GB2246486A - Programmable gain instrumentation amplifier - Google Patents

Abstract

A programmable gain amplifier 56 (shown here coupled to identical amplifier 57 in differential mode) has gain selection transistors 28(A, B) controlled by signals on gain selection conductors 42(A, B) to select gain resistors 25(A, B). The output transistor 43 is controlled by amplifier 46A whose input is, in turn, controlled by the effect of the selected gain resistor on transistor 21A or Band current source 23. Fig 4 shows a similar circuit with same bipolar transistors being replaced by FETs. <IMAGE>

Description

PROGRAMMABLE GAIN INSTRUMENTATION AMPLIFI ER The invention relates to progammable gain amplifiers, and particularly to a programmable gain instrumentation amplifier that can be conveniently implemented on a monolithic integrated circuit chip using conventional bipolar bi-fet manufacturing processes.

A variety of instrumentation amplifiers are knows.

Fig. 1 shoes a "standard" three operational amplifier instrumentation amplifier that includes teo "gain cells" connected together so that a differential input signal is applied to the positive inputs of the to gain cells, and the outputs of the two gain cells are applied as a differential input to a third operational amplifier that is connected as a difference amplifier.

A problem with the prior art circuit of Fig. 1 is that it is impractical to use this design to provide a programmable gain instrumentation amplifier in a standard monolithic integrated circuit chip. In order to provide programmable gain for the circuit of Fig. 1, it would be necessary to duplicate too many transistors and resistors to achieve an efficient design. Furthermore, the common mode rejection of the instrumentation amplifier of Fig. 1 is very dependent upon the preciseness of matching of the resistors 11, 12, 15, and 16. This is undesirable because extremely precise matching of resistors with high manufacturing yields is difficult to achieve.

Another prior approach to implementing an instrumentation amplifier is shown in Fig. 2. This figure shows a circuit used by PMI (Precision Monolithics, Inc.

of Santa Clara, California) in their AMPO1 and AMPOS instrumentation amplifiers. The input signal VIN N is buffered by emitter followers operating at constant current, with the gain resistor 25 being connected between emitters. The difference between the driving currents 23 and 24 and the signal current through gain circuits 25 then are fed into the collectors of transistors 28 and 29, the emitters of which are connected to a scaling resistor 30 and current sources 34 and 35. The difference between the collector currents and the emitter current sources then are forced to flow in the scaling resistor 30.Voltages at the emitters of transistors 28 and 29 then are used b feedback circuits 36 and 37 to force the output of the instrumentation amplifier to be a multiple of the input signal.

Again, the instrumentation amplifier of Fig. 2 cannot be easily modified to provide programmable gain.

Utilizing discrete field effect transistors (as indicated in dotted lines) to switch in additional gain resistors such as 25A appears to be the only practical technique.

This approach is very inconvenient because of the necessity of using large, expensive field effect transistors in a hybrid integrated circuit device, since it is necessary that the "on" resistances of the field effect transistors be very small compared to the resistance of gain resistors 25 and 25A. Furthermore, the temperature dependence of the field effect transistors introduces errors into the gain. The large gate-to-drain capacitive coupling of the field effect transistors would greatly limit the bandwidth of a programmable instrumentation amplifier implemented in this manner.

Also, the logic levels at the gates of these devices would need to track the imput signal, thereby requiring complex circuitry to be included to accomplish the tracking.

It is another object of the invention to provide a programmable gain instrumentation amplifier having a highly symmetrical circuitry which results in rejection of imbalances due to mismatching of transistor parameters.

It is another object of the invention to provide a programmable gain instrumentation amplifier which is substantially free from effects of "glitches" in its output signal caused by switching various gain control resistors into and/or out of the circuit operation.

It is another object of the invention to provide a programmable instrumentation amplifier having short settling times.

Embodiments of the present invention will now be described by way of example with reference to the accompanying drawings, in which Figure 1 shows a circuit diagram of a three operational amplifier instrumentation amplifier according to the prior art.

Figure 2 shows another circuit diagram of an instrumentation amplifier according to the prior art, with a possible, but undesirable modification to provide programmable gain.

Figure 3 shows a circuit diagram of a bipolar transistor implementation of the invention.

Figure 4 shows a circuit diagram of an alternative, presently preferred embodiment of the invention.

Briefly described, and in accordance with one embodiment thereof, the invention provides a programmable gain amplifier including first, second, third, and fourth gain selection transistors, the control electrodes of the first and second gain selection transistors being connected to a first gain selection signal, the amplifier also including control electrodes of the third and fourth gain selection transistors connected to a second gain selection signal, first and second output transistors the first current carrying electrodes of the first output transistor and the first and third gain selection transistors being coupled to a first constant current source, the first current carrying electrodes of the second output transistor and the second and fourth gain selection transistors being connected to a second constant current source.The amplifier further includes first, second, third, and fourth input transistors, the first current carrying electrodes of the first, second, third, and fourth input transistors being connected, respectively, to the second current carrying electrodes of the first, second, third and fourth gain selection transistors, the control electrodes of the first and third input transistors being connected to a first input terminal.The control electrodes of the second and fourth input transistors are connected to a second input terminal, the second current carrying terminals of the first and third input transistors being connected to a third constant current source, the second current carrying electrodes of the first and third input transistors being connected to a third constant current source, the second current carrying electrodes of the second and fourth input transistors being connected to a fourth constant current source. The amplifier further includes a first gain resistor connected between the first and second current carrying electrodes of the first and second input transistors. The amplifier also includes a second gain resistor connected between the first current carrying electrodes of the third and fourth input transistors.The amplifier includes circuitry for biasing the control electrodes of the first and second output transistors. First and second output currents flow through the second current carrying terminals of the first and second output transistors, respectively. The gain of the amplifier is selectable to a first value determined by the first gain resistor, by turning on the first and second gain selection transistors in response to the first gain selection signal and turning the third and fourth gain selection transistors off, or to a second value determined by the second gain resistor by turning on the third and fourth gain selection transistors in response to the second gain selection.

Fig. 3 shows a bipolar transistor implementation of the invention. The instrumentation amplifier of Fig. 3 includes two input terminals 7 and 8, between which an input voltage VIN is applied. Input terminal 7 is connected to the bases of NPN input transistors 21A and 21B. Similarly, input terminal 8 is connected to the bases of NPN input transistors 22A and 22B. The collectors of input transistors 21A and 21B are connected to constant current source 23 and to the input of an inverting amplifier 46A, the output of which is connected to the base of NPN output transistor 43. Similarly, the collectors of input transistors 22A and 22B are connected to constant current source 24 and to an input of inverting amplifier 46B, the output 50B of which is connected to the base electrode of NPN output transistor 44. Constant current sources 23 and 24 supply equal currents from +V.

The emitter of input transistor 21A is connected to one terminal of a first gain resistor 25A, the resistance of which is R1, and to the collector of NPN selection transistor 28A. The base of transistor 28A is connected by conductor 42A to receive a first gain selection voltage VRl . The emitter of transistor 28A is connected by conductor 45A to the emitter of output transistor 43 and to constant current source 34, which returns a constant current to -V.

Similarly, the emitter of input transistor 22A is connected to the other terminal of gain resistor 25A and to the collector if gain selection transistor 29A. The base of gain selection transistor 29A is connected to conductor 42A, and its emitter is connected by conductor 45B to the emitter of output transistor 44 and to constant current source 35, the current of which is equal to the current of constant current source 34 and is connected to -V.

The emitter of input transistor 21B is connected to one terminal of a second gain resistor 2sub, the resistance of which is R2, and to the collector of an NPN gain selection transistor 28B. The emitter of gain selection transistor 28B is connected to conductor 45A, and its base is connected by conductor 42B to a second gain selection voltage VR 2 . The emitter of input transistor 2-2B is connected to the other terminal of gain selection resistor 25B and to the collector of gain selection transistor 29B, the base of which is connected by conductor 42B to Or2 . The emitter of transistor 29B is connected to conductor 45b.

The collector of output transistor 43 is connected by conductor 53A to the inverting input of an operational amplifier 38 and to one terminal of a feedback resistor 48, the resistance of which is R. The output of operational amplifier 38 is connected to the other terminal of resistor 48 and b conductor 17 to VOUT The collector of output transistor 44 is connected by conductor 53B to the non-inverting input of operational amplifier 38 and to one terminal of resistor 47, the resistance of which is R. The other terminal of resistor 47 is connected to ground.

An output current Ioi flows through conductor 53A into the collector of output transistor 43, and an output current Iso 2 flows through the conductor 53B into the collector output transistor 44.

One skilled in the art can recognize that if VR1 is at a relativeiy low voltage and VR2 iS at a relatively high voltage, so that gain selection transistors 28A and 29A are off, no current will flow through the emitters of input transistors 21A and 22A, that gain resistor 25A is effectively switched out of the circuit, and the gain of the programmable gain amplifier is determined by gain resistor 25B, i.e., by R2. Similarly, if VR1 is at a high voltage and Via 2 is at a low voltage, gain selection transistors 28B and 29B are off, no current flows through input transistors 21B and 22B, gain resistor 25B is effectively switched out of the circuit, and the gain of the programmable gain amplifier of Fig. 3 is determined by gain resistor 25A, i.e., by R1.

Other aspects of the operation of the programmable instrumentation amplifier of Fig. 3 can be understood by assuming that Vi is at a high voltage, and VR2 is at a low voltage. Then it can be seen that the differential input voltage Vi. appears directly across ki. This differential voltage across resistor R1 creates a differential current I through R1, which appears both as an incremental increase in Iso 2 and an incremental decrease in Iol.One skilled in the art will recognize that this occurs because the currents flowing through input transistors 21A and 22A are constant, because constant current sources 23 and 24 are equal, and because the constant current sources 34 and 35 are equal.

The two output currents Io1 and Iso 2 and the incremental increase and decrease therein flow through resistors 48 and 47, respectively, thereby producing an incremental decrease in the voltage across resistor 47 and an incremental increase in the voltage across resistor 48. The differential decrease in voltage across resistor 47 will be simply I times R, and there will be an equal incremental increase across resistor 48, where I is equal to Iso 2 minus Ion. Since the current difference I is equal to V1 N divided by Rl, the voltage gain expression for the amplifier in Fig. 3 is equal to 2R divided by R1 if VR is at a high voltage and VR2 is at a low voltage, and is equal to 2R divided by R2 if VR2 is at a high voltage and VR1 is low.

A typical value of R might be 30 kilohms, and values of Rl and R2 might be in the range from 60 ohms to 60 kilohms.

The bandwidth of the above-described programmable instrumentation gain amplifier can be quite high, typically more than one megahertz for gains of 1 to 100.

If the amplifier gain exceeds about 100, some of the transistor parameters come into effect, resulting in reduced bandwidth. The gains of inverting amplifiers 46A and 46B can be quite low. My circuit simulations indicate that the circuit will function accurately with the gain of inverting amplifiers 46A and 46B as low as about 15.

When the instrumentation amplifier circuit of Fig. 3 is initially "balanced", i.e., when VI N equals 0, current sources 23 and 24 supply equal currents through the selected input transistors and gain selection transistors. If gain resistor R1 is selected, the currents through transistors 28A and 29A are equal to Ioi and 102. The voltages at the inputs of amplifiers 46A and 46B have established identical quiescent values.

To now understand that circuit operation, assume V1 N is increased from zero to Vi N . That produces a voltage drop of Vl N across R1 and a current equal to V1 N divided by R1 flows from the left-hand to the right-hand side of R1. That current attempts to flow into the collector of transistor 29A and tends to increase the voltage of the emitter of transistor 22A, reducing its collector current. This in turn tries to reduce the current flowing through constant current source 24, thereby producing an increase in the voltage on conductor 4913. Inverting amplifier 46B produces a corresponding decrease in the voltage on conductor 50B, tending to reduce through the current through output transistor 44.

That in turn allows more of the constant current from constant current source 35 to flow through transistor 29B, the collector voltage of which adjusts so that the current increment VI n divided by R1 now has some place to flow. Since the current of current source 35 is constant, the increased flow through transistor 29B results in a decrease of I in the output current 102 In a similar manner, essentially the opposite operation occurs in the left half of the instrumentation amplifier. The current I flowing from left to right through resistor R1 robs current that otherwise would flow through the collector of transistor 28B. This causes the emitter voltage of transistor 21A to decrease.This decrease turns transistor 21A on a bit harder, causing it to attempt to draw more current from constant current source 23. This reduces the voltage at the input of amplifier 46A, which produces a corresponding increase on conductor 50A, causing a corresponding increase in the portion of the constant current 34 that flows through output transistor 43, and producing an increase of I in I01.

Note that in "differential subsections" 56 and 57 of the programmable gain amplifier of Fig. 3, the current flowing through the selected gain selection transistor (i.e., 28A or 282) is equal to the difference between he constant current source 23 and the current flowing though the selected gain resistor (i.e., resistor R1 or R2 ) Therefore, the collector current of the output transistor 43 must be equal to the signal I current plus the difference between constant currents 23 and 24. Thus, when the two differential subsections 56 and 57 are connected together as shown in Fig. 3 to provide the instrumentation amplifier, the output current Iso 2 is equal to a constant plus the signal current I while the output current Ioi of the other side is a constant minus the signal current I.Of course, more than two selectable gain resistors and associated circuitry can be provided in the same manner as those described above.

It should be noted that the common mode signal component of V1 N are not carried through to the inputs of the output amplifier 38. Therefore, resistor mismatches in the circuit do not limit the common mode rejection.

The common mode rejection is primarily determined by the output impedances.

A major advantage of the circuit described in Fig. 3 is that only five additional devices, including two input transistors, two gain selection transistors, and one gain resistor, need to be added to provide yet another switchable gain value for the instrumentation amplifier.

Since the prior instrumentation amplifiers have utilized signals produced at the collectors of the input transistors to develop the output voltage, I believe that the instrumentation amplifier of Fig. 3 may be novel and highly useful even without the programmable gain feature.

Since the circuit operates on current steering principles, long settling times of signal "glitches" due to capacitive coupling produced by fast transitions of the switching voltages VR1 and VR 2 are avoided. The high degree of symmetry of the circuit of Fig. 3 causes errcrs due to variations in transistor parameters to be cancelled out, and allows a high degree of symmetry in the IC chip layout, resulting in excellent thermal balance of the chip.

Referring now to Fig. 4, an alternate, presently preferred embodiment of the invention is shown in which the output transistors 43 and 44, the gain selection transistors 28A, 28B, 29A, and 29B have been implemented with P-channel junction field effect transistors, which can be integrated in a monolithic "bi-fet" IC structure of many conventional bipolar IC manufacturing processes.

The NPN input transistors of Fig. 3 have been replaced in Fig. 4 by PNP input transistors designated by the same reference numerais. The reason that I prefer the circuit of Fig. 4 over that of Fig. 3 is that the base currents of NPN gain selection transistors 28A, 28B, 29A, and 29B "add into" their respective emitter currents. Therefore, variations in those base currents can cause modulation of the corresponding collector currents. Since the emitter currents referred to are respectively determined by constant current sources 23 and 24, a change in the base current of one of the gain selection transistors results in a corresponding change in its collector current, and hence a change in the signal current flowing through the collector of selected gain resistor. This can result in an error in the instrumentation amplifier gain. In the circuit of Fig. 4, such error is avoided by utilizing Pchannel junction field effect transistors as the gain resistor switching devices, because the gate current of such field effect transistors is negligible.

Claims (1)

1. Programmable gain amplifier (56) comprising in combination: (a) first and second gain selection transistors (28b,28a), each having a control electrode and first and second current carrying electrodes, the control electrode of the first gain selection transistor (28b) being connected to a first gain selection conductor (42a), the control electrode of the second gain selection transistor (28a) being connected to a second gain selection conductor (42b); (b) a first output transistor (43) having a control electrode and first and second current carrying electrodes, the first current carrying electrodes of the first output transistor (43) and the first and second gain selection transistors (28b,28a) being coupled to a first current source (34), the second current carrying electrode of the first output transistor (43) being coupled to an output conductor of the programmable gain amplifier;; (c) first and second input transistors (21b,21a) each having a control electrode and first and second current carrying electrodes, the first current carrying electrodes of the first and second input transistors (21b,21a) being connected, respectively, to the second current carrying electrodes of the first and second gain selection transistors (28b,28a), the control electrodes of the first and second input transistors (21b,21a) being connected to an input terminal (7), the second current carrying electrodes of the first and second input transistors (21b,21a) being connected to a second current source (23); (d) a first gain resistor (25a) connected to the first current carrying electrode of the first input transistor (21b), and a second gain resistor (25b) connected to the first current carrying electrode of the second input transistor (21a);; (e) means for biasing (46a) the control electrode (50a) of the first output transistor (43), the gain of the amplifier (56) being selectable to a first value determined by the first gain resistor (25a) by turning on the first gain selection transistor (28b) in response to a signal on the first gain selection conductor (42a) and turning off the second gain selection transistor (28a), or to a second value determined by the second gain resistor (25b) by turning on the second gain selection transistor (28a) in response to a signal on the second gain selection conductor (42b) and turning off the first gain selection transistor (28b), wherein the biasing means includes a first inverting amplifier (46a) having an input coupled to the second current carrying electrodes of the first and second input transistors (21b,21a) and an output (50a) coupled to the control electrode of the first output transistor (43).