The invention relates to an instrumentation amplifier com¬ prising a reference stage containing a reference amplifier and a differential stage containing a differential amplifier, a first input of the reference amplifier being connected to a first input of the instrumentation amplifier, a first input of the differential amplifier being connected to a second input of the instrumentation amplifier, and an output of the differential amplifier being con- nected to an output of the instrumentation amplifier, and a second input of the differential amplifier being connected via a first im¬ pedance to an output of the reference amplifier and via a second impedance to the output of the differential amplifier. An instru¬ mentation amplifier of this type is described in the book entitled "The Art of Electronics" by P. Horowitz and W. Hill, New York, 1989, page 428.

An instrumentation amplifier is used to amplify a small dif¬ ferential mode voltage which is to be measured and is superimposed on a relatively large common mode voltage. A measuring situation of this type is encountered, for example, in biomedical applications. When bioelectrical signals are being measured in patients, for example, a potential difference of a few volts may exist between the mean potential of the patient and the zero ("common") potential of the amplifier, whereas the bioelectrical signal to be measured is less than 1 mV. In industry, too, numerous examples of a measur¬ ing situation of this type are to be found, such as temperature measurements with the aid of thermocouples or thermistors, load measurements with the aid of strain gauges, and measurements of movements and positions with the aid of transducers suitable for the purpose.

The following requirements, inter alia, are imposed on an instrumentation amplifier: a low amplifier noise, a high common mode input impedance, a high common mode rejection ratio, a large common mode input range and an accurately determined differential gain. The instrumentation amplifier disclosed in the publication mentioned can be constructed from a small number of components and has the favourable characteristics mentioned above provided rela¬ tively high differential gain is chosen.

Additional requirements are imposed on instrumentation ampli¬ fiers for biomedical applications: the differential gain must be large (for example, 1,000 - 10,000 for ECG measurements and 10,000 - 100,000 for EEG), the amplifier must not reach saturation as a consequence of the relatively large DC differences between the electrodes ("electrode offset"), and consequently the differential DC input range should be at least 300 mV. In order to prevent satu¬ ration of the instrumentation amplifier, it is expedient if the latter has a high-pass characteristic with a cut-off frequency at about 0.16 Hz.

It is furthermore important, in particular in biomedical applications, for the instrumentation amplifier to have small dimensions and consume little power.

In order to meet the above-mentioned requirements, the usual method is to use an instrumentation amplifier having a low gain factor as input stage and then to bring about a high gain by means of a number of capacitively coupled amplifier stages. However, this results in a relatively large number of components, including fair¬ ly large capacitors. As is known, capacitors having large capaci- tance values are relatively expensive, while they occupy a large amount of space on a printed circuit board.

The object of the invention is to eliminate the above-men¬ tioned disadvantages and to provide an instrumentation amplifier having a high gain factor, a good common mode rejection ratio, and a large differential DC input range, which instrumentation ampli¬ fier is constructed from a minimum number of components, the use of large capacitors being avoided. According to the invention, this is achieved in an instrumentation amplifier of the type mentioned at the outset in such a way that a first filter circuit is incorpor- ated between the second impedance and the output of the differen¬ tial amplifier, of which filter circuit an output is connected to the second impedance, a first input is connected via a third impe¬ dance to the output of the differential amplifier, and a second input is connected to the output of the reference amplifier. As a result of incorporating a filter circuit between the second impedance and the output of the differential amplifier, i.e. in the feedback network of the differential amplifier, the desired characteristic of the amplifier can readily be achieved. As a

result of constructing the filter circuit, for example, as an integrator, maximum feedback of direct voltages is achieved, while high frequencies undergo a relatively small degree of feedback. The desired high-pass characteristic is consequently obtained. However, it is also possible in principle to construct the filter circuit as a differentiator, as a result of which, for example, a low-pass characteristic is achieved. As a result of combining an integrator and a differentiator in, for example, the same filter circuit, a suitable frequency characteristic of the instrumentation amplifier can be obtained.

The provision of a filter circuit, such as an integrator, in the feedback network of the differential circuit has, in principle, a disadvantageous effect on the rejection ratio of the instrumenta¬ tion amplifier. As a result of connecting a second input of the filter circuit to the output of the reference amplifier, not only the output signal of the differential amplifier, but also that of the reference amplifier is fed to the filter circuit. In this way, the feedback involves, for example, the difference between the output signals of the reference amplifier and the differential amplifier in fι_.;ered form (integrated or differentiated). Signals appearing at the two inputs ("common mode signals") are amplified with less than unity gain in the instrumentation amplifier accord¬ ing to the invention and therefore relatively attenuated. This achieves the result that the common mode rejection ratio is not disadvantageously affected by the presence of the filter circuit and that the common rejection ratio is not very strongly dependent on the tolerances of the components.

It is pointed out that it is known per se, for example from "Medical & Biological Engineering & Computing", part 29 (1991), page 436, to incorporate an integrating network in a feedback net¬ work of an instrumentation amplifier. These known instrumentation amplifiers are, however, always provided with a so-called current- balanced input stage in which the integrator has no effect on the common mode rejection ratio. Now, however, according to the inven- tion an integrator is used in an instrumentation amplifier which can be constructed, for example, exclusively from operational amplifiers.

Preferably, the instrumentation amplifier according to the

invention is constructed in such a way that a fourth impedance is incorporated between the first input and the second input of the first filter circuit. As a result of a suitable choice of the value of said fourth impedance, the noise suppression, the differential gain and the bandwidth of the instrumentation amplifier can be adjusted to beneficial values.

The incorporation of a filter circuit in the form of an inte¬ grating network in the feedback network of the differential ampli¬ fier provides a very low gain and therefore a relative attenuation for offset voltages from the electrodes connected to the instrumen¬ tation amplifier. It is, however, possible for components of said integrator to introduce offset voltages in turn. According to the invention, said offset voltages are compensated for by constructing the instrumentation amplifier in such a way that a second filter circuit is incorporated between the third impedance and the output of the differential amplifier, of which filter circuit an output is connected to the third impedance, a first input is connected to the output of the differential amplifier and a second input is con¬ nected to a reference potential. The zero potential of the ampli- fier, for example the potential which is present at the midpoint between the positive and negative supply voltage, may be chosen as a reference potential.

If an integrator is provided in the first filter circuit, it may advantageously be constructed in such a way that the integrator comprises an operational amplifier, of which one input is connected to the output of the filter circuit, a non-inverting input is con¬ nected to the first input of the filter circuit, and an inverting input is connected via a fifth impedance to the second input of the filter circuit and via a first capacitor to the output of the fil- ter circuit. In this way, the desired integrating action of the filter circuit is achieved with a very small number of components. The filter circuit can readily be constructed as a differentiator by interchanging the first capacitor and the fifth impedance.

The second filter circuit can also be constructed as an integrator. The latter is advantageously constructed in the same way as the first integrator, i.e. the integrator comprises an oper¬ ational amplifier, of which an output is connected to the output of the filter circuit, a non-inverting input is connected to the first

input of the filter circuit, and an inverting input is connected via a sixth impedance to the second input of the filter circuit and via a second capacitor to the output of the filter circuit.

The second filter circuit can be constructed as a differen- tiator by interchanging the sixth impedance and the second capa¬ citor. It is, of course, possible to provide both an integrator and a differentiator in a filter circuit.

The instrumentation amplifier according to the invention may furthermore be constructed in such a way that the second input of the reference amplifier is connected via a seventh impedance to a reference potential and via an eighth impedance to the output of the reference amplifier. Here, again, the zero potential of the instrumentation amplifier may be used as reference potential.

With the instrumentation amplifier according to the inven- tion, it is possible to suppress common mode signals almost com¬ pletely, i.e. to achieve a common mode gain much less than unity. For this purpose, the instrumentation amplifier is constructed in such a way that the impedances comprise resistors whose values satisfy, as accurate as possible, the relationship:

R, .R.

R,.R. + R- .R ₃ + R,.R ₃

Although the reference amplifier and the differential ampli¬ fier may, in principle, be constructed from discrete components, an instrumentation amplifier containing a minimum number of components is achieved if it is constructed in such a way that the reference amplifier and/or the differential amplifier comprises an integrated operational amplifier. A larger differential gain and/or a large bandwidth can be achieved, with a suitable choice of the compo¬ nents, by constructing the instrumentation amplifier in such a way that the differential amplifier comprises a cascade circuit of at least two operational amplifiers. An instrumentation amplifier having extremely low input noise can be obtained if it is constructed in such a way that the refer¬ ence amplifier and/or the differential amplifier comprises a hybrid operational amplifier. In this case, suitable transistor pairs having beneficial noise properties are preferably connected up-

S EET

stream of the inputs of the reference amplifier and/or the dif¬ ferential amplifier.

The bandwidth of the instrumentation amplifier can be cor¬ rected and, if necessary, reduced by constructing it in such a way that the second and/or third impedance comprises a parallel circuit of a resistor and a capacitor. A reduction of the common mode rejection ratio can be prevented in this case by constructing the seventh impedance as a resistor with a series circuit of a resistor and a capacitor connected in parallel thereto. The instrumentation amplifier according to the invention can readily be extended for the purpose of connecting more than two electrodes by constructing it in such a way that the instrumenta¬ tion amplifier comprises at least two differential stages which are each connected to a respective input and output of the instrumenta- tion amplifier and to the output of the reference amplifier. In this case, each differential stage is preferably provided with at least one filter circuit, of which a first input is connected, directly or indirectly, to the output of the respective differen¬ tial amplifier and the second input is connected to the output of the reference amplifier, the beneficial characteristics of the instrumentation amplifier according to the invention thereby still being maintained for all the differential stages.

Advantageously, the instrumentation amplifier according to the invention can be accommodated, at least partly, in an inte- grated circuit. Since the capacitors used in the instrumentation amplifier according to the invention have low capacitance values, they can in principle also be incorporated in an integrated cir¬ cuit.

The invention will be explained in greater detail below by reference to the figures.

Figure 1 shows the circuit diagram of an instrumentation amplifier according to the prior art which is not suitable as it stands for biomedical applications.

Figure 2 shows the circuit diagram of a first embodiment of the instrumentation amplifier according to the invention.

Figure 3 shows the circuit diagram of a second embodiment of the instrumentation amplifier according to the invention.

Figure 4 shows the circuit diagram of a hybrid differential

amplifier which can be used as reference amplifier and/or as dif¬ ferential amplifier in the instrumentation amplifier according to the invention.

The instrumentation amplifier according to the prior art, shown in Figure 1 , comprises a reference amplifier A, and a dif¬ ferential amplifier A ₂ . With its feedback network comprising the resistors R ₇ and R _β , the reference amplifier A _n forms a reference stage. Likewise, with its feedback network constructed from the resistors R, and R ₂ , the differential amplifier A ₂ forms a differ- ential stage. A first input terminal E _η of the instrumentation amplifier is connected to an input terminal 11 of the reference amplifier A, , while a second input terminal E ₂ of the instrumenta¬ tion amplifier is connected to an input terminal 21 of the differential amplifier A ₂ . As shown in Figure 1, the output ter- minal 20 of the differential amplifier A ₂ is connected to an output terminal U, of the instrumentation amplifier. Although this known circuit has beneficial characteristics if a high differential gain is chosen, this circuit cannot readily be used for biomedical applications because the amplifier will generally become overloaded as a consequence of the relatively large differential DC signals produced by the electrodes.

The first embodiment, shown in Figure 2, of the instrumenta¬ tion amplifier according to the invention also comprises a refer¬ ence stage which is connected to the first input terminal E _η of the instrumentation amplifier and which is constructed from the refer¬ ence amplifier A, and the impedances R ₇ and R _B . However, the dif¬ ferential stage connected to the second input terminal E ₂ of the instrumentation amplifier comprises, according to the invention, not only the differential amplifier A ₂ and the impedances R _Λ and R ₂ , but also a filter circuit F, and an impedance R ₃ . In this case, an output 30 of the filter circuit F, is connected to the impedance R ₃ , which is connected in turn to an input terminal, in the case shown the input terminal 22 connected to the inverting input, of the differential amplifier A ₂ . The first input 31 of the filter circuit F, is connected via the impedance R ₃ to the output terminal 20 of the differential amplifier A ₂ . According to the invention, the second input 32 of the filter circuit F, is connected via a connecting line L to the output terminal 10 of the reference ampli-

fier A, . As a result, a common mode signal appearing at both input terminals E, and E ₂ is never amplified more than with unity gain. Consequently, the common mode rejection ratio has at least the same value as the differential gain. In practice, this characteristic has the great advantage that a very high common mode rejection ratio can be achieved without exacting requirements being imposed on the tolerance of the components used. The filter circuit F, preferably comprises an integrator, but it may also comprise a differentiator. The impedance R ₃ may be formed, for example, by a resistor, but possibly also by a short circuit.

Figure 3 shows a further embodiment of the instrumentation amplifier according to the invention. Just as in Figure 2, the instrumentation amplifier of Figure 3 comprises a reference stage which is constructed from the reference amplifier A_, and the impe- dances R _v and R _β . The first input terminal 11 of the reference amplifier A _η is also connected in this case to the first input terminal E, of the instrumentation amplifier, while the impedance R ₇ is connected via the connecting terminal 1 to a reference poten¬ tial, such as the zero potential of the amplifier. In contrast to Figure 2, the instrumentation amplifier of Figure 3 comprises two differential stages, each comprising a differential amplifier A ₂ (or A ₂ ', respectively), impedances R, , R ₂ and R ₃ (or R,', R ₂ ' and R ₃ ', respectively) a connecting line L (or L', respectively), and a first filter circuit F, (or FJ , respectively). The first input terminal 21 and the output terminal 20 of the differential ampli¬ fier A ₂ are connected, respectively, to the second input terminal E ₂ and the first output terminal U, of the instrumentation amplifier. In a completely analogous manner, the first input ter¬ minal 21' and the output terminal 20' of the differential amplifier A ₂ ' are connected, respectively, to a third input terminal E ₃ and a second output terminal U ₂ of the instrumentation amplifier.

Each differential stage is furthermore provided with a second filter circuit F ₂ , or a completely analogously connected second filter circuit F ₂ ', respectively. The filter circuit F ₂ is incor- porated between the impedance R ₃ and the output terminal 20 of the differential amplifier A ₂ in the feedback network of said differen¬ tial amplifier, the output 40 of the filter circuit F ₂ being con¬ nected to the impedance R ₃ , the first input 41 being connected to

the output terminal 20 of the differential amplifier A ₂ , and the second input 42 being connected via a connecting terminal 2 to a reference potential, for example the zero potential of the instru¬ mentation amplifier. In the embodiment shown, both the filter circuit F, and the filter circuit F ₂ (or F_, ' and F ₂ ', respectively) comprise an inte¬ grator circuit. The second integrator circuit of the filter circuit F ₂ increases the range of the instrumentation amplifier. In the embodiment shown in Figure 3, the filter circuit F, comprises an amplifier A ₃ , a first capacitor C, and a fifth impedance R _a . The integrator circuit of the filter circuit F ₂ likewise comprises an amplifier A _Λ , a second capacitor C ₂ and a sixth impedance R ₆ . The impedances R _s and R ₆ are preferably formed by resistors, while operational amplifiers are preferably used for the amplifiers A ₃ and A _Λ . The amplifier A ₃ ensures a low total amplification of off¬ set voltages of, for example, the electrodes used in the instrumen¬ tation amplifier, while the amplifier A ₄ compensates for the offset voltage of the amplifier A ₃ . With a suitable dimensioning of the impedances R, to R ₄ inclusive and a good choice of the operational amplifiers A ₃ and A ₄ (a low current noise is essential in this case, operational amplifiers having an FET input being very satis¬ factory), relatively large values (for example, 10 MΩ) can be chosen for the resistors R _a and R ₆ without the noise of the filter circuits F, and F ₂ contributing significantly to the total noise of the instrumentation amplifier. The use of large resistance values for R _s and R _s has the advantage that (for a given value of the high pass cut-off frequency) small capacitances can be used for C, and C ₂ .

The impedances R ₂ and R ₃ may each advantageously be formed by a parallel circuit of a resistor and a capacitor. This reduces the bandwidth of the instrumentation amplifier in a simple way, as a result of which a desired frequency characteristic can be obtained. If this effect is not intended, the impedances R, to R _β inclusive may be formed by resistors. The best way of limiting the bandwidth of the instrumentation amplifier is to reduce the bandwidth of the operational amplifier A ₂ itself. In the case of most operational amplifiers, one or more of the following possibilities can be used: a) use of a suitable external compensation capacitor in the case of

uncompensated operational amplifiers, b) capacitive feedback from the output to one of the offset connections, c) choice of a suit¬ able adjustment current in the case of programmable operational amplifiers. If a high common mode rejection ratio is needed in the case of a relatively low differential gain, the rejection ratio of the instrumentation amplifier can be appreciably improved by construct¬ ing the seventh resistor R ₇ as a resistor with a series circuit of a resistor and a capacitor connected in parallel with it. In this way, it is possible to compensate for the bandwidth, which is not infinitely large, of the reference amplifier A, , which is con¬ structed, for example, as an operational amplifier.

By choosing suitable ratios of the values of the impedances, in particular of the resistors, the result can be achieved that the noise produced by the integrator does not contribute substantially to the total amplifier noise, while the instrumentation amplifier has a large input range for differential DC signals at the same time. With suitable dimensioning, the total noise level of the instrumentation amplifier is determined solely by the noise of the active components used and forming the amplifier inputs, such as an operational amplifier, a bipolar transistor or a field-effect tran¬ sistor (FET) .

If the resistance values of the impedances are chosen to satisfy the relationship

R _β R, . ^

R ₇ R ₂ .R~ R ₂ .R ₃ + R. • R 3

as accurate as possible the common mode gain of the instrumentation amplifier is nearly zero. This means that common signals appearing at the various inputs E, will not be passed on by the instrumenta¬ tion amplifier. A further simplification of the instrumentation amplifier is obtained by omitting the impedance R ₇ and replacing the impedance R _β by a short circuit, as a result of which the sec¬ ond input terminal 12 of the reference amplifier A, is connected directly to its output terminal 10 and not to the terminal 1. It has been found that this simplification does not produce a signifi¬ cant reduction in the rejection ratio provided the differential

gain is high. The latter is the case in virtually all biomedical applications of the instrumentation amplifier.

The instrumentation amplifier shown in Figure 3 comprises two differential stages. It will be clear, however, that the instrumen- tation amplifier can readily be extended with a plurality of diffe¬ rential stages which are each connected to a further input E, and an output U _A . Furthermore, the impedance R, and the line L of each differential stage can always be connected to the connecting ter¬ minal 3, these thereby always being connected to the output 10 of the reference amplifier A, . Interference is prevented because the common mode input impedance is the same for all the inputs E _i .

The output signal of the reference amplifier A _n can be used, however, not only as input signal (reference signal) for one or more differential stages, but it can also be used to drive other circuits, such as a "driven-right-leg" circuit and/or a "guarding" circuit. The addition of such circuits will considerably reduce the susceptibility to interference.

The bandwidth and/or differential gain of the instrumentation amplifier according to the invention can be increased by construct- ing the differential amplifier A ₂ as two operational amplifiers connected in cascade. In order to reduce the amplifier noise still further, the reference amplifier A, and the differential amplifier A ₂ can be constructed from an operational amplifier preceded by an input stage constructed from discrete transistors. Preferably, suitable transistor pairs having beneficial noise characteristics are used for this purpose.

Figure 4 shows an amplifier circuit of this type. Although the reference amplifier A _Λ and the differential amplifier A ₂ can each be constructed from a single integrated operational amplifier, the use of the circuit in Figure 4 provides considerably improved noise characteristics. The circuit in Figure 4 comprises an operational amplifier A ₁₀ , which is connected to an output 100 of the circuit. Provided between the inputs 101 and 102 and the oper¬ ational amplifier A ₁₀ is a differential amplifier circuit compris- ing the transistors T, and T ₂ . A current mirror constructed from the transistors T ₃ and T ₄ supplies a constant current through the balance stage. The desired adjustments of this so-called "long- tailed pair" configuration are achieved using the resistors R ₁₀ ,

R, and R ₁₂ . The positive and negative supply voltages are con¬ nected, respectively, to the connections 103 and 104. The oper¬ ational amplifier A ₁₀ is, for example, of the type TLC271, while the transistors T,, T ₂ , T ₃ and T _Λ may be of the type LM394. A suit- able adjustment with supply voltages of +5 V and -5 V is obtained if the resistance values of the resistors R ₁₀ , R,, and R ₁₂ are, respectively, 100 kΩ, 100 kΩ and 270 kΩ.

The simplest way of establishing a particular, desired band¬ width of the instrumentation amplifier is to provide a suitable combination of, on the one hand, the type of operational amplifier A ₂ and, on the other hand, the differential gain. If a compensated operational amplifier is used, the required gain-bandwidth product of A ₂ is given by the product of the desired bandwidth of the in¬ strumentation amplifier and the desired gain for differential sig- nals. It should be pointed out that, in many cases, an uncompensat- ed operational amplifier can be used for A ₂ . The circuit is stable with this type of operational amplifier provided the chosen gain factor for differential signals is high because, in that case, the ratio between the open-loop gain and the closed-loop gain of oper- ational amplifier A ₂ is relatively low. Compared with compensated operational amplifiers, an appreciable increase in the bandwidth can be achieved with uncompensated operational amplifiers for the same power consumption. Note that the low-pass filter circuits which are standard in other designs can usually be dispensed with in the circuit according to the invention because, with suitable choice of the components, the low-pass characteristic of the oper¬ ational amplifier A ₂ is utilised.

In the circuit in Figure 3, the reference amplifier A, may, for example, be of the type LT1012. Depending on the desired band- width, the differential amplifier A ₂ may, for example, be of the type LT1012 (compensated) or of the type LT1008 (uncompensated) . A good choice for the amplifiers A ₃ and A ₄ is the dual operational amplifier TLC1078. Suitable resistance and capacitance values for a differential gain of 20,000 and a bandwidth of 0.2 - 40 Hz if the LT1012 is used, or of 0.2 - 2,000 Hz if the LT1008 is used, are as follows:

In this configuration, the impedance R ₇ can be omitted, while the impedance R _β has been replaced by a short circuit. It is obvi¬ ous, however, that other resistance and capacitance values can also be used, depending on the optimisation of a particular characteris¬ tic of the instrumentation amplifier. An embodiment of this type furthermore has the advantage that the stated specifications are met with a very small number of com¬ ponents. The small number of components becomes particularly evi¬ dent if the instrumentation amplifier according to the invention is compared with two designs described in the specialist literature having comparable specifications, namely the "three-op-amp ampli¬ fier" as described in the chapter on "Biomedical amplifiers" by M.R. Neuman, in the book entitled "Medical Instrumentation, Appli¬ cation and Design", pages 307-309, Boston (1978), ed. J.G. Webster, and the "current balance amplifier" as described in the article entitled "High quality recording of bioelectric events. II: a low- noise, low-power multichannel amplifier design", Medical & Biologi¬ cal Engineering & Computing, Vol. 29, pages 433-440 (1991) by A.C. Metting van Rijn, A. Peper and CA. Grimbergen. The differences are evident from the following table: Circuit 3-op-amp Current- according amplifier balance to the amplifier invention

The instrumentation amplifier according to the invention does not impose any exacting requirements on the accuracy of the values of the components and makes fine adjustment (trimming) superfluous. It is possible to supply the instrumentation amplifier with (small) batteries, while the low capacitor values and the relatively small number of components makes a very compact construction possible.

The invention therefore provides an instrumentation amplifier which can be used for a plurality of channels, which has excellent characteristics and which can be constructed from a very small number of components.