particular phase in SR motor are symmetric with respect to its aligned position (30" in this case). Hence, the need of modelling for the test motor is effectively reduced to a zone spanning from 0" to 30". The above parameters K1(0j, ty1[8), y2y,(0) are stored in a look-up table for every 3". Linear interpo- lation is used for intermediate values (Table I). To testify the accuracy of the above modelling, the test. data and the curve following the mathematical expression in eqn (5) are plotted in Figs 3a-c for positions 0°, 15" and 3U0, respectively.
Table I Model parameters for the y- i characteristics
MODELLING AND SIMULATION OF SWITCHED RELUCTANCE MOTOR DRIVE 337
Current (A) Currcn t (A) t 0
Current (A)
FIG. 3. Compassion of experimental data (;l''i:+) and fit curve (continuous line) at (a) O", (bj So and (c) 30" positions.
3 2. Mechanical subsystem
In the present test set-up, a DC generator is used for loading the SR motor. Hence, the lnotor combined with the load completes the mechanical subsystem. For modelling the mechanical subsystem, frictional coefficient ( B ) and inertia ( J ) are required. 3 i s obtained through no-load test.
The motor i s run in no load at different speeds and the no-load torque at various speeds is noted with a torque transducer. B may be obtained hy dividing the no-load torque by the speed of the motor. It is seen that the value of B varies with the speed. The average value of B ob- tained following the above process is 0.0065 Ndradls.
The inertia of the system (load + SR motor) i s obtained by retardation test. Figure 4 shows the speed versus time plot during the retardation test. The rated speed of the motor is 1500 rpm. From the plot it can be seen that the time taken for the speed to fall from 1500 to 500 rpm is about 12.25 s, which can be treated as the mechanical time constant (z) of the total system. The inertia of the system can be computed using the expression, J = z x B. Form the given value of B = 0.0065 Nin/rad/s, J is 0.08 kgm2.
4.1. Methodology and simulation block diagram
The basic algorithm for simulating the operation of the SR motor, followed in this work, is the 2, 10 time-stepping integration of the voltage equation:
-10 10 20 30 40 50 FIG. 4. Speed vs time plot during retardation test of the Time (s) SR motor (including the DC generator).
DEBRIPBASAD PANDA AND V. RAMANARAYANAN
FIG. 5. Simulation block diagram of the SR motor.
J(vj - ~,ij)dt
where vj i s the voltage applied to the phase winding, ij the phase current and Rj the phase resis- tance. The phase voltage is a function of the DC supply voltage and the states of the switches in the controller. At any instant, the phase-flux linkage i s obtained by integrating eqn (6) over the previous time step. Now with a known position (q), the value of ( t j ) may be found using the mathematical expression in eqn ( 5 ) .
Once the value of current and the position of a particular phase is known, the instantaneous torque due to that particular phase may be found using the stored static torque data, T4 =&, 8,). Adopting a similar process, the torque due to other conducting phases also may be found. The summation of 311 the phase torques at any instant gives the instantaneous torque of the motor. The average torque over an electrical cycle can be obtained by summing and averaging the instantaneous torque over a period of 60" rotation. Alternatively, average torque may be computed through coenergy principle. 2 , 10
The instantaneous torque, T, can be used to simulate the mechanical subsystem of the drive given by the fol lowi~~g equation
where TI is the load torque, J the total inertia of the load and the motor and B the frictional coefficient of the total mechanical system. Speed i s the output of the mechanical subsystem in the simulation. The same may be used to carry out the simulation with closed loop controllers. It will be discussed later in detail.
Adopting the above method, the dynamic and steady-state performance of the drive are studied through simulation. The basic si~nulation block diagram of the SR motor i s shown in Fig, 5 . Simulation is conducted with the following assumptions:
(i) DC bus voltage is considered ripple free. (ii) Eddy current effects are ignored. (iii) Mutual fluxes are ignored. (iv) Maximum switching frequency of the converter switches is limited to 10 kHz.
4.2. Control parameter selectiolz
The control of the SR motor i s largely divided into two regions: (i) low-speed chopping control and (ii) high-speed angle control. In the former, the TON and ToFF angles are kept fixed and
MODELLING AND SIMULATION OF SWITCHED RELUCTANCE MOTOR DRIVE
Converter
Switching signal generator
FIG. 6. Controller block diagram with estimator.
torque is controlled by regulating the cunent through winding. Normally, a hysteresis control or fixed frequency chopping control is used as current controller. On the other hand, in high- speed operation, back-emf i s sufficiently high and chopping control is no longer possible. Thus the torque is controlled solely by the control of TON and ToFF angles. The low-speed chopping control and high-speed angle control are illustrated in Fig. 6. The raw enable signal is the maximum duration for which a particular phase may be excited. The envelope of raw enable signal is strictly a function- of speed of the machine. In low-speed chopping control, the phase current is maintained to its reference value by a hysteresis controller. The switching of a par- ticular phase is restricted within the envelope of the raw enable signal as illustrated in Fig. 6. Thus, in this speed range, the TON and Tow angles are fixed for a given speed but current refer- ence (I:" i s chosen as a function of load. On the other hand, in high-speed angle control mode the torque ripple is controlled by changing only the TON angle though the raw enable signal is chosen as a function of speed. Thus, the major control parameters for the whole speed range are represented as reference current (I:"), turn-on (TON) and turn-off (TOFF) angles. The success- ful and efficient operation of an SR motor largely depends on the choice of these control pa- rameters. The basic purpose of simulation is to choose the control parameters and study the perfo~mance of the drive prior to its actual implementation. The system is first simulated in
0 0 ,2 0.4 0.6 0.8 1 speed ( p . ~ . )
FIG. 7. An arbitrary set of TON and ToFF angles for different rent limit of 18 A. (a) TON vs speed and (b) Tow vs speed.
22 i 1 h I
0 0.2 0.4 0.6 0.8 I speed @.u,)
speeds to obtain rated torque (25.5 Nm) with a peak cur-
FIG. 8. Cornparsion of sin~ulation results of (i) phase current, ( i i ) phase torque, (iii) speed waveform, (iv) instantaneous and average torque at 150 rprn and full load for two cases, (a) with TON = 10.Y and ToF~: = 27.Y and (b) with TON = Oo and Ton: = 23.15".
open-loop steady-state condition. It is assumed that the motor is running at a predetermined speed, and the required average torque is obtained by adjusting the TON, T Q I T F and reference current I". It is observed that at any particular speed, the required average torque may be achieved by several combinations of parameters (Fig. 7). The combination which gives higher torque per peak current and lower torque ripple is desirable.
4.2.1. Illustration of torque ripple with digeel-ent cornbirzations of control parameters
A set of TON and ToFF angles is arbitrarily chosen for different speeds and the rated torque is obtained with these values when the peak excitation current is limited to 18 A. The torque rip- ple in Fig. 8a, 9a and 10a is typical of these set of values. On the contrary, with a different set of the above parameters the torque ripple is reduced, though the average torque and speed are the same. Phase currents, phase torques, average torques and speeds with reduced torque ripple and different values of TON and ToFF angles are given in Figs 8b, 9b and lob. It may be con- cluded that by adjusting the TON and ToFF angles, i t is possible to bring down appreciably the torque ripple in low and medium speed ranges.
MODELLING AND SIMULATION OF SWITCHED RELUCTANCE MOTOR DRIVE
60(0) 20 40 6Q(O) 20 40 Average Torque
Rotor position (degree) Rotor position (degree)
FIG. 9. Cornparsion of simulation results of (i) phase current, (ii) phase torque, {iii) speed waveform and (iv) instanta- neous and average torque at 750 rpm and full load (25.5 Nm) for two cases, (a) with TON = 5.Y and Tow = 2 5 Y and (b) with TON = OQ and TOFF = 2 1
4.2.2. Final choice of parameters and closed loop simulation
A new set of TON and ToFF angles is shown in Figs i l a and b. The duration between these is regarded as the raw enable signal for different speeds. Maximum torque is achieved for various speeds of the motor when the DC bus voltage is maintained at its rated value by choosing the TON and TOFF angles as given in Figs 1 l a and b and with a current reference of 18 A (Fig. 12). At high speed, the TON angle is varied with bath speed and demanded torque. Torque produced for different TON angles and speed are given in Fig. 13 when the ToFF angles are maintained at the values as given in Fig. I Ib. With overload condition, the TON angles may be further pushed towards the negative slope of inductance. For a peak current limit of 27 A (50% more than rated current, 18 A), torques along with corresponding TON angles for various speeds are plot- ted in Fig. 14. Thus the TON angle is varied with both speed and load. The set of values shown in Figs 11, 13 and 14 are used for coritrolling the test motor in the angle control mode up to the rated speed of the motor. The ToFF angle is set as a function of speed (Fig. 1 lb).
DEBRIPRASAD PANDA AND V.
(0
Average Torque I I I 1
Rotor position (degree)
Y
60(0) 20 40 60(0) 20 40 60(0) Average Torque
I l i I I 1
Rotor position (degree)
(b)
FIG. 10. Cornparsion of simulation results of (i) phase current, ( i i ) phase torque. (iii) speed wuvef'orm, (iv) instantaneous and average torque at 1500 rpm and R111 load (25.5 Nm) for two cases, (a) with ToN =-So and TUFF = 23.75' and ( b ) with TL>N = -5.2S' and T[JFF = 22.S0.
In simulation, the total control regime i s divided in two: low-speed chopping control and high-speed angle control. In the functions of speed only as given ing upon the demand of load. makes use of the curves in Fig.
low-speed range (0-900 rpm), the TON and ToFF angles are in Figs 1 l a and b, respectively. Current is controlled depend The torque-to-current f~~function generator in the controller 12 for this purpose. These function generators are obtained
through simulation and it i s shown that current reference is a strong function of torque and it does not vary much with speed. The current references for intermediate values of speeds are
speed @ .u .) speed (p .u .)
Rc. 1 1. Variation of TON and TUFF angles with speed to obtain maximum torque with a peak current limit of 18 A, (a) TON vs speed and (b) to^^ vs speed.
MODELLING AND S l ~ U A T I O N OF SWITCHED RELUCTANCE MOTOR DRIVE 343
0 0.2 0.4 0.6 0.8 1.0 2.2 1.4 Torque @.ti.)
FIG. 12. Current vs torque relationship nt different loads and speeds: ( i ) 900 (ii) 600, (iii) 150, (iv) general square root ( I = sqrt(Torque)) relationship for all speeds below 900 lpn (dotted lines)
-10 - 5 0 5 18 15 T,, angle (degree)
FIG. 13. Torque vs TON angles at (i) 0.6, (ii) 0.7, (5) 0.8, (iv) 0.9 and (v) 1.0 pu speeds with peak phase current limit of I8 A.
obtained through linear interpolation. A linearised function of current as the square root of torque is also szhown in Fig. 12 (iv). Obviously, with this linear relationship, the performance of the drive during transient will be poor compared to what is obtained with function genera- tors. Thus, in the speed range of 0-900 rpm, the main control is realised through hysteresis controller and the TON and ToFF angles are obtained from look-up tables given in Figs I l a and b, respectively.
In the region 900-1 500 rpm, current (torque) is controlled only by angle. The ToFF angles for this speed range are the same as given in Fig. I I b. But the TON angle varies with both speed and load and the corresponding values for this speed range are plotted in Figs 13 and 14.
Through simulation it is found that for very high-speed operations (2.5-5 pu speed), the TON angles may be pushed back to -16" and ToFE angle may be advanced up to + 1 4 O to obtain better torque speed cl~aracteristics. In this speed range, the TON angles are linearly varied from -1 6" to -3.75" and the TOFF angles from -1-1 4" to -1-22.5".
Once these parameters are finalised for all the speed and torque conditions, they are stored in a look-up table and referred by the controller. The simulation is carried out using these val- ues for variable speed and load conditions. A PI controller is used for controlling the speed of the motor. A hysteresis controller with ~naximum switching f ~ q u e n c y limitation of 10 H z is used for current controller.
4.2.3. R~sul t s
The torque and speed waveforms at different steady-state and transient operating conditions are presented in Fig. 15. The torque-speed envelope is an important criterion for any electric drive. Figure 16 gives the torque-speed envelope of the SR motor up to base speed (1 pu). It may be noticed that up to the base speed, the SR motor is capable of providing rated torque. At low speeds, with the rated phase currents (18 A for the test motor), the motor is even capable of producing 30% to 35% more than rated torque. At higher speeds, up to 2.5 times the base
DEBMPRASAD PANDA AND V. RAMANARAYANAN
i 1 I I t -10 4 - 6 -4 -2
degree
FIG. 14. Torque vs TON angles at ( i ) 0.6, (ii) 0.7, (iii) 0.8. (iv) 0.9 and (v) 1 .O pu speeds with 27 A as peak cunent limit (50% more than rated current).
0 5 10 15 20 Time (s)
FIG. 15. Speed and torque waveforms during steady- state and different transients of speed. (i)-(x) represent different transients in speed (here load is a Function of speed.)
speed, the motor can be operated in constant power mode. Beyond 2.5 pu speed, power falls with the speed. The torque- and power-speed envelopes from 1 to 5 pu speed are presented in Figs 17a and b, respectively.
5. Conclusion
In this paper, modelling and simulation of the SR motor are illustrated. Modelling of the motor is done from the measured flux-linkage characteristics. The model of the mechanical subsys- tem is obtained by measuring the frictional coefficient and inertia through no load and retarda- tion tests, respectively.
Dynamic simulation of the SR motor with the above model is obtained through time- step integration method. Simulation results of phase cun-en t, phase torque, average torque and speed are presented for various operating conditions. Suitable control parameters like TON and ToFF angles and cunent reference I#: are obtained from simulation for various operating points. It is shown that for a particular speed and load, different combination of parameters are possi- ble. By choosing suitable values for these parameters it is possible to reduce the torque ripple
'0.2 0.4 0.6 0.8 1 FIG. 16. Torque-speed envelope for less than rated speed speed (PA.) (0 to 1 pu speed) with a peak current limit of 18 A.
MODELLING AND SlMulATION OF SWITCHED RELUCTANCE MOTOR DRIVE 345
w
0 1 2 3 4 5 speed (p.u.)
0 1 2 3 4 5 speed (pu.)
of the SR motor. In this work, the above parameters are selected based on the criterion of maximising the torque per peak ampere of the SR motor.
The transient performance of the SR motor drive is studied through a closed loup simula- tion, Speed and torque waveforms during steady state and transient are presented for various operating points. A PI controller is employed for realising the speed controller. It is observed that with the given inertia (0.08 kgm" and frictional coefficient (0.0065 Nrn/rad/s), the drive using SR motor can reach from zero speed to the rated speed in 1.5 s with a peak current limi- tation of 18 A.
The torque-speed envelope throughout the operating region is found through simulation. It is seen that the SR motor is capable o f following the characteristics of a DC motor. h addition, at lower speeds, 30%-35% more than its rated torque can be achieved with the same peak cur- rent. Besides, it is observed that at 50% more than the rated current, the motor can deliver more than twice the rated torque.
1. LAWRENSON, P. J., STEPHENSON, J. M., Variable-speed switched reluctance motors, IEE Proc. B, 1980, BLENKINSOP, P. T., CORDA, J. AND 127,253-265. F u L ' c ~ , N. N,
2. RAMANARAYANAN, V., VENKATESHA, L. Flux-linkage characteristics of switched reluctance motor, Powo- AND DEDIPRASAD PANDA Elcc*tronitbs Drives and Energy System Conf (PEDES-96), New
Delhi, India, December 1996, pp. 28 1-285.
Torque ripple minimisation in switched reluctance motors via bi -- cubic spline interpolation, IEEE- PESC Conf RPC., 1992, pp. 85 1 - 85 6.
4. RAY, W. F. AND BAIIADLY, 1. H. Sensorless methods for determining the rotor psilion of switched reluctance motor, Prac. 5th E m Cortf: on Power Electlanics and Applications, Sept. 13-16, 1993, Vol. 6, IEE publication No. 377, pp. 7-13.
5 . DEBIPRASAD PANDA AND
RAMANARAYANAN, V.
(5. DEBIPKASAD PANDA AND
RAMANARAYANAN, V,
Mutual inductance and its effect on steady-state performance and position estimation method of switched reluctance motor drive, Proc. IEEE, [AS Annual Meeting, Oct. 1999, Phoenix, USA, pp.
An accurate position estimation of switched reluctance motor with srnooth starting, Proc. lEEE h t . Car$ on Indusrrial Technology (ICIT), Jan. 2000, Goa, India.
A novel sensorless method with self-measured flux-linkage charac- teristics, Proc. IEEE Power Electronics Specialist Con$ (PESC 2000), Galway, Ireland, pp. 1569-1574.
336 DEBRXPRASAD PANaA AND V. RAMANARAYANAN
9. DEBIPRASAD PANDA AND
RAMANARAY ANAN, V.
Acoustic noise cancellation techniques for switched reluctance drives, IEEE IAS Conj Proc., 1995, pp. 448-455.
A composite control strategy for low-noise and sensorless operation of switched reluctance motor, Proc. lEEE IAS Annual Meeting, 2000, Rome, Italy, pp. 1751-1758.
Switched rellccta~zce motors and their control, Magna Physics Pub- lishing and Clarendon Press, 1993.
1. I . ARUMUGAM, R., LOWHER, D. A., Magnetic field analysis of switched reluctance motor using a two- KKISNNAN, K. AND LINDSAY, J. F. dimensional finitie element model, IEEE Trans., 1985, MAG-21,
1883-1885.
Received on March 2 , 2000.
Abstract
Achieving soft switching in swit~hed-mode power supplies is an importctnt design criterion today. Converters like the sel-ies-loaded resonant topologies are preferred because they easily enable soft switching. However, the soft switching range is limited and depends on a number of parameters that are not directly under the designer's control. Special de- sign rules are used to arrive at the optimum operating points and thus maintain soft switching over nt least a lnajor portion of the operating range. This paper deals with the design issues relevant to achieve zero-voltage switching (ZVS) for the phase-modulated series resonant converter. The ZVS mechanism and the modes of operation for which the converter inherently ofTers ZVS are explained. A circuit modification to extend the ZVS range beyond the rlatural range of the converter is introduced. Further, the design procedure for achieving ZVS is discussed through a design example. A phase-modulated series resonant converter for high-voltage DC applications i s designed and experimen- tally shown to have ZVS over its entire specified operating range.
The series resonant converter (SRC) is a load resonant converter that finds usage in high- power conversion applications like ship propulsion, electric traction and in special applications like high-voltage output DC-DC converters. Large transforxner parasites are typical of these converters. Far example. the leakage inductance of the high-voltage transformer in a high- voltage DC-DC converter is large enough to prevent sufficient power flow from the source to the load. Introducing a capacitor in series with the parasitic leakage inductance results in a se- ries LC circuit. The effective impedance offerred to the power flow is then reduced if the con- verter is operated with a switching frequency dose to that o f the resonant frequency of the se- ries LC circuit. The classical SRC is a frequency-controlled converter. That is, control is achieved by varying the switching frequency with respect to the converter's resonant fre- quency. Control can also be achieved by varying the duty cycle. Such an SRC, having a fixed switching frequency but variable duty ratio, i s called the phase-modulated series resonant con- verter (PM-SRC) (Fig. 1). The term PM-SRC is adopted because the PWM inverter output voltage is obtained using the phase modulation technique.
One of the critical design issues in any switched-mode power supply is achieving soft switching over the entire operating range of the converter. A soft-switched converter has lower
348 BIJU S. NATHAN AND V. RAMANARAYANAN
T HV
Output
1
s'l gate
I I I I
I I 1 1 :
_L
Leading Leg
S I gate
52 gate
Lagging Leg
Quasi-Square Input
Voltage
FIG. 1. Schematic of the phase-modulated series resonant converter for high-voltage applic a tv 1011s.
generated EMI, switching losses and stress on the power switches. The lower switching losses promote higher switching frequency, which further results in overall size and cost reductions. Though one o f the positive features of the SRC is that it enables soft switching, the soft- switched operating regions are limited and depend on a large number of parameters, including the load. Proper design is therefore required to ensure soft switching over the defined operating range.
This paper explains the design issues critical for achieving soft switching in MOSFET- based PM-SRC operating above the resonant frequency. The paper discusses how zero-voltage switching (ZVS) is achieved and its operating zones. A modified circuit to extend the range of ZVS is described. Design relations for achieving ZVS are explained through a design example.
DESIGNING FOR ZERO-VOLTAGE SWITCHING 349
leakage inductance and any external inductor if present. The resultillg resonant tank will have a
resonant frequency vi.) given by
Switching frequency &) of the inverter is kept constant while the duty ratio (D) is varied usillg phase modulation technique. Therefore, the resonant tank is excited by a quasisquare waveform having frequency Ji and duty ratio D. The excitation results in a sinusoidal or near- sinusoidal current in the resonant tank. The tank current (@)) is rectified by a diode bridge rectifier and filtered using capacitive filter to get the required output voltage. The magnitude of the output voltage depends on tbe magnitude and wave shape of i ( t ) and the load (Rlond in Fig. 1). The magnitude and wave shape of the resonant tank current depends on&, D and the load factor ( Q ) of the converter. By definition, Q is the ratio of the resonant tank characteristic im- pedance and the resistive load as seen from the resonant tank. That is,
where Rpri = ~ ~ ~ ; ~ ~ ~ l i l ~ is load as seen by the resonant tank and n is high-voltage transformer turns ratio.
The operating frequency (ji) is gener*ally close to the resonant frequency of the tank. Opera- tion with ji <*fi= is c a k d the below resonant frequency operation. The input voltage then sees a net capacitive tank circuit? When f , >&, the operation is termed above resonant frequency and the tank presents a net inductive circuit?The above resonant frequency operation facilitates ZVS for the inverter devices while the below resonant frequency operation promotes ZCS (zero-current switching).
For phase modulation," a full bridge inverter with fully controlled devices is required (Fig. 1). Each device i s switched at 50% duty ratio with the high- and low-side devices of the same leg being complementary. As shown in waveforms of Fig. 1, conduction of switches on the
leading leg of the inverter (S,, x) is phase-shifted with respect td the conduction of switches
BIJU S. NATHAN AND V. RAMANARAYANAN
on the lagging leg (S,, F,), resulting in the quasisquare input voltage. Here, the inverter leg
containing the devices that turn on first (S, , ) is termed as the 'leading leg' and the other leg
i s termed as the 'lagging leg.'
3. Zero-voltage switc
As mentioned earlier, when the PM-SRC is operated such that its switching frequency is oreater than the resonant frequency of the tank, zero-voltage turn-on of the inverter devices is 0
possible. The latter i s preferred over zero-current turn-off in MOSFET-based converters. ZVS ensures that the inherent output capacitance in the MOSFET switch is discharged prior to switch turn-on, thus preventing turn-on losses and generated EMI. Figure 2 shows how ZVS is achieved for the switch q. The high-side MOSFET S1 is turned off at time TI - , (Fig. 2(b))
when the tank current i ( t) still tlows through it. Because of the resonant inductance, i(t) is not interrupted. The tank current is maintained through the snubber capacitors across the switches an the leading leg (S1 and $ ), thus charging CII to Vdc and discharging C12. The inherent body
diode of the MOSFET turns on (Fig. 2(c)) once the voltage across C12 reaches the diode
forward break over voltage. The tank current presently freewheels through the body diode of & . the resonant tank and the still conducting lagging-leg switch S2. The voltage across S , is
the diode drop and is nearly zero. Applying the gate pulse to turn on during this period (at
time Tl - b) results in zero-voltage turn-on. The switch $ being a MOSFET will conduct i(r) in
the reverse direction ns soon as its gate pulse i s applied (as indicated in Fig. 2(d)).
The turn-on process is therefore lossless and stress-free. Turn-off occurs when a sizeable cumnt flows through the switches. But the rate of voltage rise across the switch is limited by the snubber capacitors across it. The snubber capacitors are subsequently discharged through the load (thus preventing snubber losses) prior to the switch turn-on. Therefore, the turn-off process also has low losses.
Zero-voltage turn-on sequence of the other three switches of the converter i s similar. The basic requirements for achieving ZVS and low loss turn-off are therefore:
The device should turn off with a positive current flowing through it. The turn-off current should be large enough to completely charge/discharge the snubber capacitors across the devices and subsequently turn on the diode across the device com- ing into conduction. There should be a time delay (TD) between the turn-off of a device and the turn-on of the complementary device on the same leg. In the example above, T - = - TD should be large enough to allow the complete chargingldischarging of the snubber capacitors and turning on of the antiparallel diodes but small enough to ensure that the tank current does not change direction before the device is turned on. Snubber capacitors should be large enough to prevent high dvldt for the largest turn-off current, but small enough to be chargeddischarged by the smallest turn-off current in the given time delay.
DESlGNlNG FOR ZERO-VOLTAGE SWlTCHING
Leading Leg Lagging Leg Leading Leg
CI-
Leading Leg Lagging Leg Leadingkg t =Tq b - Lagging Leg
Fro. 2. Zero-voltage turn-on sequence for MOSFET y,. Tho currenk paths are shown in dark,
Under steady state, the PM-SRC operating ahove the resonant frequency has three modes of operation as indicated in Fig. 3 . ~ o h e I is a continuous conduction mode. Its operation results for high Q, high D or when thej& ratio is high. All the four devices turn off when a positive current flows through them ZVS i s therefore easily achievable over almost the entire range of Mode I . The exceptions arc at lower loads when the lagging-leg switches do not turn off enough current to completely chargeldischarge the snubber capacitors.
Resonant Capacitor Voli
MODE I MODE 2 MODE 3
FIG. 3. The three ~nodcs of operation of the PM-SRC operating above the resonallt frequency.
BIJU S. NATHAN AND V. RAMANARAYANAN
Fie. 4. Sequence showing loss of ZVS and turn-on stress on the lagging leg for a Mode 2 operation. The current paths are shown in dark.
The leading-leg switches turn off with a positive current fur Mode 2 operation. However, the lagging-leg switches turn off with a negative current, i.e. after the tank current has reversed direction. Figure 4 gives the sequence of lagging-leg turn-off for Mode 2 operation. The tank current i(t) reverses direction before the lagging-leg switch (S2) is turned off (Fig. 4(b)). Since the gate pulse is still present for S2, it conducts in the reverse direction. Once the gate pulse to Sz is removed, its inherent body diode takes over the tank current (Fig. 4(c)). The switch F, is turned on after a time delay from the turn-off of S7 (Fig. 4(d)). The S2 body diode is force- cornmutated and the snubber across the device turning on (% ) dumps its charge through the device. The result i s turn-on losses and generated EMI. Further, for high DC bus voltages there can be a false turn-on of S2 due to Miller capacitance, leading to a shoot through. Figure 5 shows the oscilloscope plot of a lagging-leg switch commutation during Mode 2 operation. Clearly, Mode 2 operation is to be avoided.
DES lGNING FOR ZERO-VOLTAGE SWITCHING 353
w c
!! 3 Cx
e3] a %
P Forced Commutation
I I 1 I t I I
Time
FIG, 5. O~cilluscope p l o ~ c)f the lagging switch transients for a Mode 2 oper~ition.
Snubber Capacitor Discharging Through Del ,-' / Device
FIG. 6. Simulated waveform of lagging leg device cur- rent during Mode 3. ZCS ensures low loss and low stress turn ofl; but snubbers discharge through the device dur- ing turn on.
Mode 3 is o discontinuous operating mode mukin: , when Q < ( ' /rr)~;/ f i ) . ' The leadingleg ,witches turn off with n positive current: however, the turn-off current may he too low to &aqge/dischn~*ge even the MOSFET output capacitance. The lagging-leg devices turn off at zero cull*ent. Lagging-leg turn-off stress and losses are low because of ZCS; however. the lag- fling-leg snubber capacitors dump their charge through the device, leading to generated EM1 b
and turn-on losses (,Fig. 6j. Though not as scvcse as the Mode 2 turn-off transients, Mode 3 operation too has to be avoided because of the gcner:ltccl EMI.
Figul*~ 7 gives the operating mode houndasics' for different loads (Q) and duty ratios (D) when ,f;& = I . 1 . The thick linc traversing through the plot indicates the operating point for maintaining n constant nominal output voltagc whcn the load i s changing. For the plot consid- -
ered in Fig. 7. thc nominal output volt;lgc i s 0.6 VLk and the moximum load corresponds to
354 BIJU S . NATHAN AND V. RAMANARAYANAN
Q = 6. Mode 1 operation is lost below about 60% of the nominal load. If the load is between 17.5% and 60% of the nominal, the operation falls in Mode 2, while lower load results in Mode 3 operation. Figure 8 gives similar mode boundary plots for different&/& ratios. As seen, Mode 2 region decreases as the switching frequency moves away from the resonant frequency (higher f s l f r ) *
Mode 2 operation can be avoided by selecting high values for the nominal Q or by operat- ing the converter farther away from the resonant frequency. The former results in large peak currents and voltages and lower voltage gain. The latter solution decreases the converter re- sponse time, the power delivered and the voltage gain, and has lower efficiency because of the relatively higher switching frequency.
Achieving ZVS for the lagging-leg devices is a more critical issue than the leading-leg de- vices. As seen above, the leading-leg devices turn off with a positive current under all operat- ing conditions. ZVS can then almost always be designed except when the turn-off current is too law. On the other hand, the lagging-leg devices turn off with negative or zero current in Mode 2 and Mode 3, respectively. Thus, circuit modifications are necessitated to ensure ZVS in these operating modes.
xtending ZVS range using external in
A method of increasing the ZVS range of phase-modulated resonant transition converter by including an external inductive circuit6 was introduced earlier. This method can be extended to the PM-SRC and the disadvantages of Mode 2 and Mode 3 operations as discussed in the pre- vious section can be eliminated5 (Fig. 9). An external inductor L,, is connected between point B of the lagging leg and point E, which is a node with stiff voltage V&/2 maintained by the large divider capacitors CeX1 and Crx2. Since S2 and & are complementary to each other, the voltage VEB across L,, is a squarewave switching between kVdc/2 at the switching frequency of the converter V;). The s u m of the external inductor current i,(t) and the tank current i ( t ) flows through the lagging-leg switches. The peak negative turn-off current and the peak external in- ductor current (les(peak)) coincide and by proper design, the net current flowing through the switches can be made non-negative.
2 1 MODE 2 1
FIG. 8. Variation of the mode boundaries with changing fJf,. Mode 2 region is reduced as f,!f, increases. Obtained through mathematical analysis.
DESIGNING FOR ZERO-VOLTAGE SWlTCHING
j
Lagging Leg
FIG. 9. Schematic of the modified circuit to aid in ZVS.
Then for an original turn-off current of loif I,, (which i s negative for Mode 2 and zero for Mode 3), the peak external inductor current to restore ZVS is
Using (-5) and the basic inductor equation, the value of the external inductor is
The original turn-off current I,,fl [,, i s determined from the mathematical analysis or nu- merical si~nulation of the PM-SRC. capacitor Cl:, is chosen depending on the maximum turn- off current possiblc for the Mode 1 operation. Thc actual design i s treated in the following sec- tions.
SPICE sin~ulation waveforms for a typical circuit with the external inductor ZVS aid are given in Fig. 10. The switch current i s now the sum of the tank and external inductor currents. Figure 11 shows noticeable improvement in the oscilloscope plot of switch turn-off transients for Mode 2 operation when the external inductive circuit is included.
Inclusion of the external inductor does not influence the tank current and thus the analysis for the modified PM-SRC remains the same. Since there are ideally no lossy components in the
356 BIJU S. NATHAN AND V. MMANARAYANAN
t I k I I t I I 1 J Time
FIG. 10, SPICE simulation waveforms for the modified FIG. I I . Oscillosc~pe plot of the improved lagging leg PM-SRC. switch transients on including the new modification.
path of i&, there is no power loss associated with the modification. Peak current through the switches S2 and F, increases, leading to slight increase in conduction loss. But this is more than offset by the improved commutation.
esigning for ZVS
The actual design for ZVS is explained here through an example design for a PM-SRC for high-voltage DC output. Converter specifications are as given in Table I. The converter nomi- nal operating points, currents and switch tun]-off currents are arrived at through mathematical analysis. Device output capacitance indicated in Table I corresponds to the selected MOSFET switch.
6.1. Leading-leg smtbber capacitor and time delay
The leading-leg snubber capacitor (Clcad = CI 1 or CI2 + MOSFET output capacitance) is chosen large enough to slow down the voltage rise for the largest turn-off current, thereby reducing turn-off losses. At the same time it should be small enough to be completely chargedl
Table1 .
Specifications of the experimental converter
Variable Nominal value
Output voltage Switching frequency V;) Nominal Gutput power Minimum output power Nominal voltage gain (primary side) Frequency ratio f J ' Nominal load factor (Q) Inverter device output capacitance
40 V DC 1000 V DC 100 kHz 200 W 20 W (10%) 0.625 1.1 4 1.1 n F
DESIGNING FOR ZERO-VOLTAGE SWITCHING 357
cI1 = clz = 7.4 nF (the rest provided by the MOSFET output capacitance).
The minimum turn-off current a ~ ~ ~ s p o ~ ~ d s to the minimum load (chosen here as 10% of the maximum load, Q = 0.4, Mode 3 operation). Below this, ZVS is lost for the leading-leg switches.
The time required to discharge Clcad is the delay time for the leading leg. Again, using the basic capacitor equation, this is determined as
- ~ I - k ; l d - 2C,,:dV, 2 300 ns.
IOff - lead is assumed to remain constant over the delay time TD - l,,d and the actual charge available to chargeldischarge CLeild i s the shaded area in Fig. 12. An amount of charge equalling the hatched area is now not avnil;lhle. Time T2 is the crossover time for Mode 3 and i s available from the mathematical analysis data or from numerical simulation. Approximating the area KLM to be a right triangle,
On the other hand, the charge required = 2CleacIVdc = 680 nC. The selected delay time is there- fore increased slightly to TD - leiid = 350 ns.
For the lagging leg, the maximum turn-off current fur the unmodified converter is
However, this value will be later revised to a higher value because of the additional current from the external inductor circuit. The external inductive circuit adds an extra current so as to make the negative turn-off currents of Mode 2 positive. Once again from the analysis data, the maxinlum negative turn-off current occurs for Q = 3 and D = 0.3 (this does not correspond to the nominal gain of 0.625, but is cimsen so as to eliminate the detrimental negative turn-off current for any operating condition).
358 BIJU S. NATHAN AND V. RAMANARAYANAN
FIG. 12. Leading leg switch turns off at TI with a. current l,,n-l,,d. For the design, it is assumed that the turn-ofl current remains constant for the dtlration To-lcrLj. However, an iunounl of charge equalling the hatched area i s actually not available.
Referring to eqn ( 5 ) and the capacitor equatio~l
Therefore, the minimum current required to chargeldischarge el,, is
Since the minimum turn-off current can be selected by suitably designing the external induc- tive circuit, one has the freedom to choose the delay time for the lagging-leg switches. In (10)
DESIGNING FOR ZERO-VOLTAGE SWITCHING
Table 1 Comparison of the calculated and the fine-tuned values
Initial caIcuIaf ion 8.5 3.7 19 350 AlterationafterSPICEsimulation 7.8 2.2 13 380 Final fine-tuned value 6,1 2.1 12.2 360
- -
"includes MOSFET output capacitance of 1 . 1 nF
and (1 1) above, & - I,, = TD_\ead = 350 ns. In case the control circuit has provision for setting different values for the delay times, TD - I,, can be chosen as a smaller value.
The external inductor (Fig. 9) value i s calculated from (5), (6) and (10) as
- Y , c - - 40 La - = 19 pH.
8.f:s'ex ipck) 8x100 k N ~ ~ ( 1 . 7 3 + 0 . 8 5 )
L,, is decreased by a small amount to increase the minimum turn-off cun-ent so that tile enor due to the constant current turn-off assumption is eliminated.
6.3. Fine-tuning the designed values
The design done above assumes ideal conditions and the values for the snubber capacitors, delay times and the external inductor have to be fine-tuned. The factors contributing to the non-idealities are:
The circuit parnsitics which change the switch currents and operating points.
Parasitic transformer capacitance of the high-voltage transformer, which has a signifi- cant influence on the switch turn-off currents.
Lower voltage across the external inductor circuit owing to conduction drop across the devices.
o Limited values for the
The converter with ZVS
snubber capacitors and their tolerance levels.
aiding components designed above is simulated using circvit- based simulation tools. Besides using models of the selected components, the measured circuit non-idealities are included. Based on this, the actual values used in the prototype are listed in Table 11.
FIG. 13. O ~ c i l l o ~ c ~ p e plots. (a) Lagging leg turn-on without the external inductor circuit, (b) with the external inductor circuit and (c) External inductor circuit current, resonant tank current and input voltage waveform.
360 BIJU S. NATHAN AND V. RAMANAMYANAN
FIG. 14. Oscill~scope plots. Tank current waveform for (b, d, f) showing ZVS at these loads.
100% ( c ) 50% and (e) 10% loads; Lagging-leg turn-on
Figure 13 gives the oscilloscope plot of the lagging-leg turn-off transient for a Mode 2 opera- tion with and without the external inductor circuit. Current through the external inductor circuit i s also recorded. As seen from the plots, there is a vast improvement in the turn-on transients by including the external inductor circuit.
Figure 14 shows the tank current and input voltage corresponding to loo%, 50% and 10% loads. Lagging-leg turn-on transients under the three load conditions indicate that ZVS is maintained throughout the operating range.
8, Conclusions
Soft switching is a desirable, if not an essential feature, in any power converter. This is more so for high-power switching converters where the power devices have to switch large currents at high-voltage levels. Soft switching is also needed for realizing high efficient y and low gener- ated EM1 specifications. The SRC is a preferred topology for high-power applications. Though
DESIGNING FOR ZERO-VOLTAGE SWITCHING 361
This paper addresses the design issues required for maintaining ZVS for the PM-SRC over the desired operating rmge. The different operating modes for the PM-SRC operating above resonance have been intt.~duced and those modes which inherently facilitate ZVS have been identified. It is found that of the three modes of operation, only one facilitates natural ZVS. A modified PM-SRC circuit with a few extra passive components is introduced. The modification
ZVS in all the modes of operation. Design relations obtained for ZVS are explained through a design example. The experimental results confirm that the modified circuit improves ZVS range. ZVS from ~ 0 0 % load to 60% load for the unmodified converter was improved to an extended range of 100% load to 10% load. Further, the ZVS feature enables stress-free op- eration in the otherwise undesired Mode 2 operation.
A complete DC characterization of a constant frequency, clamped- mode, series-resonant converter, IEEE Power Electronics Special- ist Conf '88, April 1988, pp. 987-996.
Series resonant converter with clamped tank capacitor voltage, IEEE Applied Power Electronics Conf., 1990, pp. 137-140.
A comparison of half-bridge resonant converter topologies, IEEE Trans., 1988, PE-3, 174-182.
A constant frequency resonant transition converter, J. hdian inst. Sci., 1996,76, 363-377.
Analysis, simulation and design of series resonant converters for high voltage applicu~ions, M. Sc. (Engng) Thesis, Indian Institute of Science, Bangalore, India, Dec. 1999.
A constant frequency resonant tmnsilion converter, M. Sc. (Engng) Thesis, Indian lnsti tute of Science, Bangalore, India, August 1995.
J. Indian Inst. Sci., July-Aug. 2000,80, 363-394. 0 Indian Institute of Science
Thesis Abstract M.Sc.(Engng)
regular tree Research supervisor: Prof. Priti Shankar Department: Computer Science and Automation
Linear regular tree pattern matching is a problem that has direct application to the generation of tools for rear ends of compilers. The problem can be stated as follows: Given a regular tree grammar, construct an automaton which recognizes trees generated by the grammar. Conven- tional approaches to this generate finite-state tree automata. We cast the problem as a parsing problem for context-free languages generated by grammars that are in general highly ambiguous, though restricted in form. This restriction enables us to generate linear time recog- nizers that are actually deterministic pushdown automata.
Usually, for the problem of code generation, each production represents a machine instruc- tion and i s augmented with an instruction cost. These costs are used during matching in select- ing an optimal sequence of patterns that match a subject tree. Cost analysis may be performed either statically, that is, at pattern processing time, or dynamically, that is, at code generation time.
Some grammars permit static cost precomputation with costs incorporated into the states of a finite-state tree pattern matching automaton? 6-8 We show that for some grammars it i s pos- sible to incorporate costs into the states of the deterministic pushdown automaton (DPDA) so that cost analysis can be performed at pattern preprocessing time. There are input instances for which, the DPDA is much smaller in size than the finite tree automaton generated. Also, for optimal matching, our approach handles a larger class of grammars than that handled by previ- ous approaches.
The results of this work can be applied to the problem of instruction selection for code generation. Previous techniques used are LR tree rewritingb7 and tree pattern match- ing*2-5y 73
Tree pattern matching has several other applications also, such as automatic. theorem prov- ing, implementation of functional languages, term rewriting systems, etc.
2. Tree pattern matching
The two basic approaches to tree pattern matching are the top down and the bottom up meth- ods based on the direction of traversal of the subject tree. Top down tree pattern matching re- duces tree matching to string matching. 1, 4, 11 It uses the string matching algorithm of Aho and
364 IISc THESES ABSTRACTS
corasick.12 For the given set of patterns, trees are constructed and used to form the matching automaton. The algorithm to construct the matching automaton takes time linear in the size of the pattern forest. The matching time is nonlinear and depends on the patterns and the subject tree. Top-down approach has been used for code generation in T'jiang.l3
Bottom-up tree matching generalizes string matching. The basic idea' is to find the set of all matching patterns and subpatterns at all the nodes of the given subject tree. The set of matching patterns at a anode is called the matchset at that node. Given the matchset at all the q children of a q-ary operator at node n, we can find the set of matching patterns at n. Since the patterns are finite, we can precompute all the possible matchsets, encode them and enter the information into tables. There are separate tables for each operator and a q-ary operator has a q-dimensional table. The q indices are the codes for the matchsets at the q children and the entry in the table is the code for the matchset at the node under consideration. The table size is therefore exponential in the maximum arity of the operators. The tables are also rather sparse. chase2 suggested that these tables could be compressed using equivalence relations between matchsets, yielding substantial reduction in sizes. The technique of Chase has been used for input specifications corresponding to regular tree
Regular tree grammars can be augmented by associating a cost with each rule in P. Differ- ential costs can be included into the states of the matching automaton to avoid cost computa- tions at matching t i n x 3 ' This avoids expensive cost analysis at code generation time.
3. New algorit r regular tree
Given a regular tree grammar, one can construct a DPDA as recognizes. In ~antai t" ' an algo- rithm for constructing a DPDA is given, which begins with the LR(0) automaton for the gram- mar, and then modifies it by merging states. A code generator which performs optimal rnatch- ing at code generation time has been automatically generated for the MC68000 series. Cost analysis is performed dynamically, i.e. at code generation time.
We present new algorithms for both regular tree pattern matching, and optimal regular tree pattern matching where a minimum cost sequence of tree replacements is obtained. The first algorithm i s based on the Earley parsing algorithm for context-free languages.'%e tree pars- ing problem is to find the set of all derivation trees in the regular tree gramnar for a subject tree starting from some specified nontenninal. Our goal is to generate a tree parser generator that takes as input a regular tree grammar and returns as output a tree parser for the grammar. The matching problem is to find all possible tree derivations for a subject tree given the regular tree grammar. Consider a new context-free grammar obtained by replacing the right-hand sides of the productions of the regular tree grammar by the post-order listings of the corresponding trees. Since operator arities are fixed, there is a unique correspondence between trees and their post-order listings. Finding all tree derivations in the regular tree grammar corresponds to find- ing all string derivations in the context-free languages. The bottom-up DPDA we construct reports a match of pattern p at pattern p after consuming the more than one production are reductions.
node n of the subject tree whenever it reduces by production for symbol at node n (in post-fix order). In general, reductions by possible at a node and the parser canies the ,effect of multiple
IISc THESES ABSTRACTS 365
The tables that are There are two tables, O((lAI + 2 ~ ~ n ) Z ~ ~ ~ i ~ e ) where IAl is the alphabet size, INTI the number of nonteminals and patsize
the total size of all tree patterns 01- equivalently, the sum of the lengths of all the right-hand
generated are two dimensional irrespective of the arity of the operators, and that encode the DPDA. The combined size of the tables is
- of production. The technique of Chase" when applied to regular tree grammars, requires
space O(IOP1(2 IN7l x mtwarity + ~naxarity x 2 patsize)) where lOPl is the total number of operators of
Using the algorithm just described, om option is to perform cost computations at code genera- tion time using dynamic p rqpmming , as in ~antai t ." Another option is to perform cost corn- putations at preprocessing time and include the information in the states of the automaton.
We extend the technique of Balachandran et aL3 and present a strategy for building a DPDA augmented with costs, thus performing cost co~nputations at preprocessing time. This DPDA will generate optimal code without the need to perform cost analysis at matching time. However, not all regular tree grammars are amenable to this technique. In some cases, the in- put may be such that the cost does not converge at any height. Thus there is no finite automa- ton with precomputed costs for such grammars. In such cases, the only solution is to perform runtime cost computations. We have derived a sufficient condition for convergence. The inclu- sion of cost into the automaton increases the number of states of the automaton and in certain cases, the blow up in the number of states renders it impractical.
The preprocessor and the matcher generator have been implemented. The preprocessor has three parts: pattern input, precomputation and table construction. In the pattern input phase, written using Yacc, the input regular tree grammar i s input from the user and converted into suitable internal representation. The precomputation phase computes certain structures that make the table construction easier. In the table construction phase, the tables representing the DPDA are constructed.
The matcher-generator takes the tables and creates the C-programs that perform matching. The automatically generated set of programs include a Lex function to read the user input, functions to recreate the productions of the grammar and the matchsets in memory and create the tree for the given input, the main matching function which traverses the tree bottom up, the function to traverse the tree top down and select the minimum cost derivation and finally the function to traverse the tree bottom up and display the selected productions.
6. Results a d conclusion
The new algorithm works well in most cases, enabling cost information to be included intu the matching automaton. However, the patterns for the MC68000 generates more than 35,000 states when compared with 482 states when costs are not precomputed. For such examples, cost computation by dynamic programming i s recommended. The patterns for the IBM R6000 yields 1 12 states with costs. For the Intel Pentiurn, the algorithm yields 565 states with costs.
366 IISc THESES ABSTRACTS
The tree pattern matcher can be easily extended to generate code. The pattern input func- tion has to be augmented to read directives to the code generator and also actions for the pat- terns. The bottom-up pass of the matcher has to be extended to emit code depending upon the selected pattern. Instruction scheduling and register allocation phases have to be added to
oenerate efficient code. &
References Pattern matching in trees, J. ACM, 1982, 29, 68-95.
4. A m , A. V. AND GANAPATHI, M. Efficient tree pattern matching: An aid to code generation, Prt,c. 12'" Anrrual Symp. orr Principles of Progwmming Lnnguu,ges, Jan. 1985, pp. 334-340.
5. HATCHER, P. J. AND CHRISTOPHER, T. W. High quality generation via bottom-up tree paitern matching, plat.
13'" A CM Symp. on Principics of' Prograwruhg Lrmgungcs, 1986, pp. 119-129.
Rewrite systslerm, paltent nmching, and rodde generrrtiotz, Ph.D. Thesis, Technical Report UCB/CSD 881423, Computer Science Division, University of Calil'ornia, Berkeley, 1 988.
7. FERDINAND, C., SEIDL, H. AND Tree automata for code selection, Artn I$, 1994, 31, 741-760. W ~LUELM, R.
8. PELEGRI-LLOPART, E. AND GRAHAM, S. L. Optimal code generation for expression trees: An application of BURS theory, Proc. 15'" A CM S ~ t t p . on Prir~cip2r.s of' Pmogmm- ming Larzgmges, 1986, pp. 119-129.
9. GLANVILLE, R. S. AND GRAHAM, S. L. A new rnelhod for compiler code generation, Proc. 5" ACM Symp. an Princip1c.s qj' Progrm~tning Lrrnylir~*qes, Jan. 1978, pp. 23 1-240.
Design of n battorn-rip rrer prrrtern ~natd~ir lg nljiorill~ttt am/ its applicutinrl to code generatiow, M. E. Project Reporl, Department of Computer Science and Automation, Indian lilstitutc of Science, Bangalore, India, l!ZJ6.
1 1. AHO, A. V., GANAPATHI, M. AND Code generation using tree rnotching and dynamic programming, TJIANG, S. W. K. ACM TOPLAS, 1989,11,491-516.
12. AHO, A. V. AND CORASICK, M. J. Efficient string matching: An aid to bibliographic search: Corn- mtn, ACM, 1975,18, 333-340.
Twig reference man~td , CSTR-120, AT&T Bell Laboratories, Murray Hill, New Jersey, 1985,
An efficient context-free parsing algorithm, Conmzrm. ACM, 1970, 13,94-102.
lISc THESES ABSTRACTS
Thesis Abstract (M+Sc.(
predicting radiated noise requires a model of the acoustic behavior of the int&/exhaust sys- tem and a model of the engine cycle s ~ ~ r c e characteristics. This is often dealt with either in the frequency- or in the time-domain analysis. The frequency-domain analysis of mufflers is done by means of transfer matrix method. This i s quite convenient and all commercial automotive
are analyzed in this domain.' Bur it is basically a linear analysis and needs a prior knowledge of the aeroacoustic source characteristics, the evaluation of which remains a chal- lenge. The timedomain analysis of the exhaust systems, on the other hand, i s a fully nonlinear analysis and is usually done using the melhod of characteristics? It offers a completely inde- pendent formalism for muffler mlalysis in as much as it obviates the necessity of a prior knowledge o f the aeroacoustic S O L ~ X characteristics. But this analysis has been primarily de- veloped and used for thermodynamic perforlnance evaluation and not for the aeroacoustic per- formance evaluation of the exhaust system. It is very time consuming. Moreover, this is limited to a few simple muffler elements only. The hybrid approaches developed recently attempt at combining the frequency-domain approach with the finite wave time-domain approach with a view to incorporate advantages of both.'-%~oever, they tend to be rather cumbersome be- cause of the repeated use of the Fourier transform pair, and do not reproduce adequately the peaks and troughs observed experimentally in the frequency spectra of the radiated sound pres- sure level.
The investigation leading to the present work is limited to filling up some of the lacunae that have been pointed out in the preceding paragraph. The extended-tube expansion chamber is analyzed by means of the two as well as three characteristics approach. The simulated results obtained by the two approaches are compared. It i s observed that the two characteristics ap- proach is a very goad approximation to the three characteristics approach, and would therefore be adequate for nodel ling complex mufiler elements. In the hybrid approach, cylinder/cavity is analyzed in the t ime domain to calculate the exhaust mass flux history at the exhaust valve by means of the mcthod of characteristics, avoiding the tedious meshing and interpolation proce- dure. Various results of this approach are validated by means of the method of characteristics and the linear acoustic theory, and various peaks and troughs in insertion loss curves are ana- lytically validated.
inite wave modeling of the exte
In the three characteristics approach (MOC-31, there are two known and seven unknown vari- ables across an extended-tube resonator junction. Mass continuity, energy, momentum and entropy equations along with compatibility equations are simultaneously solved by means of the Newton-Raphson method to evaluate the unknown variables. Interpolation technique ex- plained in Gupta and Munja16 is implemented here to calculate the characteristics variables at
IISc THESES ABSTRACTS
FIG. 1. Exhaust mass tlnx history of an extended-tube expansion chamber muffler.
all the instants. Following Gupta's experimental observations, the blowdown pressure and t en - perature were taken as 0.3436 MPa and 1300 K, respectivelyb7 Typical steady-state results of such a simulation for the extended-tube expansion chamber are shown in Figs I and 2.
From the results presented in Figs 1 and 2, it is evident that in case of mufflers such as ex- tended-tube expansion chamber, the two characteristics approach (MOC-2) is in good agree- ment with the full nonlinear three characteristics sohtion. Differences in the two approaches are more pronounced at higher frequencies only which are not so very important for designers.
for aeroaco e engine ex
Acoustic pressure and volume velocity may be written in terms of the progressive wave vari- ables A and B, where A represents the forward wave and B the reflected wave.' The cone- sponding time-domain variables are denoted by P and Q, respectively, in this work. We can interrelate these variables as follows:
and
Frequency in Hz
FIG. 2. Insertion loss of an extended-tube expansion chamber muffler.
IISc THESES ABSTRACTS 3 69
Neglecting the effects of nonlinear propagation in the exhaust pipe and assuming free expan- sion at the radiation end of the &must pipe, as the effect of radiation condition/impedance would be negligible, the fluctuating pressure would be given by
at the valve end, where .t is the time taken by the forward progressive wave to traverse the ex- haust pipe up and down. It can symbolically be written as
Substituting eqns (1) and (2) in eqn (3) gives
This means that Q at the present instant of time can be evaluated by using the P wave that started .r; seconds ago. This way, meshing and interpolation in the pipe can be avoided and all the attention will be limited to the valve end only.
During the initial phase of the first cycle of calculations, we can take
In the case of a muffler proper, expansion from the exhaust pipe into the first chamber can be assumed to be free expansion. Thus, eqns (4)-(6) may still be applicable with a new value of l,, and hence the time lag z. The same analysis can be used on the intake side also.
At all instants of time, the mass flux m(t) from the valve end is calculated using values of the characteristic variables P and Q. Taking Fourier transform of riz(t), V , .(o) is obtained. This may be combined with the linear frequency domain model of the exhaust system to calcu- late v,, o(w). The convective mass velocity vc, is then used to calculate the power radiated from the tail pipe end.' The procedure is repeated with and without the given muffler to calcu- late insertion loss of the muffler.
Exhaust mass flux history for extended-tube expansion chamber is plotted in Fig. 1. The thermodynamic analysis of an engine with the hybrid approach is quite satisfactory when corn- pared with the method of characteristics. Tedious procedure of meshing, interpolation and solving a number of equations simultaneously are avoided in this approach, which saves much time and effort.
Insertion loss values of the extended-tube resonator muffler are shown in Fig. 3. The results of the hybrid approach are compared with those of the pure acoustic theory, where the source is approximated as a constant velocity source (infinite source impedance). Prediction of the hybrid approach and pure acoustic theory are seen to be close to each other.
Thus, results of the hybrid approach tally quite well with those of the time-domain and fre- quency-domain approaches.
IISc THESES ABSTRACTS
Frequency in Hz
FIG. 3. Insertion loss of an extended-tube expansion chamber muffler.
References
1. MUNJAL, M. L. Acoustics of ducts and mufflers, Wiky Interscience, 1987.
3. JONES, A. D., MOORHEM, W. K. V. AND VOLAND, R. T.
5 . PAYRI, F., DESANTES, .f. M. AND
TORREGROSA, A. J.
The thermodynanzics and gas dynamics qf irzternul corrrbustion mgines, Clarendon Press, 1982.
Is a full nonlinear method necessary far the prediction of radir~ted engine exhaust noise? Noise Corn-01 Engrtg J. , 1986, 26(2), 74-80.
Time-domain-klnite-wave analysis nf the engine exhaust 'system by means of the stationary-frame method of characteristics. Part 1. Theory, S&Zha~~a, 1993, 18, 9 1 1-925.
Tirne-domain-finitewave analysis of the engine exhaust system by means of the stationary-frame method of characteristics. Part II. Computed results and experimental corroboration thereof, SZdhnnri, 1993,18,927-941.
Thesis Abstract (Ph.D.)
Growth and characterization of emstructures by R. Venkataraghavan Research supervisors: Prof. H. L. Bhat and Dr. K. S. R. Koteswara Rao Department: Physics
1. Introduction
Since the discovery of the III-V semiconductors by ~ e l k e r , ' indium antimonide has become one of the prominent materials in the forefront of research. On account of i t s high mobility, r m m w bandgap and ease of preparation, this material has been extensively used in IR detec- tors and high-speed devices. A variety of techniques have been applied to grow it, though the efficacy of these methods varies ide el^.^-^ This work is concerned with the growth and other
IlSc THESES ABSTRACTS 37 1
k crystal growth
0 = Overheated E = Near ~quiliibriurn
Gi = Imposed gradient (Ga)* = Actual gradient; 'oJcaso:
(Gals = Actual gradient; '<case
T~ = Interfacial t~tI3p.j b* CaSQ 1 1
T, = Interfacial temp.; S CaSQ
T, Melting point
372 IISc THESES ABSTRACTS
FIG. 2. Single-crystal InSb boules grown by the vertical Bridglnan technique.
them. Crystals were grown under a variety of conditions by varying temperature gradients and lowering velocities. Three conditions of growth, viz. overheated, supercooled and near- equilibrium, could be identified with the changing gradients and velocities. The crystals grown under the near-equilibl-ium conditions showed very good grain selection, lower dislocation content as well as very good transport characteristics. Figure 2 shows the InSb bode grown by the vertical Bridgman technique.
InSb bulk single crystals were also grown by the horizontal Bridgman technique in a sys- tem built for this purpose. Defect density in the horizontally grown crystals evaluated through etching revealed a pit count of 10~/crn~ as against etch pit density (EPD) of 1 04/cm2 in the ver- tically grown crystals. InSb crystals grown by vertical or horizontal method showed transport: properties ranging between an excess carrier concentration of 1014 to 10'~/crn' and a mobility of 105 crnWs.
in-film growth
Indium antimonide thin films were grown on InSb and GaAs substrates by liquid phase epitaxy (LPE).~ While homoepitaxial thin films were grown by the step-cooled technique, rarnp- cooling was used to grow InSbIGaAs. An additional sacrificial substrate was used during ho- moepitaxy to get films with better surface coverage and good ~norphology (Fig. 3).
FIG. 3. Photographs of epilayer and interface of 1nSblGa.A~ structure grown by LPE technique.
IlSc THESES ABSTRACTS
lnSb GaAs
In the growth of indium antimonide on gallium usenide, it is seen that the morphology of the nucleating island shows well-defined facets depending on the growth conditions. The nu- clei either showcd triangular or polygonal morphology (Fig. 4) depending 011 the ,imposed s ~ - persaturation. InTISb, a ternary of In, T1 and Sb, extends the range of usefulness of InSb-based detectors to the long w:tvelength infrared region (LWIR) which makes such structures all the more promising. LPE has been used to grow InTlSbIGaAs and the ensuing heterostructure shows a shift in the absorption band to longer wavelengths (r 10 pm).
Thin films of InSb have been grown 011 CdTe substrates by pulsed laser deposition tech- nique. The filtn quality and stoichiometry have been studied for two laser fluences and valious g~-owth pruumeters. In fil rns grown at higher laser fluence (280 mT/pulse), pitted morphology was observed but the problem could be alleviated by reducing the laser fluence. Extensive studies were made on the interfticc of InSbfCdTe heterojunction and the presence of an inter- metallic indium tcllurid~. was identified. Photoluminescence spectroscopy has been used to infer the presence of InzTc3 as against InTe. Further, therrnoe~lergetic calculations at the inter- face have been used to asccrlain the same. The studies on the interface have led to further im- pmvernent i n growth to y ielcl f ilms of optimum '"
Amorphous silicon layers were deposited by glow discharge plasma method on InSb sub- strates. Ohmic magnesium contacts were evaporated on the surface to make metalla-Si:H/InSb devices. I-V characteristics of the device showed properties similar to InSb surface barrier diodes. However, at I-evcrsc: bias, the device switches from a low to a high resistance state (Fig. 5 ) . The tl~reshold swi tchini~ is seen to shi f t with temperature and resembles ovonic switching. A simple band model has becn used to explain the observed effects in a - ~ i : ~ ~ n ~ b . ~
0.2 0+4 6 0.8 1 Reverse Voltage (V)
FIG. 5 . I -V characteristic of a-Si:H/LnSb heterostructure showing switching.
374 IISc THESES ABSTRACTS
References
WELKER, H. 2. Nntur$ A, 1992,7,744.
VENKATARAGHAVAN, R., RAo, K. S. R. K. J. P ~ s . D: Appl. Plrys., 1997, 30, LG 1.
AND BHAT, H. L.
VENKATARAGHAVAN, R., RAO, K. S. R. K. J. O y t . Growth, 1998,186,322. AND BHAT, H. L,
VENKATARAGHAVAN, R. et al. Bull. Mater. Sci., 1998, 22, 133437.
VENKATARAGNAVAN, R. et nl. Bull. Mater. Sci., 1996, 19, 123.
VENKATARAGHAVAN, R., RAO, K. S. R. K., Physic0 Stalus Snlidi (a) , 1997, 163, 93. HEGDE, M. S. AND BHAT, H. L.
VENKATARAGHAVAN, R., RAO, K. S. R. K., Solid St. Comrnun., 1997, 102, 759. BHAT, H. L., PAL, S. AND DUBEY, G. C.
Thesis Abstract (Ph. D.)
lectrical and t asses by K. Ramesh Research supervisors: Prof. E. S . Rajagopal and Dr. K. S. Sangunni Department: Physics
Binary Ge-Te chalcogenide glasses exhibit interesting structural phnsc transitions like sharp discontinuous semiconductor-to-metal transition, double T, and double T,. under high pressures and temperatures .' The stability, rigidity and structural connectivity of chalcagenide glasses are found to change in a special way when metal atoms are added.' For example, the addition of Ag to selenide glasses introduces ionic bonding and the conduction changes from electronic to ionic conduction. In contrast, Ag with germanium tellurides results only in covalent bonding and the conduction is purely electronic .3
This work deals with thermal and electrical resistivity studies of Ag- and Cu-doped Ge-Te glasses as function of composition at high pressures and temperatures. The aim of this work i s to study the network connectivity of Ge-Te glasses and the pressure-induced semiconductor-to- metal transition when the glasses are doped with Cu and Ag atoms.
2. Experimental
Bulk CuJ3e15Te8s-x (2 S x I 10) and AgxGelsTe8s-,T (2 5 x I 2 1.5) glasses are prepared by melt- quenching method, The amorphous nature of the prepared samples is confirmed by X-ray dif- fraction (XRD).
IISc THESES ABSTRACTS 375
3.1. DSC studies on Cu-Ge-Te and Ag-Ge-Te glasses
Both Cu-,Gel~Tegs-.~ and AgU,Ge~~Te~~, glasses show a single glass transition T, followed by a single crystallization T,. upon heating whereas binary Ge-Te glasses exhibit a double T and
6 double T,.. The composition dependence of T, of Cu,Ge15Te8s-x samples exhibits a minimum at x = 5, whereas the Tc and E,. show maximum at x = 5. X-ray studies on the samples annealed at their respective crystallizatioi~ temperatures yield only hexagonal Te and rhombohedra1 -Ge-Te phases. This indicates that Cu atoms do not form their own structural units with Ge or Te. It is likely that Cu atoms randomly replace Ge as their atomic radii, electronegativity and coordina- tion are similar."
The variation of the thermal parameters T,, T,, E,. and thermal stability (OT) with composi- tion for Ag,Cel STe8s-.x glasses show a minimum at x = 5. Ts shows a maximum at x = 18.5, whereas Tq, Tc, E,. and KIT exhibit a minimum at x = I 8.5. The X-ray patterns of the devitrified samples can be indexed to cubic Ag8GeTe6, cubic GeTe4 and hexagonal Te phases. According to the present results the network structure of these glasses can be viewed as having three structural regions consisting of Te chains, Ag-Te and Ge-Te tetrahedral units. For x I S , flexi- ble Te chains are more active resulting in a minimum in Ts at x = 5. In the second region (7.5 i x 2 15) Ag-Te and Ge-Te tetrahedral units influence the structure more and the struc- tural rigidity grows as these units are rigid causing an increase in T,, Tc and E.. At x = 15, Ge- Te and Ag-Te units are equally populous; so a crossover from Ge-Te to Ag-Te results in a pla- teau region around x = 15.
The thermal crystallization studies provide a clue to understand the different ranges of glass formation in Cu.,Ge lsTexs-.y (0 i x < 10) AgYGel 5Te8S, (0 S x 1 23). To improve the glass-forming ability the added third element should increase the network connectivity by in- teracting with the existing parent glass matrix. The added Cu atoms do not improve the net- work connectivity of the parent Ge-Te glass matrix as evident from the X-ray studies. The ad- dition of Ag to Ge-Te glasses promotes the network connectivity resulting in a wide range of glass formation!
It has been suggested that the network of covalent network glasses undergoes a transition f ~ n floppy to a rigid structure in a percolative manner (rigidity percolation). This change oc- curs at a critical composition corresponding to the average coordination number Z,,, = 2.40. h u s u a l changes in the properties of covalent network glasses are expected at this critical composition.' The unusual change in T,, Tc3 Ec and OT at x = 5 is the indication of the rigidity percolation in these glasses. In addition to the rigidity percolation, a chemical ordering also
376 IISc THESES ABSTRACTS
occurs at a higher average coordination number. The unusual change at x = 18.5, in A ~ ~ - G ~ ~ ~ TesS_, glasses is due to chemical ordering in the structural networl~.~
3.2. High-pressure stzidies on Cu-Ce-Te and Ag-Ge-Te glasses
The resistivity as a function of temperature (300-450 K) of C ~ S G ~ ~ S T Q W glass at various pressures shows a subtle change around the glass transition temperature. 'ls, measured from the log(p/p, ,) vs 1/T is found to decrease with an increase in pressure. The nod el developed by deNeufville and Rockstad finds a linear relationship between glass transition temperature and optical band gap (E,). This model can be used to explain the observed behaviour. In chalco- eenide glasses, OE and E, are related by n E = E42. Decrease in n E at high pressures in& LC'
cates a reduction in E,?. The decrease in E, shifts the T, of Cu5GelsTeso to lower values.
Under lligh electric field Cu,GelsTess-,r and Ag.yGelsTe8s-., glasses are found to exhibit memory switching. The memory switching and the composition dependence of switching fields are explained on the basis of thermally induced phase transitions in these glasses at high fields. The switching fields of both the glasses decrease with the composition and exhibit anomalies at compositions corresponding to rigidity percolation and chemical ordering thresholds.
The present studies show that the thermal, electrical and structural properties of chalco- genide glasses as well as the anomalies expected at the topological and chemical thresholds are to a large extent determined by the atomic size, relative covalency and coordination of the added impurities.
References
3. FERHAT, A*, OLLITRAULT-FICHET, R., MASTELARO, V., BENAZETH, S. AND
RIVET, R. J.
Phil. Mag. R, 1988, 57,49.
Glassy semiconduc-[om, Plenum Prcss, I 985, p. 436.
J. Plzys. IV, t 992, C2, 201.
4. RAMEsH, K., ASOKAN, S., SANGUNNI, K. S . J. Phys.: Condmsfd Matter; 1996, 8, 2755; Pl1y.s. CC'm Glasses, AND GOPAL, E. S. R. 1996,37, 17.
378 IISc THESES ABSTRACTS
conjugate may be the erythrocyte receptor for PPRV. The RDE for PPRV has been shown to be a neuraminidase (NA) and conclusively proven by showing NA activity with purified H
3 -2. Phylogenetic analysis of PPR V based an the hemaggluti~lill-neur~~yt~it2 idase (JJN) gene and protein sequences
The H gene of various PPRV isolates was amplified by RT-PCR and the products were cloned. PCR products corresponding to 2.0 kb (the full length gene) were obtained for 8 African and 3 Indian isolates. Complete sequence of both the strands was determined. Partial amplification of tl-le H gene could be achieved for 4 Indian isolates and their sequence was also determined. A large single open-reading frame in the H gene encodes the 609 amil~o-acid-long H protein. The H protein is a type II surface glycoprotein, with the hydrophobic transmembrane domain at the amino terminus of the protein. Multiple sequence aligmnents were computed to estimate distances. The distance data obtained was conected using Kimum-2 parameter method and Gamma distances using the MEGA software for reconstructing phylogenetic relationships. The distallces were grouped according to the neighbor-joining method. Bootstrap analysis was pel-- formed to esti~nate the robustness of the grouping of the clades. Based on such rigorous phy- logenetic analysis the relationships between various circulation isolates of PPRV were recon- structed.
Both NA and HA activities were conclusively demonstrated using the purified PPRV H protein. In the PPRV H-deduced amino-acid sequence, residues corresponding to viral neura- minidase active site could be identified. The 2D model generated for PPRV H using various secondary structure prediction methods was folded into 3D using homologous MV-W nlono- clonal antibody epitope data. With the determination of molecular and biochemical basis of PPRV H neuraminidase activity, the use of the property 'lack of neurruninidase activity' as a parameter to group viruses under Morbillivirus should be discontinued. The H gene showed scattered variation of up to 13.4% at the nucleotide and 9.4% at thc amino-acid level. The mu- tation rate in the PPRV H gene was estimated to be 6.65 x 1 substitutions per rnlcleotide position. Predominant occurrence of amino acid changes whenever there i s a nucleotide change demonstrates the role of immune selection in the evolution of the virus. The extent of variation observed in PPRV is similar to that observed foi- RPV or CDV. Unlike RPV, the PPRV isolates grouped into four different lineages are not always geographically restricted. N o epidemiologi- cal center could be identified for PPR in Africa or India. The numbcs of PPIiV lineages in Africa i s four while that in India is only three.
References
The order Mononegavirales: evolu t ionxy re1 ut ions hips and mecha- nisms of variation. In Molccitlir r hmis c!f' ~v'rlrs evohrtion (A. J+ Gibbs, C. H. Calisller and F. Garcia-Arenul, cds), Cambridge University Press, 1995, Vol. 29, pp. 426337.
IISc THESES ABSTRACTS 379
2. &MACHANDRAN, S., SHAILA, M. S. Hernagglutination and hemadsorption by Peste des petits ruminants AND SHYAM, G. virus (PPRV), Immunobiology of virus infections, Proc. 3r" Con-
gress Eur. Soc. Vet. Viral. (M. Schwayzer et nl., eds), 1995, pp. 5 13-5 15.
Thesis Abstract (Ph. D.)
by Joshi Chirag Vinaychandra Research supervisor: Prof. P. Ajit Kumar Department: Molecular and Cellular Biology
Interfkron gainma (IFN-y) exerts a variety of biological effects, which include inhibition of cell proliferation, immunomodulation, induction of antiviral state, apoptosis, and differentiation.' The IFN is believed to mediate these pleiotropic effects through the modulation of expression or activity of a variety of genes and proteins. Binding of IFN-y to i t s specific receptor on the cell surface initiates a cascade of events, which result in the activation of the latent cytoplasmic transcription factor, STAT 1 a through phosphorylation. The phosphorylated STAT 1 a mi- grates to the nucleus and binds to Gamma-IFN Activation Sequences (GAS), which are present upstream to the IFN-inducible genes, leading to their expression.2 Although several intracellu- lar changes that occur in response to IFN- y have been studied, the exact mechanism by which the IFN arrests proliferation of cells is not yet known. This work is an attempt to characterize the nature of IFN- y-mediated growth arrest of human promonocy tic leukaemic U937 cells. Previous studies from this laboratory have shown that IFN-yinhibits the growth of nontrans- formed, human amniotic WISH cells by arresting cell cycle progression at the GUS boundary in a reversible manner.3 In the present, but parallel study, investigation was focussed on the growth inhibitory action o f the IFN on the human prornonocytic leukaemic U937 cells, a trans- formed cell line unlike WISH. The study was undertaken with the objective of finding out whether the growth inhibitory action of IFN-yon a transformed cell line such as U937 would be similar to that exerted on the nontransformed WISH cells. Therefore, the principal goals of this study were to determine: (i) whether the growth inhibition of U937 cells by FN-y oc- curred in a cell-cycle phase-specific manner; (ii) whether the arrest of proliferation was re- versible, and (iii) the characteristics of the growth arrest o f U937 cells by IFN-I/.
2. Results an
U937 cells were confirmed to be sensitive to the growth inhibitory effect of IFN-y at a mini- mum concentration of 100 unitdm1 by the determination of viable cell number using the trypan blue dye exclusion method. The cells were found to get growth-arrested from the 24th hour of treatment onwards. Viable cell count of the cells, which were released from 48 hours of IFN treatment (after complete growth inhibition), did not show any increase indicating that the growth inhibition was not reversible unlike in the case of WISH cells. This observation was confirmed by the finding that the cells, which were released from growth inhibition, did not incorporate 3~-thymidine over a time period of 48 hours after release, indicating the
380 I ISc THESES ABSTRACTS
Since ~ 9 3 7 cells are quasidifferentiated myeloid cells, i t was likely that IFN treatment led to the induction of tel-minal differentiation resulting in an irreversible arrest of proliferation. In other words, C g;rowth al-rest might have been a consequence of the induction of terminal differ- entiation, as the two phenomena are mutually exclusive. In order to verify this possibility, sev- eral parameters, such as morphological changes, respiratory metabolic changes, and biochemi- cal changes, were used to monitor for myeloid differentiation. Phorbol- 1 2-Myristate- 13- Acetate (PMA), which is known to induce terminal differentiation in U937 cells, was used as the positive control. While PMA treatinent did bring about teminal differentiation-related ~norphalogical changes in U937 cells, IFN-ytreatment did not result in any o i such changes. Si~llilarly, biochemical changes such as generation of free radicals and reducing power as monitored by nitro blue tetrazolium reduction could be observed upon treatment of U937 cells with PMA but not with the IFN. Treatinent of cells with PMA and IFN together brought about all the above changes in a much shorter period. Thus it was established that 'IFN- ytreatment does not effect terminal differentiation on its own in U937 cells. However, the presence of IFN- yalong with PMA was to hasten the PMA-mediated terminal JifFcrentiation process.
Another possible reason for* the lack of resumption of growth coi~ld bc that IFN- y might have induced apoptotic or necrotic cell death. Several parameters such as trypan blue and propidiu~n iodide dye exclusion, DNA ladder& Hoechst 33342 and propidiu~n iodide dual- staining of cells, and flow cytometric determination of rnitocl~ondrial tr-ansmclnbrune potential, were used to find out this possibility. Etoposide VP-16, a known inducer of apoptosis it] U937 cells, was used as the positive control in these experiments. Although about 30% (of the cell population could undergo some of the changes associated with apoptosis and necrosis, majority of the population did not show any such changes. This 'indicated that the isseversibi lity of inhi- bition of growth of U937 cells by IFN-ywas not due to induction of cell death.
The growth-arrested cells, which were released from ZFN- y block, could not be stimulated to proliferate by enriched growth medium or by co-culture with untreated cells. However, these cells could be induced to undergo differentiation or apoptosis using PMA or etoposide, sespec- tively . The rate of protein synthesis in IFN-mediated growth-arrested ccl i s wits seduced by 40%. These results indicated that probably IFN-ytreatment had effected a state of quiescence in U937 cells from which they can be induced to differentiate or die, but crm-mt be induced to resume proliferation.
Simultaneous study on the possible modulation of expression of genes such as cyclins, cdks, cdk inhibitors, tumour suppressors, and so on, that are likely to be involved in growth regulation, did not show significant changes in the Northern analysis of total RNA derived
IISc THESES ABSTRACTS 381
from gowth-arrested cells. However, Rb gene was found to be upregulated by 1.7 fold in 12- 24 h of IFN-I/ t reatn~nt . Unlike in other reported systems, p21 was not upregulated in U937 cells following IFN- y treatment.
The observations made for U937 cells are strikingly different from those made for WISH cells which are arrested by IFN-)I in a reversible, cell-cycle phase-specific manner at the Gl/S boundary. The induction of quiescence could represent a novel mechanism for the inhibition of proliferation by I F N - ~ ~
PESTKA, S., LANGER, J. A., ZOON, K. C, AND SAMUEL, C. E.
DARNELL, J. E* Jr.
SUPRIYA, P,, JQSNI, C. V, AND
AJITKUMAR, P.
JOSHI, C. V., SLJPRIYA, P. AND
AJZTKUMAR, P,
interferons and lheir actions. A. Rw. Biockem., 1987,56,727-757.
STATs and gene regulation, Science, 1997, 277, 1630-1 635.
IFN-y inhibits growth of WISH cells in a cell cycle phase-specific manner, J. Interjeron Cytokine Res., 1998, 18,2 15-2 1 8.
Growth inhibition of human promonocytic leukaemic U937 cells by interferon gamma is irreversible and not cell cycle phase-specific, Cytokine, 1999, 11,673-678.
Thesis Abstract (P11.D.)
r virus resistance by C. S. Sree Vidya Research supervisor: Prof. G. Lakshrni Sita Department: Microbiology and Cell Biology
Tomato belongs to the family Solmaceae and is considered a very important vegetable crop worldwide. Tomato plants are severely infected with several viruses, bacteria, fungi, nema- todes and insects leading to tre~nendous crop loss. Conventional breeding techniques to obtain disease resistance are too laborious and inadequate. To introduce desired gene or a set of genes by conventional breeding requires a sexual cross between two lines and then repeated back crossing between the hybrid offspring and one of the parents until a plant with the desired characteristic is obtained. Production of disease-resistant plants by genetic engineering has increased to obtain resistance against insects, fungi, bacteria, virus and nematodes. The present study was initiated with the aim of genetically manipulating tomato plants to produce disease- tolerant plants. Two approaches are adapted. One is by overexpressing defense-related genes, which are expressed during infection called pathogenesis-related (PR) genes to obtaiwbroad range resistance. It has been reported that in many plants overexpression of defense-related genes like chitinase,' coexpression o f chitinase and glucanase, polygalacturonase inhibitor2 brings protection against disease. The second approach is the pathogen-derived resistance which involves transferring the part of the viral genome-like coat protein (CP)? movement protein4 and altered form of replicase5 to obtain viral resistance.
IISc THESES ABSTRACTS
We have undertaken both the above-mentioned approaches to obtain diseaseresistant plants. Coat protein gene (CP) of Physalis mottle virus (PhMV) (provided by Dr Sslvithri et al*) was used to obtain pathogen-derived resistance in tomato plants. Pathogenesis-related gene from tomato has been cloned to obtain host-mediated resistance. These results are presented here.
21. Cloning and characterization of diseuse-resistance gene
To clone pat;h.hogenesis-related (PR) genes from tomato, tomato leaves were treated with salicylic acid @A) to induce PR genes. Total RNA was isolated and 1tlN"4A was purified. cDNA library was constructed in AMOSElox (Amelasham) vector system. Recombinant plaques were subcloned to plasmid MOSElox by automatic subcloning and random analysis of cDNA library was performed. Putative recombinant plasmids of size 700 bp to 1 kb were fur- ther taken for sequence analysis which led to the identification of full-length 700 bp intracellu- lar PR (IPR) gene and named as TSI-I (tomato stress induced-I). The deduced amino-acid sequence of TSI-1 codes for a 178 amino-acid polypeptide with a 534 bp open reading frame starting with the first translation initiation codon ATG at position 40 and ending with a stop codon T M at position 574. TSI-I is acidic and the calculated isoelectric pH is 5.8. Its pre- dicted molecular weight is 20.4 ma. TSI-1 shows maximum ho~nology to potato IPR genes, i.e. 71 % to STH-2 and STH-21 at nucleotide level which are expressed against fungal infec- tion, 30%-40% with the tree pollen allergens and other IPR members from soybean, pea, bean, alfalfa, lily, carrot and asparagus. Tomato genomic DNA was digested with various restriction enzymes and subjected to Southern analysis which revealed that TSI-I belongs to multigene family like other reported IPR members. Northern analysis was performed to analyze the in- duction pattern of TSI- 1 for SA. The levels of TSI-1 transcripts increased as the concentration of SA increased and were maximal at 10 mA SA after 48 hours. TSI-I was not expressed for water treatment and an extremely faint signal was obtained for mild concentration of SA. Ex- pression was very well correlated with the concentration of SA, which indicates that SA in- duces expression of TSI- I . After stripping and reprobing with 1 8s rRNA, similar signal inten- sities were obtained which indicates equal loading of RNA in all lanes. Northern ~rnalysis was pelfor~ned with RNA isolated from Fusariurn mysporurn fungal-treated tomato leaves and fresh tomato leaves as control. Highly intense signals were obtained in fungal-treated leaves after exposure for 24 h but no transcripts were detected in fresh tomato leaves even after expo- sure for 2 days.
2.2. Genetic transformation of ?ornuto with Phyalis mottle virus coat pro.~t~i~z ( PhMV CP)
Tomato plants are infected by many viruses and PhMV virus is one ~ ~ m o n g them. Tomato plants infected with PhMV develop chlorotic and necrotic symptoms and hampers the growth of plants. Viral resistance was obtained by genetic transformation of tomato with CP gene of PhMV through Agrobacterium-mediated transformation.
Initially, efficient regeneration was obtained through organogenesis by culturing two-week old cotyledonary leaves on the MS medium (MS) supplemented with 0.5 mg/l thidiazuron (TDZ). After 3 weeks, the callus which appeared at the cut ends was subcultured on the fresh
IISc THESES ABSTRACTS 383
MS medium with 0.5 mgfl TDZ for differentiation into shoot buds. The shoot buds eloqpted to shoots in the same regeneration m x h m and subsequently transferred to half MS medium supplemented with 0.5 mg/l illdole acetic acid (IAA) for root induction, Fully regenerated plants were transferred to soil after acclimatization in tissue culture conditions, After establish- ing regeneration system, A~r~l?d~terium-mediated transformation was standardized using GUS marker genes. Tomato leaves were transformed with pBI121 binmy vector and transgenic plants were selected for kallalllycin resistance (50 lngll) and the transformed plants were sub- jected to GUS assay and PCR analysis using npt II gene-specific primers to amplify 700 bp fragment of rzpt II. After establishing the transformation with GUS, marker gene tomato plants were transformed with the coat protein gene of PhMV. CP gene of PhMV was cloned into pB1121 binary vector under CaMV 35s promoter by deleting GUS gene and transformed to &robactericrm by freeze thaw method and tamatto was transformed with CP gene using the same conditions that were used to transfer GUS gene of pBI121 vector. Transformed shoots were screened for kanamyoin resistance phenotype and subjected to molecular analysis. 600 bp fragment of CP gene was amplified using CP gene-specific primers. Southern blot analysis revealed the integrity of the transgene in tomato genome. Total proteins were extracted and Western blot analysis was performed using monoclonal antibodies to CP. All the transgenic plal~ts showed expression of CP protein. To plants were self-pollinated to produce fruits. TI seeds were sawn in green house and challenged with PhMV to characterize the degree of resis- tance obtained due to CP expression by observing symptom development.
Representative SA-induced cDNA library was constructed in AMOSElox vector system. Ran- dom analysis of recombinant plilsmids revealed the presence of tomato TSI- I . Southern blot analysis revealed TSI-1 is a member of a gene family. They occur as ~nultigene families in potato, carrot, nlhlfa, soybean, potato and pollen allergens to which it shows considerable ho- mology. IPR genes are reported to express during stress conditions like pathogen invasion, wounding, asachidonic acid tmd alxicisic acid treatment. Exogenous application of SA induces PR genes in many systenx and they act as pathogenic signals. Exogenous application of SA clearly shows that SA has significant induction over TSI- 1. We would see less-intense signal after treatment with SA for 12 11 31 1 mM and the expression increased as the concentration of SA increased and clear signals were obtained in all the SA-treated leaves. RNA was isolated from fresh leaves and probed with TSI-I to check the constitutive expression of TSI-I, which did not give detectable signal in Northern analysis. It behaves like any other SA response genes expressed during hypersensitive reaction. The function of this TSI-1 remains unknown. A stretch of amino acids from 83rd to 126th of TSI-1 which i s considered as the signature mo- tif l~ad homology with the sttucturally related Bet VI-type pollen allergens and other related IPR proteins from plants. The recent reports on white birch pollen and grass allergens are known to have RNase activity. It could be possible that TSI-1 may be involved in degradi~~g invading pathogenic RNA. We conclude that tomato IPR is a low-molecular-weight protein expressed against stress. It is organized as a gene family in tomato genome and highly induc- ible by salicylic acid. This gene can be overexpressed in tomato to obtain broad-range resis- tance against pathogens. The nucleotide sequence data reported will appear in EMBL nucleo- tide sequence database under the accession number Y 15846.
384 IISc THESES ABSTRACTS
As a second strategy to develop resistant plants, transformation of tornat0 plants harboring PhMV CP gene was achieved. This is the first report on tomato (Indian variety) sl1owing resis- tance t~ PhMV (tymo viral group). Regeneration on tomato was attempted t~ obtain quick and efficient protocol, which can be used for further transformation experiments. Several plant- growtll hormones were tried using different explants such as cotyledons, hypocotyl and leaves for regeneration. Among the several cytokinins and auxins tried best regelleration was obtained in 0.5 rng/l of TDZ. Cotyledonary leaves inoculated in MS medium supplemented with 0.5 mg/l TDZ gave high nu~nber of shoots (5 per explant). Established plants were obtained within a period of 3 months. Standardization of transformation was carried out by Agrobucte- rz'urn-mediated transformation harboring pBI121 binary vector. Trnnsformation of tomato with PBI 12 1 binary vector was carried out. to standardize the conditions suitable for tomato trans- formation. Initial standardization was carried out with the GUS repol-ter gene. Only the trans- oenic plants showed GUS activity showing blue colour and the control plants were ~olourless a showing abserlce of GUS activity and it is specific in transgenics due to integration of GUS oene. Ful*ther confirmation was done by PCR analysis using npt gene-hpcifk primers. Ex- b
petted 700 bp size products were obtained showing further confirmation of integrated T DNA in the tomato genome. Out of 20 plants growing in kana~nycin medium, we obtained 6 inde- pendently regenerated To plants expressing high-level CP protein. Transgenic nature was con- finned by molecular analysis such as PCR and Southern l~ybridization. Intensity and pattern of integration varied among plants showing the occurrence of random integration. All the T* plants analyzed showed the same level of CP expression. Southern and ELISA analysis were undertaken in TI plants to check the successful inheritance of CP. Interestingly, Southern analysis showed a similar pattern of signals for plants originating from n singlc fruit. ELISA results showed that out of 30 plants analyzed, 20 expressed CP that is close to the Mendelian segregation ratio of 3: 1. The 20 plants inoculated with the virus initially stlowed low levels of virus accumulation but virus titer increased 30 days after inoculation. Twelve out of 20 plants developed symptoms as control plants. In 8 of the transge~~ic plants, the infection was however restricted to local lesions (chlorotic spots) on the inoculated leaves only. These plants ex- pressed high levels of CP initially. In some of the low CP-expressing plants, systemic infection was avoided by cutting primary leaves. Transgenic plants expressing low levels of CP accumu- lated more virus compared to transgenics which expressed high-level CP and the vir~ls titer was low compared to nontransgenic plants. It i s possiblc that thc constitutively expressed CP which forms the capsid blocks the site for disassembly of inoculated virus and ribosome bind- ing to undergo initial translation needed for replication. The possibility of CP RNA interacting with the negative strand of the replicating virus to interfere virus replication crinnot be ruled out since tobacco plants transgenic to PhMV CP were symptom free after infecting with PhMV JWA (Ranjit et d, unpublished report). This can be applied to tomato since both tonlato and tobacco belong to Solanaceae. This method can be exploited for frequent and successful pro- duction of transgenic plants harboring other CP gene and pathogen-derivcd gcncs of harmful viruses to produce disease-resistant plants.
1. BROGUE, K. et al.
2. ZHU, Q. et al.
IISc THESES ABSTRACTS 3 85
In Advances in niolecular ge~zrtics of plant-rnicw.2-w imwncrions, Vol. 3, (M. Daniels ~t a]. eds), Kluwer Academic, 1994, pp. 399- 402.
J. Theor. Biol., 1985,113, 395405.
Thesis Abstract (Ph. D.)
C6W5) by M. Ganesan Research supervisor: Prof. S. S. Krishnalnurthy Department: Inorganic and Physical Chemistry
Diphosphinoamines of the type RN{PX212 (R = allcyl or aryl; X = F, C1, alkyl, aryl, alkoxy or aryloxy) are versatile ligands capable of forming a variety of complexes with transition metals in which they exhibit chelating, bridging and monodentate modes of coordination.'" In the present investigation, several new aspects of the organometallic chemistry of diphosphino-
nds bearing electron-withdrawing substituents, MeN {P(OR)2)2 (R = CH2CF3, L have been explored. The topics investigated are: (1) Reductive carbonylation of
CoCL, ( 2 ) Synthesis of Group 10 metal complexes, and (3) Synthesis of Pd-Mo and Ni-Mu heterometallic comnplexes.
Reduction of CoC12 by NaBH4 in ethanol in the presence of MeN{P(OR)2)2(R = CH2CF;) and carbon monoxide at one atmospheric pressure gives an unusual mononuclear complex, [co(CO> {P(OR)2(NWMe) {P(0)(OR)2) {P(0)(OR)2 ) {P(H)(OR')2 )I, (R = CH2CF3, R' =CH2
) formed by cleavage of a P-N bond and by franks-esterification. When the reductive
2.2. Group 10 metal complexes
Treatment of Ni12 with MeN{P(OR)2)2 ( R = C6H5 or CH2CF3) gives the mononuclear [N~I&&M~N{P(oR)~)~)] (R = C6Hs) (5) and dinuclear [Ni212(p-I)2(p-MeN {P(0R)2)2)21 (R = CH2CF3) (6) complexes respectively. The reactions between MeN(P(OR)2)2 (R = CH:!
386 11% THESES ABSTRACTS
The reaction of 5 with [ M ~ ~ ( ~ ~ - - C ~ H ~ ) ~ ( C O ) ~ - gives the heterodinuclear complex [($- CSHs)Mo (I)(p-I)(p,b-CO) ( p M e N {P(OR)2 ),)NiI] (R = ChHs) (I ) and the rnononuclear cum- plex [ ( r l S - ~ s ~ , ) ~ o ( ~ ~ ) ~ ( $ - ~ e ~ {P(oR)~)~)] (R = C&) ( ). The reactions of the dipalla- dium complexes [Pd2C12(p.l-MeN{P(OR)2)2)21 (R = CH&F.? or ChH5) with [MO~($- C5H5)2(C0)6] in boiling benzene give Mo-Pd dinuclear and MoPd2 trinuclear complexes [(I$ C~Hs)(CQ)M~(~-Me~{P(OR))~)~)~PdC~I (R = C H K h or 5 ) (w, [ ( q 5 - ~ s ~ s ) ~ ~ ( p 3 s b - C0)2(p-MeN{P(OR)2)2)2Pd2C1] (R = CH2CF3 or C6H5) [ and [ ( I ~ ~ - C ~ H ~ ) ( C ~ ) M O ( ~ , ~ - CO)(p-Cl)(p-MeN {P(OR)2)2)PdC1] (R = CH2CF3 or C&) as well as the Mo(1I) corn- plexes [ ( q 5 - ~ S ~ 5 ) ~ o ( ~ ~ ) ~ l ( q 2 - ~ e ~ { ~ ( ~ ~ ) ~ ) 2 ) ] (R = CH2CF3 or C6H5) (15) which call be separated by column chromatography. The reaction between [ M ~ ~ ( ~ ~ - C ~ I I ~ ) ~ ( C O ) ~ : ] and MeN{P(OR)2)2 (R = ChHS) in boiling toluene gi the dimolybdenul~~ complex, TMQ~C~)&~~-C~HS)Z(~~ '-M~N{P(oR):! ) 2 h 1 (R = ( 3 3 5 )
The complexes have been characterized by C, H, N elemental analysis and IR and NMR spectroscopic data. The structures of 1, 2, have been con- firmed by single-crystal X-ray diffraction studies.
IISc THESES ABSTRACTS
All the reactions were carried out using standard Schlenk-line techniques and solvents were purified prior to use. The ligands 2 were synthesised by the treatlnent of J~ leN(Pc1~)~ with CF3CH20H and C&OH in the presence of Et3N in diethylether. Metal precursors such as [ M ~ ~ ( ~ ~ - C ~ H ~ ) ~ ( C O ) ~ and [Pd2C12(p- 2)2 were prepared by literature procedures. The NMR spectra were recorded using Bruker AMX-400 or ACF-200 spectrometers. Single-crystal X-ray diffraction data were collected ushg Enraf-Nonius CAD4 diffractometer.
. Conclusions
The results obtained in the present investigation demonstrate the considerable scope and versatile nature of diphosphazane ligands for the synthesis of a range o f transition metal complexes, in particular heterodi- and trinuclear complexes. Significant differences in the reactivities and the structural features of the resulting complexes are observed between the two ligands MeN {P(OR)2 )2 (R = CH2CF3 or CbH5). The trifluoroethoxy-substit~~ted Iigand shows a more pronounced tendency to form bridging type of complexes rather than of chelate type formed by the phenoxy-substituted ligand. The structural data on the complexes show that the mean M-P distances are shorter than those found in dppm or dppe complexes. This trend is in support of strong n-acceptor nature of the ligands which i s also reflected in the magnitude of carbonyl-stretching frequencies observed for these complexes. The P-N distances fall in the range 1.63-1.68 A and are shorter than the accepted value for a P-N single bond distance (1.77 A) found in ~-T~N'-PO-~. The mean P-N-P angles of the bridging diphosphazanes vary from 113 to 122O; in the chelate complexes, the mean P-N-P angles are 96-97". In spite of this large variation in the P-N-P angles, the planarity around the nitrogen is maintained in all cases as observed in numerous other diphosphazane complexes. The structural data for complexes
IISc THESES ABSTRACTS
throw much light on the nature of metal-metal containing semibridging carbonyl ligands interactions.
I . BALAKRISHNA, M. S., REDDY, V. S . , KKISHNAMURTHY, S. S., NIXON, J. F. AND
B URCKETI. ST LALIKENT, J. C. T. R.
2. Wllr, M. AND ROESKY, H. W.
3. BHATTACHARYYA, P. AND WOOLLINS, J. D.
4. KRISHNAMURTHY, S. S.
5. KING, R. B.
6. GANESAN, M., KRISHNAMURTHY, S. S . AND
NETHAJI, M.
7 . GANESAN, M., KRISHNAMUI~TTHY, S. S., NETNAJI, M. AND RAGHURAMAN, K.
Chem. Rev., 1994,94, 1 163.
Polyhedron, 1995, 14, 3367.
Proc. Indian Acnd. Sci. (Chem. Sci. ), 1 996, 108, 1 1 I .
Acc. Chem. Res., 1980, 13,243.
J. Orgetwmet. Cltem., 1998, 570, 247.
Paper presented at the Poster Section of XIV-hit. Con$ on PAOX- phorus Chemistry, Cincinnati, Ohio, USA, July 12- 17, 1998.
Thesis Abstract (Ph. D.)
Resonance ies on strue by Nandita Biswas Research supervisor: Dr. Siva Urnapathy Department: Inorganic and Physical Chemistry
Resonance Raman (RR) spectroscopy has become a widely used technique to probe the struc- ture and dynamics of short-lived excited state intermediates of various nhotochemical and nho- tophysical prooesses.l~ * Under pre-resonance or resonance conditions,Ai.e. when the excitation
J.
wavelength approaches or coincides with a particular electronic transition of the system under investigation, the RR intensities essentially contain information about the excited state poten- tial energy surface and hence, in turn, can dictate the initial dynamics of nuclear motion in Franck-Condon (FC) region on that surface. Wavelength dependence of the Raman scattering intensity, viz. Raman excitation profiles (REPS) provide information on the origin of RR inten- sities and also the dynamics associated with the resonant excited state. Thus, RR spectro- scopy can be used to study the initial excited state dynamics occurring in the range of tens of ferntoseconds, which cannot be studied by time-resolved resonance Raman spectroscopy3 due to the limitation resulting from broad spectral bandwidth of the laser pulse (dictated by the Uncertainty principle). This work deals with various applications of resonance Raman spec- troscopy, viz. isomerization dynamics, resonance deenhancement and electroll transfer reaction in am dyes, the structure of which has attracted considerable attention, in recent times, lnainly due to its wide applicability for reversible optical data storage4 and molecular switching de- v i c e ~ . ~ The basis for majority of these applications is the photoisomerization of am dye6 be-
- J
tween the stable tram and the metastable cis conformation.
IISc THESES ABSTRACTS 389
In the present investigation, it is shown that the use of commercial optical parametric oscilla- tors (OPO) enable a simple method for determining the absolute Raman cros s-sect ion s, ~tilizing the measurement of integrated intensity ratios.7 Standard curves for the absolute Ra- man cross-sections of the commonly used internal standards, viz. acetonitrile (918 cm-l). ni- trate ( 1045 cm-l), benzonitrile (1 000 cml), n-hexane (1 454 cm-') and carbon tetrachloride (79 1 cm-') peaks, in the wavelength range, 450-650 nm, are presented.
A timedependent quantum ~nechanical (TDQM) method of wavepacket propagation in com- puting RR intensitiesB for polyatomic systems i s developed and presented here. The usefulness of this method is shown by applying it to cis-stilbene (a well-studied system). Using the TDQM technique, the absorption spectrum and the REPs of polyatomic molecules can be collz- puted with ease and one could obtain information on the displacements in the excited state at any time after photoexcitation, but within the electronic dephasing time.
iscussion
The equilibrium structures, harmonic frequencies and mode assignments for azobenzene and its substituents are presented, since accurate vibrational analysis o f the am dyes are essential in order to understand illode-dependent structural distortions involved during the isomerization process. The vibrational analysis for trans-azobenzene (TAB) and cis-ambenzene (CAB) and their isotopic analogues" are presented using restricted Hartree-Fock (RHF), hybrid Hartsee- Fockldensity functional (HF/DF) and pure density functional theoretical (Dm) methods utiliz- ing 6--31G* basis set. In addition, a comparative study of the equilibrium geometry and har- monic vibrational frequencies of 6nitroazobenzene (NAB), 4-dimethylaminoazobenzei~e (DAB) and 4-nitro, 4'-dimethylaminoazobenzene (NDAB) using hybrid Hartree-Fockldensity functional method are presented. The effect of substitution on the para position by an acceptor group, by a donor group and by a donor-acceptor group of the parent azobenzene molecule on the vibrational structures and harmonic frequencies is also presented.'0
Although isomerization of a m dyes has been studied by various people in relation to the marked dependence of the photoisomerization quantum yield on excitation wavelength6 and n~olecular structure, still the mechanism of isomerization in general remains a subject of con- troversy. The reason for this is the presence of nitrogen (N) lone pair in the azo group. Hence, in azo dyes, there are two possible routes of isomerization, namely, rotation (torsion or twist) about the central N = N bond which involves a n bond rupture and inversion (in-plane lateral shift) through a linear transition state (the 7~ bond remaining in tact). In order to study the mechanism of isomerization under ( 2 ' ~ , t l ' ~ , ) electronic excitation, the REPs for the 10 Franck-Condon active fundamentals of trans-azobenzene in carbon tetrachloride solution are presented. It i s observed that various modes of tmrzs-azobenzene show reduced intensity near the maxima of the resonant electronic ( 2 ' ~ , t 1 'A,) transition. 11, 12 The observed deenhance- ment indicates that the resonant electronic state is not the sole source of intensity for various fundamental vibrations, instead vibronic activity arising due to preresonant state, results in deviation from the expected excitation profiles. We have used Heller's wavepacket dynamical
390 IISc THESES ABSTRACTS
The concept and determination of inertial s o k n t reorganizatio~l energy for an intramolecu- lar charge transfer process, using RR spectroscopy i s presented.13 These reorganization energy terms play a major role in determining electron transfer rates. RR spectra have been recorded for 4-nitro, 4'-dimethylamino-azobenzene (NDAB) within the charge-transfer transition, in a range of solvents of widely varying polarity, viz. n-hexane, carbon tetrachloride, benzene, ace- tonitrile and benzonitrile. Changes in the polarity of the solvent is thus expected to have a &a- matic influence on the excited state structure and dynamics of the solute and this change in geometry, due to solvent effect, can be easily monitored using RR spectroscopy. It has been shown that RR intensities in combination with the over- and underdamped Brownian oscillator model provides a reasonable estimate to the values of solvent reorganization energy as well as modespecific vibrational reorganization energies. From a detailed experimental and theoreti- cal analysis of the Raman excitation profiles, it is observed that in nonpolar solvent (n-hexane), NDAB exists in the locally excited (LE) state whereas in other solvents, viz. carbon tetrachlo- ride, benzene, acetonitrile and benzonitrile, it exists in the charge-transfer state. The distortions along various normal vibrations (mainly along N = N and C-N stretching coordinates) of NDAB in different solvents infer isomerization via inversion mechanism in benzene and rota- tion in n-hexane. This study indicates that the solvent polarity drastically effects the molecular structure in svstems undergoing intramolecular charge transfer in the excited electronic state.
AND MATE.
3. HAMAGUCHI, H.
In Laser techniques in t*lzcmi,stry (A. B. Myers and T. R. Rixm, eds), Wiley, 1995, Vol. 23, p. 325.
IIES, R. A. In Biological upplictr $ions of Ranzi-n ,spei-iroscnr~py (T. G. Spi ra, ed), Wiley, 1987, Vol. 2, p. 1.
In Vibrationcd spectra utd strrcc?rrm (J. R. Durig, ed), Elsevier, 1987, Vol. 16, Ch. 4.
4. RAMANUJAM, P. S., HVILSTED, S . AND Appl. Ph.ys. Lett., 1993, 62, 1041. ANDRUZI, F.
5. SEKKAT, 2. AND DUMONT, M. Appl. Phys. B, 1992,54,486.
6. Rsu, H. In Photochrornism. Molecules and Systems (1-1. Durr, H. I ~ W - Laurent, eds) Elsevier, 1990, Ch. 4, p. 165.
7. BISWAS, N. ANDUMAPATHY, S. Appl. Spectrosc., 1998,52,496.
8. Brsw~s, N., UMAPATHY, Proc. Indian Accrd. Sci. (Chern. Sci, ), 1995, 107, 233. KALYANARAMAN, C. AND SATHYAMURTHY, N.
9. BISWAS, N. AND UMAPATHY, S. J. l%ys. Chem. A, 1997,101,5555. 10. BEWAS, N. AND UMAPATHY, S+ Submitted.
IISc THESES ABSTRACTS
1 1. BISWAS, N. AND UMAPATHY, S.
12. BISWAS, N. AND UMAPATHY, S.
13. B i s w ~ s , N. AND UMAPATHY, S.
Chem. Phys. Lett, 1995,236, 24.
J. Chem. Phys., 1997,107,7849.
Chem. Phys. Lett., 1998,294, 18 1.
Thesis Abstract (Ph-D.)
Research supervisor: Prof. Ashoka G. Sarnuelson Department: Inorganic and Physical Chemistry
Ligands play an important role in coordination chemistry and organometallic chemistry. Coor- dination of the ligand to a metal ion induces a variety of physicochemical changes that result in reactions that are unique to each ligand set. The present study i s focused on the role played by the ligands in three systems involving zinc-promoted allylation in aqueous media, dioxygen activation by copper and tuning metal-metal interactions in platinum dimers.
The Barbier-type reaction involving the allylation of carbonyl compounds promoted by metals like Sn, In, Bi and Zn in aqueous medium provides a convenient method for C-C bond fom~a- tion in organic synthesis.' This methodology offers relatively inexpensive, environmentally benign reaction procedures with altered selectivities. The mechanism of the Sn-, In- and Bi- promoted reactions are quite clear, while that promoted by zinc is less understood.
A radical species generated by a single electron transfer from the metal surface is believed to be the intermediate? A complete investigation of the mechanism of the zincmediated allyla- tion of carbonyl compounds in aqueous medium was carried out to understand the nature of the intermediate formed. The investigation comprises trapping and competition experiments, CO insertion and asymmetric addition.
Attempts were made to trap the intermediate with reagents that could interfere with radical as well as organometallic intermediates formed during the reaction. Control experiments with- out the traps were performed simultaneously to compare the yields. Reactions in the presence of radical traps like BHT, 4-test-butylcatechol and O2 proceed smoothly without decrease in the yield of the product alcohol suggesting that radical species may not be intermediates.
392 IISc THESES ABSTRACTS
The reactivity of the carbonyl group could be altered by substituents present on the ring, While electron-withdrawing groups increase the reactivity, electron-releasing groups decrease it. Organozinc intermediates being electron rich would react much faster with the aldehyde having an electron-withdrawing group compared to the one with an electron-releasing group leading to the observed large difference in rates. For radical intermediates (electro- and nu- cleophilic), similar differences in rates of addition to substituted aldehydes would still be ob- served but will be of smaller magnitude.
Competition reactions between differently substituted aldehyde pairs were carricd out using half equivalent of zinc and ally1 halide. Aldehydes containing electron-withdrclwing groups were found to react much faster than those with electron-releasing ones. In fact, p-hydroxy benzaldehyde failed to react with the intermediate and was recovered unreacted. However, in the absence of competing aldehydes it did react with the intermediate to give rise to the corre- sponding alcohol. The reactivity pattern is clearly discernible and is compaxxble to that ob- served by Zhang et aL3 for the addition of diethylzinc to various substituted aldehydes. This experiment also lends support to the formation of an allylzinc intermediate.
In the presence of high pressures of CO (500-650 psi) small amounts of sy~nmetric dial- lylketone was obtained. The formation of the ketone suggests the insertion of CO across the zinc carbon bond of the allylzinc intermediate.
Another interesting outcome of this investigation is the development of a methodology for asymmetric allylation. While reports on the addition of allylhalides to c h i d carbanyl com- pounds under aqueous conditions are readily available, similar addition to achiral substrates are scarce. The only attempt with c h i d auxiliaries like mannitol or camphoric acid resulted in racernic products.4 However, in the present investigation chiral l iga~~ds like L-(+)-tartaric acid and diethyl-Martrate were successfully used for asymmetric induction. The stereoselectivity was found to be dependent on the reaction temperature. The chiral auxiliaries are believed to coordinate to the intermediate zinc species resulting in the formation of diasteseomeric trmsi- tion states whose ratio depends on the reaction temperature and is reflected in product selectiv- ity. The observed selectivities in the presence o f chiral auxiliaries could well be explained by invoking the formation of a Zn-carbon intermediate.
IISc THESES ABSTRACTS 393
A variety of metalloenzymes activate oxygen towards various biochemical oxidations, which otherwise be forbidden due to the triplet state of oxygen. Considerable efforts have been
~nade to understand the role of ligands in these systems to develop synthetic models that could perform the same under laboratory conditions.' The synthetic models developed so far suffer a serious limitation in that they are stable at very low temperatures and decompose at ainbient temperature. There are instances wherein the peroxo species have been stabilized by steric hin- drance or by hydrogen bonding%~oever, details regarding the oxidizing ability of such stabi- lized species are unknown.
Two lignnds, 2, containing nitrogen and oxygen donor sets have been used to syn- thesize copper(1) complexes that activate oxygen. The oxygenation of dichloromethane solu- tions of these complexes results in the formation of a green and a bluish green complex, re- spectively. In the presence of oxygen these complexes are in equilibrium with a complex con- taining 3 perguxo bridge. The peroxo bridge was detected by a chemical test for peroxide. They are reasonably stable at room temperature. Stability is attributed to hydrogen bonding by the ligand. However, attempts to crystallize these complexes failed. Instead, a colourless unstable copper(1) complex crystallized out from dichloromethane.
These complexes exhibit selective oxidation ability. The addition o f two equivalents of PPh3 to dichloromethane solutions of these complexes results in the formation of PPh@ and the corresponding Cu(1)-PPh3 complex. These complexes were also found to oxidize 3, 5-di- .
text-butylcntechol catalytically to the corresponding quinone with a turnover number of 7.5. Similarly, benzylamine was found to be catalytically oxidized to the corresponding aldimine which after hydrolysis results in the formation of Schiff base adduct. This reactivity is different compared to the oxidation of amines to nitriles by CuCl/02/pyridine system.
The reactions observed with PPh3 and 3,5-di-tert-butylcatechol suggest the presence of a p-$ : ~ ~ - ~ e r o x o moiety as the active species. However, the catalytic oxidation of catechol was found to be hampered by competing ligand oxidation, a limitation often encountered in cata- lytic oxidations. The pmduct due to the ligand oxidation was obtained as a colourless liquid whose structure was not elucidated.
C
394 TlSc THESES ABSTRACTS
short metal-metal distances are quite often observed in the solid-state ~ t s~c tu re s of many di- and plynuclear metal clusters. It was earlier believed to result in metal-metal interactions in these systems. Metal-metal interactions in such systems depend largely on the coordinated &an&. They serve as a bridge through which the metals communicate. In many cases, the ligands force the metal centers to be in close proximity only to satisfy the bite angle requilx- ments. Ab irzitio and Extended Huckel Theory (EHT) calculations serve as tools in understand- ing the nature of these metal-metal interactions?
The role of bridging ligands (acetate, amidinato and triazenido) in mediating Pt-Pt interac- tions in dinuclear platinum(I1) complexes were probed by EHT calculations. The computed reduced overlap population (ROP) suggests that M-M interactions are primarily affected by the electronic nature and relative disposition of the bridging ligands. Fragment molecular or- bital analysis of these systems suggests that good n-acceptor ligands, triazenido for example, stabilize the metal-metal bond while the amidinato ligand destabilizes the metal-metal bond due to the lack of suitable n* orbitals.
References
LI, C. J. Chenz. Rev,, 1993, 93, 2023.
Lucw~, J. L., ALLAVENA, C., PETRIEK, C. Tetmhedron Lett., 1998,29, 5373. AND DUPIIY, D.
ZHANG, H., XUE, F., MAK, T. C. W. AND J. Org. Chem., 1996,61, 8002. CHAN, K. S.
EINHORN, C. AND LUCI-IE, J. L. 1. &gnnornet. Chew., 1987,322, 177.
~ T A I I M A , N. AND MORO-OKA, Y. Chem. Rev*, 1994, 94, 737.
EL-SAYED, M. A., ISMAIL, K. 2. AND I?zlorg. Chim Acta, 1994, 217, 109. EL-ZAYAT, T. A.
AULLON, G. AND ALVAREZ, S. horg. Chenz., 1996,35,3 1 37.
J. Irtdian Inst. Sci., July-Aug. 2000, $0, 395-403. (3 Indian Institute of Science
verse serraigrou
The initial introductory chapter introduces the definitions of the two basic notions above and discusses (in a 'mn-formal way', as the author himself says) the connections between theln and their associated operator algebras, using the illuminating special case of the so-called Cuntz groupoid, the Cuntz semigroup and the Cuntz Czalgebra. The quickest definitioll of a groupoid is 'a small category with inverses'. An illuminating example is that of the fundamen- tal groupoid of (hoinotopy classes of paths in) a topological space; thus composition is re- stricted, but there are inverses. The occurrence of quasicry st als and Penrose-tilings might be cited as reasons for the inadequacy of just groups to describe 'all possible symmetry in nature'.
Similarly, 'inverse sernigroups' are a precise abstraction of 'partial bijections', and not sur- prisingly, they admit representations as classes of partial iso~netries in Hilbert space (rather than unitary operators, as in the case of groups). The second chapter discusses inverse- semigroups in detail and their natural 'left-regular representations' and the naturally associated (reduced as well as universal) C'kalgebras (defined as the completions of the complex 'semi- group-algebras' with respect to appropriate norms, as in the case of groups). In order to discuss the representation theory properly, the author then goes on to discuss the classes of r-discrete groupoids, locally compact groupoids and finally Lie groupoids (as well as the important spe- cial cases of the holonorny and tangent groupoids).
The third chapter is primarily devoted to the representation theoly of locally compact grou- poids; and the book strives to strike an acceptable balance between wanting to give enough detail, and not wanting to get too entangled in a morass o f detail, which is not too easy in a suhject which is intrinsically fraught with various technical problems and complications. (For instance, unlike the group case, neither existence nor uniqueness of Haar measure is available in the general case.)
For this, the author uses the device of paying a lot of attention to the so-called r-discrete groupoids in which several potential problems either vanish or considerably diminish.
The final chapter is devoted to the problem of determining the representation theory of an inverse semigroup in terms of associated r-discrete groupoids. Using this machine, the author proves in the last section that the von Neumann algebra generated by the left-regular represen- tation of an inverse semigroup S is amenable, if every maximal subgroup o f S is amenable.
Finally, the book ends with an Appendix containing six sections, devoted to various topics treated in the body of the hook.
The author has attempted the difficult task of presenting a technically demanding subject in as reader-friendly a manner as possible. For this, he certainly deserves a pat on the back. If the
396 BOOK REVGEWS
reader finds the p i n g a little rough and seemingly esoteric at times, @)he might take solace in the fact that some of this material is what is needed to understand some of the meatiest exam- ples relating to foliations or Penrose tilings, for instance, of Alain Comes' noncommutative geometry!
The Institute of Mathematical Sciences (Matscience) Taramani Chennai 600 1 13
7, Parts I and 11, edited by Kathleen C. Howell et nl., Published for the Arnelican Astronautical Society by Univelt Inc., P. 0. BOX 28 1 30, Stin Diego, California 92198, 1997, pp. 1168, $240.
The AAS/AIAA Space Flight Mechanics Conference is one of the most popular annual events in aerospace industry drawing scientists and engineers from all over the world. The seventh allnual conference has drawn nearly 100 papers contributed by more than 200 authors, The proceedings published in two parts contains 71 full-length papers and 14 abstracts a~+rc?nged under 14 sessions. Part I includes six sessions on: Orbital mechanics; Global positioning sys- tem; Attitude dynamics; Satellite theory; Navigatiot~/tracking/debris; Multi-body dynam- icsltethered systems. Similarly, part I1 holds 30 papers and 13 abstracts under eight sessions on: Trajectory design and optimization; Control theory; Mission design-near earth; Attitude determination and control; Mission design-Lunar and interplanetary; Dynan~ics and control of flexible structures; Orbital transfers, Orbit determination and filtering theory. Tl~erefore, six sessions are devoted to spacecraft orbit, five to rigid and flexible spacecraft attitude motion, two to spacecraft missions and one to applied control tl~eory. Majority of' the technical papers are indicators of the latest developments in their respective field.
The orbital motion of heavenly bodies (i.e. planets around Sun and Moons of Earth, Jupiter, Mars, Saturn, etc) are studied ever since the days of Kepier for over 300 yeus. However, the dawn of space era and evolution of digital computers spurred the shnspening of time-tested method of orbital analysis and emergence of new techniques. During the space age that ex- ceeds four decades (first artificial satellite was launched four decades ago), a large number of computer software was produced. A very systematic compilation and assessment of orbit analysis software (1998 updates) from leading software f i r m from USA and west European countries has been presented as a first paper of the conference (rightfully so). This paper i s certainly very useful reference for researchers from space industry and faculty of academic institutions. The first session on orbital mechanics also contains six more papers on alternate representation of orbital elements, comparative study of asteroid collision avoidance on to the earth, closed loop guidance control of a satellite around a slnall primary, control of satellite constellations, analysis of halo orbits around Sun-Earth Lagrange points defined by the re- stricted three-body problem. The two-orbit control-related papers and the four-orbit analysis papers are only of academic interest and are unlikely to be used in the near future. On the other hand, the last five papers on orbit analysis and orbit control in the session an Satellite theory have significant practical utility. These papers represent the progress made on improving the
BOOK REVIEWS 397
state-of-art on-orbit analysis and control. Similarly, orbit transfer problems also studied extensively in the past, and a certain level of maturity is attained by early 89s. Two sessions on ~ ~ ~ j e c t o r y design and optimization, and orbital transfers have five papers which are samples taken from low-key research topic being pursued at present. A paper on optimal lllaneuver se- quence for recovery of satellite from the lower launch orbit (due to under-performance of the launch vehicles) to desired final orbit through a series of optimal orbit-raising maneuvers using the onboard fuel shows that a systematic study of maneuver can improve the emergency ma- neuvers conducted by the practising engineers from industry. In future too, there exists need for using fuel optimal transfers and hence the need for even more optimal solutions. With in- crease in space-activity-related permanent space stations and space colonies, soon satellite ren- dezvous will become important activity. A run-down of space rendezvous between two satel- lites in close by orbits is surveyed well by Thomas Carter. This author also presents a simpli- fied approach for obtaining solution to terminal rendezvous described through linearized equa- tions of motion. These papers indicate that there is still a little scope left for researching into orbital transfer problems.
Orbit determination of planets, moons, asteroids and other objects in space has been under- taken from time immemorial. Beginning of Kepler's era and satellite era saw quantum jump in the development of orbitltrajectory determination know-how. With the increase in the popula- tion of earth satellites (low to high altitude), and associakd space debris, and also arrival of debris from outer space as a random event, navigatiodtrachng of satellites v i s - h i s debris (and their population count) has become need of the hour. The seven papers in the session on Navigation/Tracking/debris cover diverse spectrum of issues on the theme topic and are useful to researchers. The advent of Kalrnan Filter-based orbit-determination algorithms has helped in getting highly accurate estimates by rejecting the bios and systematic errors and smoothening of the effects due to noise in measurements. The last session on orbit determination for life- extended Topex/Poseidon mission, ill-conditioning of covariance matrices of the two-step nonlinear estimator for orbit navigation problem, and development of the modified detection filter for sensor faults (failures). However, three plus seven papers on two sessions on Mission design-near Earth and Lunar and interplanetary are likely to disappoint some readers.
The papers presented in five sessions on spacecraft attitude dynamics and control are well focussed (partly due to the fact that they deal with only one aspect of space flight rncchanics). Out of six papers in the session on Attitude dynamics, there are two interesting papers on atti- tude motion with varying mass plus two papers on fomwlation of attitude motion. Even though lots of research has been done in this field, these four papers indicate that there is further scope for research. With the ground work on building th6 International space station under progress, academic research on tethered satellite dynamics and control has turned out to be of practical value in the near future. The last three papers on the session on Multi-body dynamicsftethered systems are representative of the continuing investigation on tethered satellite by the academia. Hopefully, such systems will be successful in future not withstanding the past failure of such mechanisms in orbit. Amongst the five sessions on satellite attitude dynamics and control, pa- pers presented in tile two sessions on Attitude determination and control and Dynamics and control of flexible structures are the most interesting ones and are also useful in the ongoing spacecraft development. The paper on airborne vehicle attitude determination using GPS is a fine example of spin-off of space technology to allied area of research.
398 BOOK REVIEWS
The progress of mankind and its well behg hinges on the success in the pursuit of excel- lence. Space mechanics is one such branch of technology where the frontier research and &- velopment can make a qualitative and quantitative difference. The proceedings of space flight mechanics over the years have maintained the tradition of bringing out the samples of state of the art. These proceedings certainly meets the expectations o f the readers.
Department of Aerospace Engineering M. SEETHARAMA BHAT Indian Institute of Science B angalore 560 0 12
ars V, AAS Science and Technology Series, Vol. 97, edited by Penelope J. Boston, Published for the American Astronautical Society by Univelt, Inc., P. 0. Box 28 130, San Diego, California 93 198, USA, 2000, pp. 552, $80.
The Case for Mars conferences were started as informal fora for the discussion of human ex- ploration of Mars in 1981. Interest in Mars exploration grew steadily in the US in the Post- Apollo days. In 1969, the Space Task Group recommended the goal of a manned Mars Mission before the end of the 20th century, but was deferred. Again in 1989, the then US President Bush constituted the Space Exploration Initiative (SEI), which advocated for a program to send humans to Mars but the US Congress didn't approve. Interest in Mars was waning fast in 1993 when the Case for Mars V was organized to maintain a continuous dialogue amongst people to explore the red planet. The title volume is a collection of the papers of the proceedings of the fifth Case for Mars conference held in Boulder, Colorado. The 33 papers in the volume are covered in six chapters. It has two appendices. Relevant aspects like Planning for missions to Mars, Surface exploration and Human missions to Mars are covered in this issue.
The chapter on Planning for missions to Mars opens with an article on Mars colonization by the former NASA Chairman and a visionary, Dr. Thomas Paine (now deceased), as a ges- ture to honor him. This paper was published earlier in Vol. 86, AAS Science and Technology Series, 1996, which the present viewer had the privilege of reviewing for this journal. Unlike the Moon, Mars is too distant; going there is too risky and expensive. Americans landed man on the Moon essentially, 'to beat the Russians'. With the ending of the Cold War, the political justification of the case for Mars is over. A human mission to Mars will not be undertaken simply because it is a good idea from an international standpoint, says D. A. Day in his care- fully crafted article, 'The third era o f space exploration'. He cites several scientifically sound, nonpolitical justification in favor of going to Mars, but concludes with a pessimistic note that 'the only thing likely to change the situation and lead to an increase in human exploration ef- forts and a decision to send humans to Mars is the discovery by robotic probes of past or pre- sent life there, But such a discovery many never happen and, unless it does, the likelihood of a human mission to Mars in the near future is grim indeed'.
Is there a short-term economic and social justification for manned solar system explora- tion? The question is not answered. Rather, the article provides a rationale for initiating coor- dinated efforts by the exploration community to convince the government and the public. An-
BOOK REVIEWS 399
other article examines the question, IS bone loss in space a problem? Skeletal mineral loss ex- perienced by humans in micro-gravity has been considered as one of the prilnary nledical con- cerns which could limit mission duration. After a careful analysis of the bone mass of the Rus- sian cosmonauts who have fkwn on space station Mir for long periods, it is concluded how- ever, that the bone loss occurs in space at a relatively slow rate compared to the time frame of planned mission operations to Mars.
The Space Exploration chapter describec an interesting proposal for a global reconnais- sance of Mars using super-pressure balloons. The purpose of the so-called 'Mws aerial plat- bm' mission i s to generate high-resolution pictures of the Martian surface, map the global ciI-culation of atmosphere, and examine the surface and sub-surface with ground penetl-ating radar, and other remote-sensing techniques, which are carried by balloons flying about 7 km over the Martian surface. The balloon probes could last for thousands of days producing an immense amount of data. Other issues dealt in this section include identification of a preli~ni- nary site list for Mars Exobiological exploration, a 'jumper' Having possibility to make jcl~nps over the natural obstacles of the rugged Mars surface, systems for Mars Extravehicular Mobil- ity Unit (or space suit) design, and over-the-horizon communications on Mars via HF radio propagation.
Novel transportation concepts have been proposed to save propellants during the return journey from Mars. The idea of magnetobraking to dissipate hyperbolic excess velocity for Mars return vehicles has been proposed. The Earth's magnetic field produces a force on elec- tric current in the tether deployed fsan the spacecraft, which can be used to brake the space- craft speed without expending the reaction mass. Another concept makes use of the liquefied inert gases from the Martian atmosphere as inert reaction mass to inject into the combustion chamber of an Earth-return vehicle. The process results in a greater mass of propellant used, but can reduce the requirement for fuel brought from the Earth.
A significrmt fraction of the duration of a human exploration mission occurs in the Earth- to-Mars iransFer orbit. Failures during this period may require immediate orbital maneuvers in order to ensure the safe return of the crew. An investigation of a powered abort to the Earth orbit as an emergency response to life or mission-threatening failures not affecting the propul- sion system is presented.
Methanol i s an excellent candidate fuel for both Earth and Mars because it can be produced from atmospheric carbon dioxide and is a source of hydrogen on Mars, and industrial carbon dioxide waste on the Earth, biomass and methane. It can be used in fuel cells to produce elec- tricity fc-r Martian habitats and surface vehicles, and as a rocket fuel for the return journey to the Earth. Its relevance to rocket propulsion and production on the Mars and the Earth by vari- ous techniques has been discussed in detail.
A major objective of the human exploration of Mars might be the permanent settlement on the planet. This has been a consistent theme of The Case for Mars conferences also. An elabo- rate article on the long-term habitation of Mars examines the issues of acquisition of matter and energy on Mars and their organization in life-support systems. Several aspects of the life- support system needed for long-term habitability are discussed. It is argued that although any
400 BOOK REVIEWS
initial life-support system for Martian colonists is going to be a small, enclosed biosphere, the notion of eventually tenaforming (endowing Mars with a biosphere that mimics that of the Earth) will be debated by the earliest settlers. Various approaches to terraforming and parater- rafoming M a s have been discussed. Paraterraforming differs fro111 terraforming in that the intention is to achieve conditions suitable for human habitability even though the transforma- tion of the pl~inet to a fully Earth-like gravity-bounded state may either be impossible or is not achievable on a realistic time scale. Transforming Mars for terrestrial life appears to be a favor- ite topic of several authors. The AAS Science and Technology Series, Vols 91 and 92, lists many papers on this topic, which were reviewed by this reviewer earlier. Apparently, terra- forming Mars represents a distant, perhaps an improbable, drea~n-ethically too controversial, and not worth giving too much of an importance, at present.
Undoubtedly, the construction of a Mars base is one of the major tasks the travelers to Mars confront. Careful planning is definitely required before an actual base development is realized. A report to evaluate mission design and exploration technology proposals to ensure that the means support the goals and do not obstruct them numerates the factors that may detract from success. The success of a Mars Mission architecture depends on how well it defines the pmb- lem, then solves it, while combating institutional inertia and political entropy, according to this report.
A M a s base study proposes the presence at the sull'ace, rather than in a hollowed-out cave dwellings. Concrete construction may be used on Mars for perxnanent enclosures. Portland cement material has been identified in soil samples, and may be recovered to provide building materials. Another article describes ways by which impact craters, abundant on Mars, could be converted into living biomes by providing them with a flattened dome-like cover consisting of +an underlying plastic or fiber glass seal and loaded down with Martian soil to balamx the at- mospheric pressure trapped within. Yet another type of base proposed i s a mobile or 'roving' base. The Nomad Explorer, primarily designed as a lunar base, is claimed to vastly inlprove the exploration range.
Abstracts (93 in all) of all the presentatior~s of the conference are put together in Appendix A. Appendix B lists all the major publications of the AAS. In general, the papers compiled in the present volume are well chosen. However, because of the delay in publication of this vol- ume, same papers have already appeared elsewhere while a few others, like the lengthy status report on the mission Mars Pathfinder in 1993, have become redundant as the mission is al- ready completed.
Interest in Mars was dimmed when the images from Mariner spacecraft in the 1960s re- vealed barren, lunar-like landscapes, and no canals! Mars seemed geologically lifeless, i t s mea- ger portion of water locked beneath a deeply frozen surface. Stunning images from the Mars Global Surveyor (MGS) which has been in orbit since 1997 are breathing new life into the red- planet. Camera aboard MGS had sy ied signs of geologically recent-possibl y even ongoing- water seeps has caught everyone's attention. From these findings, it is imminent that interest in Mas exploration will grow steadily in the coming years. As a matter of fact, the enthusiasts expect the first manned flight to Mars in the first half of this century. In this context, I am sure
BOOK REVIEWS 30 1
text, I am sure The Case for Mars V conference papers will make an interesting reading to the space researchers and students.
Department of Aerospace Engineering Indian Institute of Science Bangalore 560 012
n: AAS Science and Tech- nology Series, Vol. 98, edited by Kelly R. McMillen. Published for the American Astronautical Society by Univelt, Inc., P.O. Box 28 130, San Diego, California 92 l98,2OOO, pp. 560, $SO.
The title volume is a collection of 42 papers and 12 abstracts presented in the six.tl1 case for Mars Conference held in 1996 at Boulder, Colorado. Grouped into five chapters, these articles cover issues like Building support for low-cost missions to Mars, Essential technology and proposed infrastructure, Reducing the cost of precursor missions, Designing a habitable planet, and Working group reports. A couple of the abstracts were earlier published as full-papers in tlze AAS Science and Technology Series, Vol. 92 (1997), which was reviewed by this reviewer for t h i s journal.
The volume begins with an interesting article describing the heritage and legacy of the Vi- king mission to Mars by two of its main architects, G . Morgentlmler and W. Lowrie. The two Viking spacecraft, which successfully landed on Mars in 1976 within a span of 40 days cover- ing 460 million miles, were designed, built and landed on Mars by Martin Marietta Company by its own launch vehicle, Titan III. In the next paper, Robert Zubrin, an ardent protagonist of the Case for Mars, portrays the significance of the Martian frontier, advocating strongly for terraforming Mars. He clear fusion reactors by
Issues like How to
foresees the use of deuterium available on Mars as fuel for thermonu- the Martian colonists.
'sell' Mars to the public'? What set of conditions will align to enable human exploration of-' Mars? Who funds it? are the topics of several articles. While these can- not be a single answer to these questions, it is argued that premises like 'exploration for exploration's sake', 'pioneering spirit', 'human destiny' or 'because it is there' as rationales for Mars exploration wouldn't work for the public at large. Ways suggested to gain a high level of public support include persuasive discussions of pivotal audiences including children, media personnel, and opinion leaders. Large-scale public support, however, i s only expected if there is a commercinl motive in settling on Mars,
Unlike a trip to Moon, the human exploration of Mars will be a complex undertaking in- volving commitment for scveral years, and there is a very narrow window within which return is possible. NASA has formulated a 'Reference Mission' addressing human exploration o f Mars using approaches that are technically feasible, have reasonable risks, and have relatively low cast. The plan emphasizes the surface exploration of Mars; the crew will travel to and from Mars on fast transit (4 to 6 montlzs) and will spend long time (18 to 20 months) on the surface. The single largest mission cost is transportation. The Marshall Space Flight Center has developed a Space Transportation Plan to explore ways to enable a low-cost human Mars ex-
402 BOOK REVIEWS
pedition by 201 8, at a program cost of no more than $25 billion. The most cost-effective Inems of transport proposed is a nuclear thermal propulsion system. However, in 1996 the Congress - -
terminafed the last remaining US space rmclear research and development TOPAZ International Program. It will take a major and costly effort to resurrect clear rocket programs.
program, the the thermonu-
'One-way to Mars' is a new slogan. A few authors believe that bringing back the crew after sending them millions of miles and landing on Mars negatively affects the cost and risk. Be- sides, a n ~ return sanlple from Mars could cause a microbiological disaster on Esu-th. Tomas Gunn, an expert in managing large-scale space projects, warns sternly of this possibility and argues that humans on Mars nus st and should be one way for the foreseeable future. According to him, no agreeable case can be made for the need to bring anything but the data back to Earth.
The cllapter on reducing the cost of precursor missions lists some novel means and tech- nologies that may be cost-effective. A comparison of alternative methods for the Mars sample return mission reveals that the one employing irz-situ (produced on Mars) propellant offers the lowest cost, as well as risk. A high-power Mars subsurface radar operating from an orbiting spacecraft is described which could map the planet in search for undergrourd ice, water and other geologic phenomena up to 1 to 5.6 km below the surface. A study of human Mars explo- ration alternatives using in-situ propellant production and current technology, carried out at the University of Washington, examines two scenarios to provide an econo~nically feasible ap- proach to manned Mars exploration. Both use carbon monoxide as an in-situ-produced propel- lant, and three Energia launches, but differ in other details. Another article relates to prospects for using C02/metal propellants in Mars missions.
The use of the International Space Station in preparing for the human exploration of Mars is explored in detail. Another paper reviews the design of the surface mission for a Mars Ex- ploration program. The mission goals being to perform scientific exploration and to assess the habitability of Mars. These could be achieved by (a) using pressurized rovers capable of sup- porting a crew o f three for up to a month duration and (b) by growing food and using in-situ resources to support the base. It is surmised that 'Being There on Mass' will turn out to be much like being in remote places on Earth.
As in The Case for Mars V (AAS Science and Technology Series Vol. 97), which the pre- sent reviewer had reviewed here, and other Volumes, Nos. 86 and 92, reviewed earlier by him, there are several papers on terraforming (transforming for terrestrial life) Mars, in this volume also. An ethical perspective of terraforming Mars points out various issues in detail. One view is that humans should terraform Mars, if and only if, there is no indigenous Martian life. Con- trary to this, R. Zubrin is strongly opposed to any ethic which would allow moral standing on Martian biota. ". . .bacteria don't have rights.. . And the idea of denying humanity a new world to provide a reservation for extraterrestrial bacteria is - simply ludicrous." Another paper re- views the ongoing research on terraforming Mars. The main premise of terraforming is that Mars might be transformed into a habitable planet by planetary engineering, which could alter the environment so that terrestrial organisms could survive and grow.
As pointed out in one of my earlier reviews, terraforming is a pet topic of many authors. It basically involves warming the climate of Mars to cause a green-house effect. Methods sug-
BOOK REVIEWS 403
gested for warming the planet are most fascinating to say the least. A payer presented in this voluine proposes the use of weapons-grade plutonium that has become surplus to defense needs with the end of the Cold War as a fuel for nuclear reactors for Mars exploration and ter- raforming. Further, it recommends to assemble such a reactor in space by injecting the fuel rods into orbit using the gas gun like that being developed at the Lawrence Livermore National Laboratory. Another paper suggests seeding by microorganisms to create a self-regulating an- aerobic biosphere, thereby, facilitating the planetary engineering o f Mars. Anyway, both terra- forming and ecopoiesis (creation of a self-regulating biosphere) are a dream (or nightmare) of a distant future.
The volume includes extracts of the conference workshop organized to focus on ways to mount an initial mission to be assembled over an eight-year period for a total cost of $32 bil- lion. Topics included the rationale, innovative technologies and strategies, management and organizational approaches, and international cooperation.
Of course, interest in the exploration of Mars is rekindled by NASA's recent finding that life may have once existed on the red planet. NASA has planned six robot missions to search for evidence of life and for water, and the return of rock samples by 201 1. There are no current plans, however, for a lnanned mission. The papers listed in this volume argue for a more inten- sive initiative possibly by actively i~~volving the academic institutions, private industries and international participation. Several relevant ideas have been put forth to reduce the mission cost. However, too much importance is being given to terraforming and ecopoiesis aspects, which are not going to matter in a foreseeable future. Overall, the plans and projects described for the exploration of Mars are informative and make a highly interesting reading. I am sure space scientists and students will enjoy going through this volume.
Department of Aerospace Engineering Indian Institute of Science B angal ore 5 60 0 1 2
Published by Prof. T. S . Ramamurthy, Executive Editor, Journal of the Indian Institute of Sci- ence, Bangalore 560 012; Typeset by Creative Literati Pvt. Ltd., Bangalore 560 043, Te1.-cum- Fax: 5454047, 5455808; Printed at Lotus Printers, Bangalore 560 044, Tel.: 3380167, Fax: 3357378.
MANUSCRIPTS should be in English and written in a concise form. Three type-written double- spaced copies with one set of original Indian ink drawings and two sets of prints are t o be submitted. The pages should be consecutively numbered and securely fastened.
TITLE PAGE should contain the following: (i) a brief title with suitable words for indexing; (ii) the names of the authors and the institution(s) where the work was carried out; (iii) a footnote with the present address of the authors, if different from (ii); (iv) a 75-word Abstract which summarizes the significant results of the communicated paper; ( v j keywords for indexing and information retrieval; (vi) major discipline; and (6) a runninghhort title.
TEXT should begin on page 2. It is preferable t o break up the text into different sections, with suitable numbered headings, such as; 1 . Introduction, 2. Experimental, 3. Theoretical analysis . . . , and 7. Conclusions. Acknowledgments should appear at the end of the paper, but before references.
REFERENCES should be indicated by IndiadArabic numerals with superscript letters, e.g., "Earlier Ehse6 has measured . . . ,". References should be typed in double space on a separate sheet in the order of their occurrence in the text and appended at the end of the paper. They should be given with full details as in the following examples:
3 . RAMAKRISHNA NAIDU, G, Isotopic exchange study of nickel xanthate in the presence of AND NAIDU, P. R. toluidines, Proc. Indian Acad. Sci. A , 1978, 87, 443-446.
Convergence of state distributions in multi-type Bellman-Harris and Crump-Mode-Jagers branching processes, Ph.D. Thesis, Indian Institute of Science, Bangalore, India, 1978.
LENGTH of papers: Up to 25 pages (about 8,000 words) for full papers and 6 pages (2,000 words) for short communications.
TABLES should be included only when considered essential, They should be numbered consecutively with Roman numerals and typed on separate sheets. Every table should contain a descriptive title.
FIGURES should be drawn with Indian ink on good tracing paper. It i s most important that the lines and letters are sufficiently bold to permit reduction to the Journal size. Captions should be typed on a separate sheet.
MATHEMATICAL SYMBOLS should be identified with pencil on the margins. Standard mathematical notations should be adhered to. Please distinguish between kay and kappa, ell and one and other similar symbols likely t o cause confusion.
ABBREVIATIONS such as e.g., et al., and i .e. , can be used. If non-standard abbreviations or acronyms are used they should be explained where they appear first in the text.
GALLEY PROOFS will be sent to the authors for correction. They should be carefully scrutinized and returned within three days of receipt.
REPRINTS: Fifty reprints are supplied gratis. Additional reprints (up to 50) can be ordered while returning the corrected galley proof.
PAPERS received for publication are subjecied to rigorous review process.
MANUSCRIPTS should be sent to the Editor, JOURNAL OF THE INDIAN INSTITUTE OF SCIENCE, C/O IISc Library, Bangalore 560 0 12, India. ernai 1 : jou~nal@;l ibra1y.i iscCemet .in
MANUSCRIPTS should be in English and written in a concise form. Three type-written double- spaced copies with one set of original Indian ink drawings and two sets of prints are t o be submitted. The pages should be consecutively numbered and securely fastened.
TITLE PAGE should contain the following: (i) a brief title with suitable words for indexing; (ii) the names of the authors and the institution(s) where the work was carried out; (iii) a footnote with the present address of the authors, if different from (ii); (iv) a 75-word Abstract which summarizes the significant results of the communicated paper; ( v j keywords for indexing and information retrieval; (vi) major discipline; and (6) a runninghhort title.
TEXT should begin on page 2. It is preferable t o break up the text into different sections, with suitable numbered headings, such as; 1 . Introduction, 2. Experimental, 3. Theoretical analysis . . . , and 7. Conclusions. Acknowledgments should appear at the end of the paper, but before references.
REFERENCES should be indicated by IndiadArabic numerals with superscript letters, e.g., "Earlier Ehse6 has measured . . . ,". References should be typed in double space on a separate sheet in the order of their occurrence in the text and appended at the end of the paper. They should be given with full details as in the following examples:
3 . RAMAKRISHNA NAIDU, G, Isotopic exchange study of nickel xanthate in the presence of AND NAIDU, P. R. toluidines, Proc. Indian Acad. Sci. A , 1978, 87, 443-446.
Convergence of state distributions in multi-type Bellman-Harris and Crump-Mode-Jagers branching processes, Ph.D. Thesis, Indian Institute of Science, Bangalore, India, 1978.
LENGTH of papers: Up to 25 pages (about 8,000 words) for full papers and 6 pages (2,000 words) for short communications.
TABLES should be included only when considered essential, They should be numbered consecutively with Roman numerals and typed on separate sheets. Every table should contain a descriptive title.
FIGURES should be drawn with Indian ink on good tracing paper. It i s most important that the lines and letters are sufficiently bold to permit reduction to the Journal size. Captions should be typed on a separate sheet.
MATHEMATICAL SYMBOLS should be identified with pencil on the margins. Standard mathematical notations should be adhered to. Please distinguish between kay and kappa, ell and one and other similar symbols likely t o cause confusion.
ABBREVIATIONS such as e.g., et al., and i .e. , can be used. If non-standard abbreviations or acronyms are used they should be explained where they appear first in the text.
GALLEY PROOFS will be sent to the authors for correction. They should be carefully scrutinized and returned within three days of receipt.
REPRINTS: Fifty reprints are supplied gratis. Additional reprints (up to 50) can be ordered while returning the corrected galley proof.
PAPERS received for publication are subjecied to rigorous review process.
MANUSCRIPTS should be sent to the Editor, JOURNAL OF THE INDIAN INSTITUTE OF SCIENCE, C/O IISc Library, Bangalore 560 0 12, India. ernai 1 : jou~nal@;l ibra1y.i iscCemet .in
