Chapter 9
Realization of OTA-C filters using OTA-based FP AA
9.1 Introduction
Signi Cicant attention has been paid to active continuous-time filters over the last
three decades. Tens of thousands of journal articles and conference papers may have been
published and presented over the yeurs. The changes in technology hnvc required new
approaches. Thus as cheap, readily available integrated circuit operational amplifiers
(Op-Amps) replaced their discrete circuit (based on vacuum tubes), it became feasible lo
consider actjve-RC ti lter circuits using large numbers of Op-Amps, and new improved
architectures emerged (61,63.65).
Similarly the development of integrated operational transconductance amplifiers
(OT A) led to new filter configurations which reduced the number of resistive
components, and allowed transconductance-mode rather than voltage-mode. This givt:s
rise to OT A-C filters, using only active devices and capacitors, making it more suitable
for integration (63,65-66).
The demands on filter circuits have become ever more stringent as the world of
electronics and communications has advam;ed. For cxnmple, greater demands on
bandwidth uti lization have required much higher performance in filters in Lerms of their
attenuation characteristics, and particularly in the transition region between passband and
stopband. This in tum has required filters capable of exhibiting high quality factor but
having low sensitivity to component changes, and offering dynamically stable
121
performanc~. In addition, the conLinuing increase in the operating frequencies of modem
circuits and systems increases the need for active filters that can perform at these higher
frequencies; an area where the Q-enhanccd LC active filter emerges.
9.2 Types of Analog filters
9.2.1 Active-RC Filters
Acclvc- RC filters how been around for some time as a means Of overcoming the
disadvantages nssociated with low-frequency pussive RLC fillers (of which the use of
inductors is one). They offer the opportunity to integrate complex filters on-chip, and do
not have the problems that the relatively bulky. lossy. and expenc;ive inductora bnns in
particuJar their stray magnetic fields that can provide unwanted coupling in a circuit or
system [45).
The Sallcn and Key circuit, shown in Fig. 9.1, (which uses a voltage amplifier,
resistors, and capacilors) is one of the most popular and enduring active-RC filter
architectures. It has been around for about 48 years. Active-RC filters have been widely
used in various low frequency applications in telecommunication networks, signal
processing circuits, communication systems, control, and instrumentation systems.
However, they cannot work at higher frequencies due to OpAmp frequency limitations
and are not suitable for full integration if large resistors arc n:quired. They are also not
electronically tunable and usually lmw complex structures. The most successful approach
to overcome these drowbacks is to replace the conventional OpAmp in active-RC filters
122
Cl
Rt 2 R7 3 5
Vo
4 XOP1 :, r RB
0
RA
Figure 9.lSallen and Key circuit
by an OTA giving rise to OT/\-C filters. Programmable high-frequency active filters can
therefore be achieved by incorporating the OT A. OT A-C filters also have simple
structures, and can operate up to several hundreds of Miiz [65,67].
9.2.2 Switched-Capacitor Filters
Another type of filter, called the switched-capacitor filtt::r, has become widely
available in monolithic form. The switched-capacitor Apprm1ch overcomes some vf the
problems inherent in standard active filters, while adding some interesting new
capabilities.
Integrated switched-capacitor filters need no external capacitors or inductors, and
their cutoff frequencies are set to a typical accuracy of ::1:0.2% by an external clock
frequency. This allows consistent, repeatable filter designs using inexpensive crystal-
controlled oscillators, or filters whose cutoff frequencies are variable over a wide range
simply by changing the clock frequency. In addition, switched-capacitor filters can have
low sensitivity to temperature changes.
123
Switched-capacitor filters are clocked, sampled-data systems; the input signal is
sampled at a high rate and is processed on a discrete-time, rather than a continuous basis.
This is a fundamental difference between switched-capacitor filtt:rs and conventional
active and passive filters which ore also referred to "continuous time" litters.
The operation of switched-capacitor fillers is basecl on the ability of on-chip
capncilors and MOS switches to simulate resistors . The values of these on-chip capacitors
can be closely matched to oLher capacitors on the IC, resulting in integrated filters whose
cutoff frequencies are proportional to. and delt!nnined only by, the external clock
fre4ucncy. Now. lhese in 1c grntcd filters a rc ncurly olways bascu on swte-vnrluble uclivc
filter topologies, so they are also active filters , but nonnal tenninology reserves the name
" active filter" for filters built using non-switched or continuous active filter techni4ues.
The primary weakness of switched-capacitor filters is that they have more noise at their
ourput:J both ram.lom noise anti clock feedthroug11 than that of stanc.tarc.t acuve filter
circuits [65].
9.2.3 OT A-C Filters
ln recent years OTA-based high frequency integrated circuits, filters and systems
hnvc been widely investigated. This is due to their simplicity, electronic tunnbility nnd
suitability for high frequency operation due to open loop configuration.
The OTA has been implemented widely in CMOS und bipolar and also in
BiCMOS and GoAs technologies. The typical values of transconductanccs are in the
rungc of tens 10 hundreds of µS in CMOS and up to mS in bipolar technology. The
CMOS OT A, for example, can work typicully in the frequency range of SOMHz to
124
several I OOMHz. Linearization techniques make the OT A able to handle input signals of
the order of volts with nonlinearities of a fraction of one pcrccnl.
Although OT A-C filters have the potential to operate at rt!latively high
frcyuencic:s the liuc::,ar range of the used transconductance limits the dynamic range. Also
the OTA-C filtt:rs become very power hungry ut Gll7 range. At high fn:quencics
sensitivity to parasitic components becomes very prominent r65}.
9.3 Elementary Transcondutor Building Blocks
With gm-C filters we need n variety of clememnry building blocks wi1h which the
filtcr:s are constructed [45]. As the name "gm-C filters" suggests, we wish to employ only
transconductors and capacitors as basic components [68]. In general. there arc two design
approaches. The first approach begins with a synthesized prototype design and the
necessary transfonnation to obtnin the lumped-element prototype. Ba:scd on lhis
prototype design, each passive element is replaced by its active equivalent. for instance,
an inductor is replaced by n gyrator, which can be implemented using transconductors
and capacitors. Capacitors in lwnped design are nonnally not replaced since they can be
readily implemented in intebrrated circuits. Source and load resistances are replaced by
self-feedback transconductors. Thus Gyrator filters are an alternative for active
implementations of the LC band pass ladder. This realization implies the replacement of
all the inductors and the resistors with active structures that emulate their functionality . Jn
addition, it is helpful to have a way to realize or simulate resistors because these are
difficult to implement with sufficient accuracy and over an udequate range of
values[63,65,67).
125
9.3.I Resistors
There is generally little need for resistors in the area of gm-C filters with the
notable exception of source and load resistors in doubly tem1inatcd LC ladders. We need
at least lwo resistors in a lossless ladder. For sensitivity design, source and load resistor
should be incorporated into the design strategy (45).
To realize a resistor, the difli:rcntial tnmsconductance was taken to connect two
inputs to two different voltages and feed the output back to the input as shown in Fig. 9.2.
Always paying attention that feedback is negative, the two equations (9. 1) and (9.2) art:
obtain
Ii= lo = I::. gm• (Vl-V2) (9.1)
Thu!:, the resistor is
R -(VI- V2)fl = II gm ('>.2)
We should be ct:nain to 1m1intain the negative feedback when fonning the gm-
based resistor. If the feedback connection becomes positive, we simulate a negative
resistor. Vi I Ii = - R = - l I gm, in a differential form. Such a resistor is used, for
instance, to compensate the transconductor losses or to illuminntc inductor losst:.c; when
very smnll but real spiral- wound inductors are used on ICs for filters at the high
frequencies.
Figure 9.2 Active implementation of a resistor using "gm" module with one terminal grounded
126
The implementation of a floating resistance is more complicated since two
different reference voltages must be created. Therefore, a 1.:ommon practice is lo
transform the input voltage source to an input current source (64). The transformation is
shown in Fig. 9.3. It can be seen that the noating source resistance hos been transformed
into a grounded resistance that can be implemented with the structure from Fig. 9.2. The
input current Iin • is detennined by forcing the equivalence of the two circuits. The
functionality of the sources is equal if the current and the voltage drop on the input
impedance of the LC network are the sume after the transformation (64]. Then it holds
true.
r . Zv: = [' . R~Zr.,~ m R Z 1
' 1 R Z S + l <" lt: + LC (9.3)
R. 11:1 .. 1,.
! v,I \
z~, c ) (
Cl)
b )
Figure 9.3 a) Transformation of the input voltugc source to un input current and
b) The OTA-C implementation
Relation (9.3) results that the equivalent input current source must be scaled with
the reciprocal of the source resistance RS and will be equal to Vin/RS. The
127
transformation yields the OTA-C structure from the fi g. 9.3 (h), where the fir:>l
transconductance amplifier implements the sculeu current source.
An advantage of the I/gm resistors is that the resisturs are dectronically variable
because they depend on a bias current (or voltage). A traditional integrated resistor.
implemented say, as n diffused or a deposi ted layer, does not offer this possibility. It is
most helpful in gm-C filter design to have the " resistors" technologically match and track
the active devices. The OTA-based resistor offers a resistance of 19.2 K Olun to 11.07 M
Oluns for transconductance of OT A varying from 0.45 uAN to 7. 7 uAIV.
9.3.2 Gyrators
Another element t11at must be replaced in the uctive implcmenwtion is the
inductance. Inductors are implcmcn1.:d by using an Impedance conversion. The
impedance conversion is u procedure in which an impt.:dan~ 7. is inverted and mirrored
to the input of nn active structure. The inversion ond the impedance mirroring is
implemented with a gyrator whose input current is proporlional to the output voltage and
the output current is proportional to the input voltage through a conversion factor r, called
the gyration resistance [66 j. The general structure of a gyrator as n two-pon and the
equivalent input resistance are shown in Fig. 9 .4.
Figure 9.4 Cen ernl s tructure of a gyrator represented as n two-port
128
The equations of the circuit. shown in Fig. 9.3, can be written as follows.
{
~,:: : ~ r 11
1,
v; =-I~ . z (9.4)
The input impedance of the structure is proportional with the reciprocal of the impedance
Z and with the squared gyration resistance. lt re~ults:
I . ,.J I , I ' z ... - T. = z (9.5)
Creating a floating node whose voltage is referenced to the cvound does the
inversion of floating impedance. This procedure can be con:;idered as the replacement of
the Z impedance ground connection from Fig. 9.4 with a floating voltage. The
transformation is done by using a second gyrator that creates the floating nodes at the
terminals of Z, as shown in the Fig. 9.5. It ctm be dcmonstruted that both the input and the
output impedances of the circuit are equal to r /Z. From the analysis it results that nn
inductance cnn be implemented by the inversion of a capacitance. In the OTA-C design
techniques the gyrators are realized as a pair of transconductance amplifiers connected in
a feedback loop. The structures used to simulate the functionality of a grounded and a
noating inductance is presented in Fig. 9.6 and Fig. 9.7. Ilcrc 1hc gyrution r<:sistancc is
equal to I/gm. The value of capacitance can be calculated according Lo Lhis equation: C =
gm2.L. The OT A-based inductor offers nn inductance of 7.6ml1 to 33H for
trw1sconductance of OTA varying from 0.45 uAN to 7.7 uAIV and programmable
capacitor array having a capacitance from 0.45pF to S.75pF.
129
I . v
1 ..
1. Z r~ V.~V:
Figure. 9.5 The inversion of a floating impedance with two gyrntors
.ZQ + + g.,
c I ...-·' c
¢ ~-·1 .--·' + + g,,, ._ -
...........
Figure 9.6 The OT A-C Implementation of a grounded inductance
c
Z. ---~ Q
...... ··
~~ .......... ~__.__::_ + >----+--.--
...........
~~~.+~
+ ~
i:
Figure 9. 7 The OT A-C implementation of a floating inductance
130
9 .4 hnplcmentation of 6th order Butterworth Band-Pass filter
A 6th order Butterworth Band-pass filter hnving fo llowing specifications.
Center frequency = 51.961 Kl lz.,
Bandwidth= 60KHz.
Stopband = 24dD,
Jnput Impedance = 300K Ohm
and Output Impedance = 300K Ohm.
The prototype of 61h order filter is shown in Fig.9.8. Fig.9.9 shows the realizution
of basic components like ' register', ' floating inductor' and ' grounded inductor' [65-66]
The Gm-c realization of 61h order Butterworth band-pass filter is presented in
Fig.9.10. in which each OTA is realized using on~ cross-coupled OTA. The capadto1
value also includes pnrasitic capacitance uf switches and connectioni:. The
transconductance of ench OTA is adjusteu individually. By adjusting the analog voltage
Vetri, one can change the center frequency of the filter. The Fig. 9.11 shows
implementation of 6th order Butterworth band pass filtt:r in our fPAA using clusters. In
Fig. 9.11 , red lines show used OT As.
dpolo , dpo6o 3 dipole 5
R1 L 1 C1 L3 C3 ~--~~~
L2
Rl• 300K R2 • 300K
Cl • 11.79859 pf
C2• J7 .66919pF C3 • ll .79859pF
ll • 795.145mH
l2• 5J0.937mH l3 • 795.14SmH
Figure 9.8 Prototype of 611' order Buttcnvorth bond-pass filter
131
The above structure of 61h order Butterworth Band-pass filter will be
implemented.
L 'b~~f(f
R=-2._ g,,,
c L = -g;_
c L =-
R!
Figure 9.9 Realization of register, floating inductor and grounded inductor
F igure 9.10 Gm-C realization of 6'h order Butterworth ba nd-pass filter
132
" . . '
Cluster-1
Cluster-4
Cluster-2 Cluster-3
Cluster-5
.figure 9.11 Implementation of 6' ~ order Butterworth band-pass filter
9.5 Implementation of 121h o rder Butte rworth Bandpas." filter U!jiog FPAA
A 12"' order Ruuerworth Band-pass filter having following ~-pecifications.
Center frequency = 51. 96 1 K Hz.
Bandwidth = 60K.H7.,
Stop band "" 24dB,
Input lmpedwicc '" 300K Ohm,
Output lmpedance - 300K Ohm.
The prvloi.ype filter is shown in fig. 9.12.The Gm-c realization of 12lh order
Butterworth band-pass filter is presented in Pig.9.13 Titis realization implies the
replacement of aJI the inductors and the resistors with gm-C structures that emulate their
functionality. The transconductance of each OTA is adjusted individunlly. By adjusting
133
the analog voltage Vetri, one can change the center frequency of filter. The Fig. 9.14
shows implementation of 12th order Butterworth band pass filter in our FPAA using
clusters. In Fig. 9.14, red lines show used OT As.
dipole 5 dipole 7 dipole 9
L6
<ipole 6
"R1• 8001< R2 · !DOI< C1 • U.71411 pF C2 • 12.-..;pl' ca c e.10498of' C4 • 17 JJ745CllDF
C5 •~ ce s Uf611pF
R2 L1 =411.7ClmH l2 • TIIO.H1ml I U • 1.597df l.4 • 6411..«emH L6• 1,12!iH l6-2.0CiU1
dipole 10
Figure 9.12 Prototype of 12•h order Buttcrwo1-th baud-pass filter
Figure 9 . 13 Gm-C realization of 121h order Butterworth bond-pa.ss filter
134
Cluster 1 Cluster~ C'lu.;ter 3 Cluster 4
Clus ter .=' Cluster 6 Clu:> te1 -
Figure 9.l 4 Implementation of 12•h order Butterworth band pass filter in FP AA
\ 135