Sunteți pe pagina 1din 6

Operational floating amplifier

J.H. Huijsing

Indexing ferm: Amplifiers

this, t.: OF has been given a floating output port,


Abstract: The operational floating amplifier which means hat the leakage or commonmode current
(OFA) is the most general ‘nullor’ implementation, +
I, = ( I o , IO2)/2out of the internal output-port circuit
which is a basic active two-port network element. is zero, and, consequently, the two output currents I,,
The definition, applications, and design of OFAs and - I o 2 are equal:
are described. The OFA accurately equates two
voltages connected to the terminals of the input I,, = - I 0 2 (4)
port, like a standard operational amplifier, but in
addition, an OFA accurately equates two currents
connected to the terminals of the output port.
Specific applications are given, e.g.: current ampli-
fiers, voltage-to-current converters, and gyrators.
The design of class-A and class-AB OFAs based

I
102
on several topologies is covered.

1 Definition

The concept of a universal active network element was Fig. 1 Operationalfloating mplijier (OFA)
introduced in 1954 by Tellegen under the name of ‘ideal v;, = 5 2
amplifier’ [l]. The name ‘Nullor’ was given to it by I,, = ~ I,,
Carlin in 1964 [2]. It is defined as a two port device with
input port voltage F and current I ; , and output port This fundamental current equation ensures that the
voltage V , and current I,, obeying the relations output-port can accurately perform operations on cur-
F=O and Ii=O, (1) rents.
When one output terminal is connected to one input
V , and I, arbitrary (2) terminal a simplified three-terminal nullor implementa-
The nullor concept is only applicable if an external tion can be defined. This device has been named the
network provides for a feedback from the output port to negative current conveyer’ by Sedra [4].
the input port and if the nullor provides for a high and The requirements for an OFA are more precisely
stable gain [3]. depicted by the equivalent circuit of Fig. 2. The equiva-
The operational floating amplifier (OFA) is the most lent circuit distinguishes between differential-mode input
general implementation of a nullor. It can best be defined voltage c,
= q1- q2 and current I, = (Iil - Ii2)/2,
as an operational amplifier equipped with a floating differential-mode output voltage K, = El - V,, and
input port, like that of a standard operational amplifier current I,, = ( I , , - IO2)/2,common-mode input voltage
(OA), but additionally equipped with an independently Vk = (F, - y2)/2 and current I, = (Iil - Ii2)/2 and
floating output port. ‘Floating’ is used here in the sense common-mode voltage V,, = (E, K2)/2 and current +
that no current is internally leaking away from a port, I,, = (Id + 1,2)/2.
and that the current and voltage of a port is not inter- Three elements are common to all active devices and
nally dependent on the common-mode voltage of that can be represented by a voltage controlled current source
port. Fig. 1 shows the schematic symbol for an OFA. If with transconductance I;, a differential input admittance
the gain A is infinite, the voltage difference F, = Fl - F2 xddr and a differential output admittance yodd. All other
between the two input terminals will be zero (eqn. 1) and elements represent parasitic errors and should be kept
hence the two input voltages VI and F2 are equal with small : Firstly, we have the differential-mode input offset
respect to ground: voltage Go,,. and current Ii0,,. . Second, we have the ele-
ments representing non-idealities of the floating character
b1 = v2 (3) of the input port. These represent the offset-voltage com-
This fundamental voltage equation makes the OFA suit- ponent Vk/H, which is the common-mode input voltage
able to perform operations on voltages accurately, just vc divided by the common-mode rejection ratio (CMRR)
like the standard OA. Because of this property, the OA H , two equal common-mode input bias currents libins,
has received widespread acceptance. Now in addition to and two common-mode input conductances Ykl and
Ec2. Third, we have the elements representing the non-
Paper 71226 (E10, E16h received 23rd August 1989 idealities of the floating character of the output port.
The author is with the Electronic Instrumentation Laboratory, Lkpart- These are two equal common-mode output bias currents
ment of Electrical Engineering Delft University of Technology, Delft, I,,, , two equal common-mode bias components lod/Hd,
The Netherlands which are the differential output current I,, divided by
IEE PROCEEDINGS, Vol. 137,Pt. G, No. 2, APRIL 1990 131
the differential-mode reiection ratio (DMRR) H , of the 2 Applications
output current, and two common-mode output- admit-
As we have seen before, the OFA has a floating output
tances ,Y and X c 2 .
port. This allows the feedback to be connected in series
Besides the equivalent circuit, the OFA can also be
with the output port, which results in circuits with a high
described by a matrix equation. The input and output
output impedance Or an output with a current-source
variables have been chosen according to the differential-

I I I I I I I I /
I-
Fig. 2 Equiualent circuit of an OFA

character. This opens up new applications in addition to


those of the standard OA.
A basic application of the OFA is the current amplifier
OFA: as shown in Fig. 3A. Due to the parallel connection of
the feedback to the input port the input impedance is
low, and the input voltage V, is nearly zero. The input
current I, passes through the admittance Y, and is
adsorbed by the negative output terminal. The output
port transfers the input current without losses to the
f positive output terminal. The voltage V,, = ll/Ylacross
Fig. 3A Current amplifier
Y, is nearly equal to the voltage across Y, . Consequently,
the current I,, = IlY2/Y,through Y, is added to the
1, = 1,K + Y2)K
output current I,. The total output current equals

mode and common-mode variables, as described above.


Due to the serial feedback connection at the output, the
The voltages and currents that should be made negligible
output impedance will be high, and the output current is
will be called error quantities. These are expressed as
nearly independent on the output voltage. If Y2 is elimi-
being dependent on the main signal quantities only. In
nated, the current amplifier becomes a current follower,
this way we obtain what we will call an error matrix:
obeying I, = - I,. The meaning of the negative sign is
that a positive physical current that is being sunk into
the input will subsequently be sourced from the output.
It is interesting to see that the current amplifier of
Fig. 3A is complementary, in terms of voltages and cur-
rents, to the well known voltage amplifier shown in Fig.
3B. The voltage transfer of the voltage amplifier is
V, = Vl(Z1 + Z , ) / Z , (7)
(5)

All elements of the error matrix should be made small.


The two-by-two upper left submatrix in the error matrix
represents the four elements common to all operational
amplifiers. These four elements can be made small by
using a large transconductance k;. The elements in the
third column represent the non-ideal floating character of
the input port. The CMRR H should be large and the
input ground admittance yi, small. The elements in the
fourth row represent the non-ideal floating character of
the output port. The DMRR Hd should be large and the
ground admittance , Y small. Further more, the additive
offset and bias elements should be small.
The equivalent circuit and error matrix both show The input impedance is high due to the serieal feedback
that no accurate elements are needed to obtain accurate connection to the input and, thus, the input current is
signal operations for voltages (eqn. 3) and currents nearly zero. The output impedance is low, and the output
(eqn. 4). The only requirements are a large amount of voltage is nearly independent on the output current. If Z,
gain, a large degree of isolation at the input and output, is eliminated the voltage amplifier becomes a voltage fol-
and low offset and bias values. lower, obeying V, = VI.
132 IEE PROCEEDINGS, Vol. 137, Pt. G , No. 2, APRIL 1990
Another basic and yet simple application of the OFA matched resistor bridge to obtain a high CMRR,
is the voltage-to-current convertor of Fig. 4. The input whereas, in our case, the current-source character at the
port of the OFA equates the voltage on the top side of output of the input OFAs provides for a high CMRR.
the admittance Y, with the voltage VI at the input. Due
to the serial feedback connection to the input port, the
input impedance is high and the input current I is nearly

Fig. 4 Voltage-to-current Convertor Fig. 5 Voltage-to-current conuertor with operational amplifiers


I , = r, v,
I , = - R , V,/R,R,
RJR, = RJR,

zero. The current IY, = V,Y, through admittance Y2 is The resistor R , has been split into two parts R,, and R 2 ,
picked up by the negative output terminal of the OFA, to balance parasitic effects, but basically this is not
and transferred without loss to the positive output ter- needed.
minal. Due to the serial feedback connection to the If the current-to-voltage convertor is placed in front of
output port, the output impedance is high and the output the voltage-to-current convertor, a current instrumen-
current I, is independent of the output voltage V,. It can tation amplifier arises that is able to measure the short-
be said that a ‘hard’ output current is generated, equal to circuit current of a floating source.
I , = Y, VI (8) Another application of the balanced voltage-to-
current convertor is the fully floating gyrator shown in
Note that the voltage-to-current operation has been Fig. 7. It can be applied to the simulation or inductors in
implemented with the minimum number of elements, viz. filter designs [SI.
one active element and one passive element, to determine
the transfer. If we had implemented the same voltage-to-
current function with standard operational amplifiers, we
would have needed three operational amplifiers O A ,
through to OA,, a bridge of four matched resistors R,,
through to R,, and a transfer determining resistor R , , as
shown in Fig. 5.
From the basic circuits mentioned, a large number of
applications can be derived. General-purpose voltage-to-
current convertors as shown in Fig. 4 can be applied, for
example, to accurately transfer signals to a different
common-mode voltage level, to overcome ground-loop
interference, and to engage current driven actuators.
A balanced version of the voltage-to-current convertor
+ - I
with OFAs has been applied to the instrumentation Fig. 7 Floating gyrator with two balanced voltage-recurrent conuer-
amplifier to Fig. 6. The two input OFAs, together with tors
RI,, perform a floating voltage-to-current conversion,
while the output OFA, together with RZ1 and R,, , per- Some other applications of OFAs are: simulated
forms a floating current-to-voltage conversion. The transformers, balanced integrators and differentiators and
widely used solution with op amps needs an accurately hybrids in telecommunication links.

R21-R22 v1
Rl

tronsducer voltage instrumentotion amplifier


bridge
Fig. 6 Instrumentation amplifer with balanced voltage-to-current conuertor and current-to-ooltage conuertor
V2 = V,(R,, + R , M ,

I E E PROCEEDINGS, Vol. 137, Pt. G , No. 2, APRIL 1990 133


3 Design the current in the output stage is 120 dB over a frequency
band of 10 kHz, and its untrimmed signal-to-offset ratio
An operational amplifier can be provided with a floating is 40dB. The simplicity of the output stage allows a

*++
output in several ways: For example, one could think of bandwidth of several tens of MHz.
a transformer coupling at the output, which is clearly not
a desirable solution. Another way is to use a standard 3.2 Class-AB output stages
OA with floating power supply, as depicted in Fig. 8. A disadvantage of class-A output stages is the poor
power efficiency of the output stage and also the relative
I 4vsp low dynamic current range, expressed by the ratio
between maximum output current and output-bias and
output-bias-noise current. To improve these quality
out factors, a class-AB output stage would be a better choice.
However in that case, signal dependent bias-current
source can no longer be used for isolation purposes,
because matching errors appear as a signal-dependent
leakage current from the output port. The general situ-
ation of a class-AB push-pull output stage is drawn in
Fig. 8 OFA design based on the application of aflonfing power supply
Fig. 10.

source V,,or the negative one V,, to the negative


output terminal. Hence, we have obtained our objective
as expressed in eqn. 4 without using accurate or matched
elements. However, we have now carried over our iso-
vGIHr 01
/IQl 1.314
03
w v G 3

lation problem to the power supply, which is also not I out -


desirable. c _ -

101 102

isolation, which is exemplified in the basic circuit of a


class-A output stage for an OFA in Fig. 9. I
IlSN
r I I
+Vrr
~~
v~E
Fig. 10 General push-pull class-AB output stage of an OFA

To obtain the equality of I,, and - I o , we can apply


three methods: First, equate the diagonal currents I,, =
I,, and I,, = IQs;secondly, equate the total supply cur-
rents I,, = Ism; thirdly, directly equate the two output
currents I,, = - Io2.

3.2.1 Current mirroring (Method I )


Fig. 9 Class-A OFA with current-source isolation of floating output Current mirroring can be used to equate the diagonal
port currents, or simpler, the paired currents I,, = I?, and
I,, = I,,, which results in the operational mirrored
The signal current at the positive output terminal amplifier (OMA) [7], as shown in Fig. 11. The sign of the
cannot flow through the fixed current source I, and must
be picked up by the collector of Q, and transferred to its
emitter. The fixed current sources I, and I, cannot
absorb the signal current, and so the emitter of Q, must
catch the signal and conduct it to its collector, which is
connected to the negative output terminal. The fixed
current source I, cannot hinder the signal flow. However,
an imbalance of the four current sources I, through to I,
causes a fixed output bias current I,,,,. A convenient
advantage of the long-tailed-pair connection of Q, and
Q, is that the base leakage current of one transistor com-
pensates for the other so that the equality of the output
currents is accurate to within about O.OOO1. If a higher
degree of accuracy is required, Darlington transistors can
be added to the output pair. The full design of a class-A VEE
OFA has been presented in Reference 6. Fig. 11 Class-AB operationnl mirrored amplifim ( O M A ) using two
For many applications the class-A OFA is adequate. supply-current mirrors
A differential-mode rejection of the output current of
Hd> 100dB can be obtained for frequencies up to current at the negative output terminal has been reversed
10 kHz. Filters can be built with Q-factors higher than with respect to that of an OFA, but this does not matter
1OOO in that frequency range. The signal-to-noise ratio of for a large number of applications. With this relatively
134 IEE PROCEEDINGS, Vol. 137, P f . G, No. 2, APRIL 1990
simple class-AB OMA circuit, a signal-to-noise ratio of However, there is an important disadvantage in using
130 dB in a band of 10 kHz, and an untrimmed signal-to- the OMA implementation based on Method I and in the
offset ratio of 60 dB can be obtained. The negative and OFA based on Method 11, namely that positive and
negative output currents have different deviations from
the ideal A, = - IO2/I,, = 1 slope, as shown in Fig. 13.

Fig. 15 Class-AB OFA realisation with output-current equalisation


using an instrumentation amplifier
Io*= - Ie,RmiRJRm,R,
VE;
Fig. 12 Class-AB OFA using supply current equalisation
I = I .I
.3

-101
"E E

Fig. 16 Class-AB voltage-to-current convertor directly based on the


instrumentation amplifier
I , = V,R21R,R.

Fig. 13 Different deviations of positive and negative currents from This is the result of different physical components for the
ideal slope using Methods I and I1 control of the positive and negative currents. When trim-
ming is not applied, the differences between the values of
integrated components may be of the order of twenty
percent to two percent, causing a distortion of the same
order when harmonic signals are being processed. When
these methods are applied in filters, no stable quality
factors can be expected of an order higher than 50 to 500.
At higher Qs, instability of the filter characteristic will
occur, as illustrated by Fig. 14.

3.2.3Output-current equalisation (Method 111)


This can be implemented by inserting a measuring
resistor R,, in the positive output lead and a resistor R,,
f/fo in the negative lead, and using an instrumentation ampli-
Fig. 14 Filter characteristics. Methods I and I! produce instability
fier with floating input and output [9] to equate the
for Qmhigher than 50 to 500 voltage over the measuring resistors, as shown in Fig. 15.
This third method does not have the disadvantage of
positive current conveyers are three-terminal class-AB the first and second methods (in which the transfer is dis-
nullor implementations using current mirroring [4,8]. torted at zero crossing) because the measuring resistors
R,, and RnZ process the positive as well as the negative
3 2 2 Supply-current equalisation (Method 11) output currents. The relation between the two output
This can be used to equate the total positive and negative currents is
supply currents I,, = I,. This has been used in the
example of an OFA design according to Fig. 12. If the =1 Rm1R2/Rm2Ri
~ 0 2 I ~ o (9)
total positive and negative supply currents are equal, the the DMRR is
common-mode output current I, = (fol + IO2)/2must be
zero, and thus the OFA requirement IS fulfilled such that H = ~ o *= /L1A1 - Rm,Rz/R,2R1) (10)
I,, = - I,, . The dynamic range of this class-AB OFA and the output impedance Z,,, is not dependent on com-
equals that of the OMA. ponent matching and can therefore be high.
IEE PROCEEDINGS, Vol. 137, Pt. G , No. 2, APRIL 1990 135
The ratio between maximum output signal current 2 CARLIN, H.J.: ‘Singular network elements’, IEEE Trans., 1964,
and output offset current is of the order of 50 dB, and the CT-11, pp. 67-72
3 TELLEGEN, B.D.H.: ‘On nullators and norators’, IEEE Trans.,
signal-to-noise ratio is 120dB. These values are of the :966, Cl-13, pp. 4 M 9
same order as those of the class-A OFA because the mea- 4 SEDRA, AS., and SMITH, K.C.: ‘A second generation current con-
surement and control circuit of the output current are veyer and its application’,IEEE Trans., 1970, CT-17, pp. 132-134
biased as in class-A. The remaining advantage of the 5 VOORMAN, H.O., and BIESHEUVEL, A.: ‘An electronic gyrator’,
class-AB OFA based on the instrumentation amplifier IEEE J., 1972, SC-7, (6). pp. 469474
6 HUIJSING, J.H., and de KORTE, J.: ‘Monolithicnullor - a univer-
over the class-A OFA is that better power efficiency, as sal active network element’, IEEE J., 1977, SC-12, (l), pp. 5 9 6 4
well as a good linearity is being maintained. 7 HUIJSING, J.H., and VEELENTURF, C.J.: ‘Monolithic operational
For a large number of applications where the OFA is mirrored amplifier (OMA)’,Electron. Lett., 1981.17, (3), pp. 119-120
used as voltage-to-current convertor, the instrumentation 8 HASLE’IT, J.W., and RAO, M.K.N.: ‘A high quality controlled
current source’, IEEE Trans., 1979, IM-28, (2), pp. 132-140
amplifier with floating input and output can be directly 9 HUIJSING, J.H.: ‘Instrumentation amplifiers: A comparative study
applied without the OFA construction, as shown in on behalf of monolithic integration’, IEEE Trans., 1976, IM-25, (3),
Fig. 16. The transfer is I, = 4 R,/R,R,. pp. 227-321

4 Conclusion

The OFA is the most general operational amplifier that


allows accurate operations on voltage signals as well as
on current signals. A high quality class-A OFA can easily
be realised with a signal-to-noise ratio of 120 dB in a fre- Author biography
quency band of 10 kHz and a bandwidth of several tens
of MHz. Three types of class-AB OFAs can be distin- Joban H. Huijsing was born in Bandung,
guished as having a better power efficiency than the Indonesia, on 21st May 1938. He received
the ingenieurs (MS) degree in electrical
class-A version. The OMA and the circuit based on engineering from Delft University of
supply-current equalisation have a better dynamic range, Technology, Delft, The Netherlands, in
however, these types show a significant nonlinearity. The 1969, and the PhD degree from the same
third type, based on the instrumentation amplifier, com- university in 1981 for work on oper-
bines the same dynamic range as that of the class-A OFA ational amplifiers.
with a good linearity. Since 1969 he has been a member of
the Research and Teaching Staff of the
5 References Electronic Instrumentation Laboratorv.
Department of Electrical Engineering, Delft University of Tech-
1 TELLEGEN, B.D.H.: ‘La recherche pour une serie comlete nology.
d’elements de circuit i d h u x non-lineaires’. Rendicouti Del Seminario His field of research is integrated signal conditioning circuits
Matematico e Fisico di Milano, April 1954.25, pp. 13+144 and integrated sensors.

IEE PROCEEDINGS, Yo1 137, Pt G, No 2, APRIL 1990

-. ~-

S-ar putea să vă placă și