Control of a naturally commutated inverter-fedvariable-speed synchronous motor
N.A.H. Issa and A.C. Williamson
Indexing terms: Controllers, Inverters, Synchronous motors, Velocity control
Abstract: This paper describes the design and construction of the control circuits of a naturally commutatedcurrent-source inverter feeding a synchronous motor to give a variable-speed drive. The main contributionsof the paper deal with the circuits required to control the inverter from rotor position, and to control thesystem for all modes of operation including regenerative braking. The performance of a prototype is illustratedby test results and recordings.
1 Introduction
Much interest has been directed towards the use ofrotationally induced e.m.f.s in the stator windings of anexcited-rotor synchronous machine for the natural commu-tation of a directly coupled thyristor inverter to give avariable-speed drive. The attractions of this scheme are thatrelatively inexpensive power handling devices can be usedwith a conventional machine to give a drive with inherentlyflexible characteristics.
The principal elements of the drive are shown by Fig. 1.A conventional thyristor converter is supplied from themains and referred to as the supply converter. The outputof the supply converter is connected, through a choke inthe d.c. link, to a similar bridge (referred to as the machineinverter) connected at its a.c. terminals to the stator of a 3-phase excited-rotor machine.
Under motoring conditions, the power flow is frommains to machine shaft, the machine inverter being com-mutated naturally by the a.c.-line voltages. Maximumutilisation of the machine and equipment, in terms oftorque per ampere, is obtained if the rotor excitation is atits maximum value. The drive is inherently capable ofregenerative braking by reversing the roles of the twothyristor bridges and the d.c.-link voltage polarity, whilstreversal of direction of rotation simply requires a reversalof the firing sequence of the machine-inverter devices.
The machine-inverter firing pulses must be controlledwith respect to machine-winding voltages in both phase andfrequency. The conventional converter-control technique,using a.c.-linevoltage zero-crossing instants as references, isnot attractive in this case owing to a high harmonic contentof the voltage waveform and a wide frequency rangeincluding zero. At zero speed the rotational voltages arezero so that, if maximum torque capability is to be realised,rotor position must be used as a reference for control ofthe machine inverter. Rotor position sensing is equivalentto reference ot the machine voltages on no load.
At zero and low speeds, natural commutation of themachine inverter is impossible. However, the d.c.-linkcurrent can be pulsed by the supply converter and advantagecan be taken of this to change the machine-inverter conduc-tion pattern at each pulse of link current. A transition fromthe pulsed-link commutation mode to natural commutationcan be made at a suitable speed.
Paper T478P received 27th April 1979Dr. Issa was and Dr. Williamson is with the Department of ElectricalEngineering & Electronics, University of Manchester Institute ofScience and Technology, PO Box 88, Sackville Street, Manchester 1,England. Dr. Issa is now with the Department of ElectricalEngineering, College of Engineering, University of Baghdad, Iraq
ELECTRIC POWER APPLICATIONS, DECEMBER 1979, Vol. 2, No. 6
Many articles have dealt with this subject1"6 and a recentpublication7 presented and validated a method of analysiswhich is simple to employ yet satisfactorily predicts theperformance of a machine/converter combination. Thereare few details in the literature of the exact manner inwhich the necessary control and timing circuits are imple-mented. This paper describes one way in which such controlhas be2n achieved together with the measured performanceof a prototype system.
machine
Fig. 1 Power components of drive
30r
20
10-
500 1000speed,revs per minute
1500
Fig. 2 Maximum permissible d.c.-link current for natural commu-tation at full excitation of machine
20r
10
0 30 60 90 120no-load angle of advance (y),degrees of electricity
Fig. 3 Variation of maximum possible torque with yat 1500 rev/min and with full excitation
199
0140-1327/79/060199 + 06 $01-50/0
2 Operational requirements of control circuits
The machine inverter must be controlled by reference torotor position, and this will require a rotor-position sensortogether with decoding logic to obtain a no-load angle ofadvance 7 of the inverter. Reference 7 describes tests on adrive of this type using the machine detailed in Appendix 9.It is shown that the machine-inverter commutation abilitydeteriorates as speed decreases in the manner shown by Fig.2 which plots, against speed, the maximum current whichcould be commutated. Above a speed of about 200rev/minthe behaviour is virtually independent of speed and themaximum torque available is determined by excitationlevel and 7 (no-load angle of advance). Fig. 3 gives ameasured variation of this torque with 7 and shows that,for the machine tested, the optimum is for 7 = 90°. It isnoteworthy that 7 = 90° would give zero torque for amachine with a linear magnetic circuit and for zero overlapangle in the inverter; these two important effects can beincluded by the analysis of Reference 7.
At zero and low speed the technique of reducing thed.c.-link current momentarily to zero3 was used to commu-tate the machine inverter. This requires that the supplyconverter be phased back to inversion at each pulse; inaddition advantages accrue from simultaneously firing thethyristor TL across the choke (Fig. 1) since this reduces theamount of energy to be pulsed in the system.
The pulsing of the d.c.-link current gives a notchedrectangular waveforms of machine-winding current, theamplitude of which is not limited by machine commutationability and the fundamental component of which is notshifted in phase by overlap. Consequently, at standstill, theoptimum value of 7 will not be the same as that for naturalcommutation. Its value, which depends upon the currentamplitude, can be calculated as described in Reference 7,and for the prototype was 30°.
The control circuits must change 7 from its optimum atstandstill (pulsed-link commutation) to its optimum fornatural commutation above a speed Nm. For the prototype,these values were 7 = 30° at zero speed and 7 = 90° atNm = 230rev/min and it is expected that these angleswill not be greatly different for other machines. A singlestep of 60° in angle of advance could cause a commu-tation failure and this is avoided by using an intermediatevalue of 7 = 60° for speeds between 0-5Nm and Nm.
For regenerative braking, the simplest strategy is to con-tinuously trigger the thyristors of the machine inverter,thus rectifying in an uncontrolled manner, and controllingthe d.c.-link current by means of the supply converter whenoperating in the inversion mode. In addition to theserequirements, the control circuits must sense a speedreversal and modify the machine-inverter firing-pulsepattern accordingly.
3 Machine-inverter control circuits
The machine-inverter firing-pulse generator (f.p.g.), therequirements of which are described in the previous section,is shown in schematic form in Fig. 4. The major componentsof the f.p.g. are considered below.
3.1 Rotor position sensor (r.p.s.)
The construction of the r.p.s., based on a four-pole machine,is shown in Fig. 5. A disc is slotted on four concentrictracks so that with four photodiode-phototransistor couplesinformation changes at intervals of 30 degrees electrical.
Excess-three Gray code is used to avoid the generation ofspurious pulses in the event of misalignment of the photo-devices (this is the only code having one bit change betweenconsecutive states and between the first and last states).The outputs LMNO are processed by the decoding logicdiscussed later.
This disc is slotted at its outer rim so that, with twomore pairs of photodevices, two trains of pulses areproduced with frequency proportional to speed. Thesetrains are converted to a bidirectional analogue speed signalwhich is required for inverter firing control as well as overallclosed-loop speed control.
brakingsignal
speeddemand
directionsignal
reference
voitages
i
firingcircuits
1
•»—
speed rangediscriminator
decodelogic
- r D s
itspeed
Fig. 4 Components of machine-inverter firing-pulse generator
speed pulsetrains
Fig. 5 Rotor position sensor for 4-pole machine
speed
EX OR AND
n>Nr
(Z)
Fig. 6 Speed-range discriminator
200 ELECTRIC POWER APPLICATIONS, DECEMBER 1979, Vol. 2, No. 6
3.2 Speed-range discriminator
The operation of this component, which is required toprovide a logic signal for each of the three speed ranges0-0-5Nm, 0-5Nm-Nm and >Nm, is described by Fig. 6.Four operational amplifiers (two for each direction ofrotation) are connected to compare the actual speed signalwith reference signals corresponding to the limits 0-5 Nm
and Nm. The amplifier connections provide hysteresis inthe switching characteristics to avoid instability betweenmodes. The outputs of the amplifiers are limited anddecoded as shown to give the required logic signals X, YandZ.
3.3 Decoding logic
The logical relationships between the firing pulses (PyP6) for the thyristors Tx T6 of Fig. 1 and the outputsignals of the r.p.s. are given in Table 1. The decoding logicimplements Table 1 by means of groups of AND and ORgates as shown in Fig. 7. As well as having y set accordingto the speed range, the thyristors are supplied with continu-ous pulsing when the braking signal applied to the lastgroup of OR gates is logic 1. The braking signal is generatedby a comparison circuit in the speed-loop logic when thespeed signal, in either direction, exceeds the speed referenceand initiates regenerative braking.
4 Overall control circuits
The basic control loops of the drive are similar to those of aconventional thyristor/d.c. machine drive and the overallstrategy is shown by Fig. 8. An outer-speed regulating loopgives an error which is limited for use as a reference for thecurrent control loops. The circuits for generating pulses for,and controlling the angle of, the supply converter areconventional.
=pulse—generators
(Z)
Fig. 7 Decoding of rotor-position sensor
50 Hz _3-phase d.c.c.t.
firingcontrol
currentlimit
speedydemand
_speedsignal
Table 1: Relationships between logic states from position sensor andmachine-inverter firing pulses
Pulse 7 = 30° 7 = 60° y = 90° Rotation
/»,p*p,p.
Psp6
pl
p,p3
p.
Psp.
LMMNLMLNMNLNMNLNLMMNLMLN
LMOMNOLMOLNOMNOLNOMNOLNOLMOMNOLMOLNO
+ LNO+ LMO+ MNO+ LMO+ LNO+ MNO+ LNO+ LMO+ MNO+ LMO+ LNO+ MNO
LNLMMNLMLNMNLNLMMNLMLNMN
c.w.c.w.c.w.c.w.c.w.c.w.c.c.wc.c.wc.c.wc.c.wc.c.wc.c.w
4.1 Current and speed loops
The arrangement of the control circuits is shown in Fig. 9.Operational amplifier Al produces a speed error signalwhich is then limited to a value corresponding to a current-limit reference in the motoring mode. Three independentcurrent loops are used, only one of which is active at anytime. The loops with A2 and A3 are for motoring in thec.w. and c.c.w. directions, respectively, whilst the loop withA4 is used for regenerative braking. The braking loop has acurrent limit which is externally applied providing forindependent adjustment.
The logic block comprises interfacing circuits and logicgates which decode information provided by the d.c.pulsing circuit (discussed below), the d.c. c.t., the speedreference, the speed and the speed-error signals. The logicblock outputs are appropriately levelled signals which freeone current loop and block the other two, as well as thebraking signal for the f.p.g. The outputs of the three currentloops are combined by means of diodes so that the loopwith the highest output voltage is effective in controllingthe supply converter.
4.2 D.C. currentpulsing circuit
This circuit produces a pulse which starts at the leadingedge of each machine-inverter firing pulse and terminates atthe instant that the d.c.-link current is reduced to below thethyristor holding current. The pulse, when levelled, isapplied by the logic block to the current-loop amplifierswhere it overrides the current error and causes the supplyconverter to phase back to inversion; at the same time thethyristor TL across the choke is fired.
braking currentreference
speeddemand
supply-converter
link current
speed
braking signah
rotation sense Jcurrent pulsing
Circuit
-fp.g.
Fig. 8 Basic control strategy Fig. 9 Current and speed loops
ELECTRIC POWER APPLICATIONS, DECEMBER 1979, Vol. 2, No. 6 201
The circuit is shown in Fig. 10; the flip-flop B is set bymeans of the six monostables m.s.l m.s.6, providedthat the AND gate G is enabled which is only so at speedsless than Nm. The flip-flop is reset by a pulse from the zerocurrent detector, which results in thyristor TL beinginhibited and control being regained by the appropriatecurrent loop to re-establish link current for the nextmachine-inverter conduction period.
Transition from pulsed-link to natural commutationtakes place when the speed passes through Nm and gate G isinhibited. At the same time the machine-inverter firingangle 7 is advanced, and it is possible for this to occur afterthe last pulse for which G is enabled. If this happens, thefirst natural commutation will not be at the optimum valueof 7 and commutation failure could result. A smoothtransition is ensured by the monostable m.s.? which has anoutput pulse duration of 10 ms and which thus extends thetime for which G is enabled by 10 ms from the instant atwhich the speed passes through Nm. The 10 ms was chosento cover the worst possible situation in the prototye (i.e. achange of 30° in 7 at 230 rev/min).
monostables
Fig.9
link current zerocurrentdetector
Fig. 10 Current-pulsing circuit
5 Performance
In the prototype, for all tests through the complete speedrange, the motoring current limit was set at 17-5 A.
5.1 Torque-speed relationship
The variation of the maximum torque with speed has beengiven in Reference 7 which shows an increase at low speedscompared with the torque available with natural commu-tation. The speed-torque curves for various no-load speedsettings are shown in Fig. 11. At all speed settings there isa regulation which increases sharply when the current limitis reached. This regulation, less than 5% for currents belowthe limit, could be modified by adjusting the gain of theloop.
5.2 Lo w-speed opera tion
5.2.1 Effect of pulsing the choke thyristor: In Fig. 12 areshown current waveforms which illustrate the effect offiring thyristor TL during d.c.-link pulsing. The lower traceis for TL inhibited and the upper trace is for TL fired and
shows the substantial decreases in the times required toboth reduce current to zero and to re-establish current level.The resulting effect on winding current, and hence torque,is indicated by Fig. 12 where measured variations of torquewith speed are plotted, for two values of 7, with andwithout TL being pulsed. It is apparent that firing thechoke thyristor yields considerable advantage. A low valueof choke resistance will not only reduce the system lossesbut will decrease the time required to complete a currentpulse.
1500
11000cE
* 500
10 20torque, Nm
Fig. 11 Speed against torque characteristics
30
Fig. 12 Effect of firing link-choke thyristor TL upon link current
202 ELECTRIC POWER APPLICATIONS, DECEMBER 1979, Vol. 2, No. 6
5.2.2 Torque capability at low speed: There is no limitimposed upon link current in the pulsing mode by consider-ations of machine-inverter commutation ability. At lowspeeds, therefore, torques higher than those possible withnatural commutation can be obtained. Fig. 14 presents thetorques measured over the low speed range for 7 = 30° andfor several values of link-current limit setting. It is seen thattwice rated torque could easily be obtained at starting byuse of a control circuit which sets the current limit inaccordance with speed.
5.2.3 Commutation failure: The machine inverter can fail tocomutate in the pulsed link mode only if the link current isnot reduced sufficiently for blocking ability to be restoredin the outgoing thyristor. Such a failure is avoided bysuitable design of the zero-current detector of Fig. 10. If afailure does occur, however, although violent torquetransients might arise, recovery is possible without thedisastrous consequences which might be felt with a voltage-source inverter. An example of a commutation failure isillustrated by the recording of Fig. 15.
30r
20
10-
y=
0 100 200 300speed, revs per minute
400
Fig. 13 Effect of pulsing-choke thyristor on torque
A ~ TL pulsed
~_A TL inhibited
U Or
30
£2aT20
10
21A
17-5A
100 200 300speed, revs per minute
Fig. 14 Torque-speed characteristics for various link-current limitsin low-speed range (7 = 30°)
phasecurrent
'WVWWwvrvn linkcurrent
Fig. 15 Example of recovery from commutation failure in thepulsed-link mode
-speed
line
rr. -%,.?,,*t it 4
link"current-phasecurrent
-speed
M ' i i currentn. -phaseu c u r r e n t
Fig. 16 Starting from standstill
I f °n ?adH^rqUe ) d i f f e r e»t Paper speedsb full-load torque I v v
5.3 Transien t behaviour
Recordings of the important parameters during a start fromstandstill for conditions of no-load and full-load torques aregiven in Fig. 16. These show that only 19 current pulses(corresponding to about 3 cycles of machine windingcurrent) are required to accelerate to speed A m̂ on no load.When loaded, the number of pulses is increased to 26 butthe transition between commutation modes is equallysmooth. A feature to be noted is the distorted nature of themachine terminal voltage waveforms, which justifies thecomments made in Section 1 with regard to the method ofcontrol of the machine-inverter firing angle.
ELECTRIC POWER APPLICATIONS, DECEMBER 1979, Vol. 2, No. 6 203
A more severe test of the system is illustrated by Fig. 17which gives recording taken following a step in speed settingfrom +1500 to — 1500rev/min. Fig. 17a is the responsein the absence of regenerative braking, and Fig. 176shows the improvement gained by setting the brakingcurrent loop limit to 12 A (amplifier A4 of Fig. 9). Again,a smooth transition between all operating modes is evidentand the inherent flexibility of the drive is demonstrated.
-speed
-speed
*W&'-voltage
--speedzero
j i nkcurrentphasecurrent
Fig. 17 Reversal from + 1500 to —1500 rev/min
a without regenerative braking I ,.,„ . .b with regenerative braking j d l f f e r e n t p a ? e r s P e e d s
6 Conclusion
A method has been described for the successful control of anaturally commutated inverter-fed variable-speed synchro-nous machine. The supply converter (together with a firing-pulse generator and phase-angle control) and the speed- andcurrent-regulating loops are identical to those used for theconventional thyristor-fed d.c. machine drive. The machineinverter is of the same type as the supply converter, butrequires special components and circuits for its control andfor control of the system in the starting and braking modes.The design and implementation of these components arethe main contributions of this paper.
It has been demonstrated that the drive system consideredcan operate with any combination of directions of speedand torque, and that satisfactory starting and low-speedperformance is obtained by pulsing of the d.c.-link current.A smooth transition can be achieved between all modes ofoperation indicating the suitability of this drive for indus-trial applications. The performance and economics are bothenhanced as the rating of the drive increases; per-unit resist-ances decrease, resulting in improved efficiency and low-speed commutation, and the machine-inverter cost is lessthan would be a forced-commutated system.
7 Acknowledgments
The, authors are grateful to Joseph Lucas Co. for financialsupport and to the Department of Electrical Engineeringand Electronics, UMIST for the research facilities available.
8 References
1 SATO, N.: 'A study of commutatorless motor', Elect. Eng. Jpn.,1964, 84 pp. 42-52
2 SATO, N.: 'Induced-voltage commutation type commutatorlessmotor', ibid, 1971, 91 pp. 114-124
3 SATO, N.: 'Adjustable speed drive with a brushless d.c. motor,IEEE Trans., 1971, 1GA7 pp. 539-543
4 SATO, N.: 'A brushless d.c. motor with armature induced voltagecommutation', ibid., 1GA Conf. Publ. 71CI, October 1971
5 ENGLISH ELECTRIC: 'Improvements in and relating to thecontrol of synchronous machines', British Patent 1246970
6 OHNO, E., KISHIMOTO, T., and AKAMATU, M.: 'The thyristorcommutatorless motor', IEEE Trans. 1967, MAG 4 pp. 236-239
7 WILLIAMSON, A.C., ISSA, N.A.H., and MAKKY, A.R.A.M.:'Variable-speed inverter-fed synchronous motor employingnatural commutation', Proc. IEE, 1978, 125, pp. 113-120
9 Appendix
The machine used in the prototype drive was a 4-pole 50 Hz200 V 3kW induction motor. The rotor was wound with aconcentric field winding and d- and g-axis damper windings.The measured parameters (all £2 or mH referred to thestator per phase) were:
stator resistance = 0-24 £2field winding resistance = 0-57 £2tf-axis damper resistance = 2-3 £2g-axis damper resistance = 5-4 £2unsaturated magnetising inductance = 93-0 mHstator leakage inductance = 1-4 mHtf-axis transient inductance = 6-7 mHcf-axis subtransient inductance = 6-2 mHq-axis subtransient inductance = 12-7 mHpull-out torque as synchronous motor with200 V 50 Hz supply and with fullexcitation = 43-0 Nm
204 ELECTRIC POWER APPLICATIONS, DECEMBER 1979, Vol. 2, No. 6