GB2424330A - A transimpedance amplifier with a shielded feedback resistor - Google Patents

Abstract

A transimpedance amplifier for very low currents comprises a high value feedback resistor (1 TOhm) located inside a cylindrical screen. The screen is driven at a potential VH divided from the output voltage VG. The resistor and screen assembly (figure 4) is modelled as a distributed RC transmission line 314. The screen circuit allows the capacitance effects of the feedback resistor to be more accurately defined and limited. The capacitance between the screen and resistor is effectively nulled by driving the screen at one half of the output voltage so that the overall feedback capacitance comprises C30 and C20 in series. Low feedback capacitance permits faster amplifier response. The measurement of very small currents is required in electrometers, vacuum ion gauges, mass spectrometers, optical spectrometers, chromatographs and semiconductor characterization apparatus.

Description

Improved Transimpedance Amplifier for Low Currents The present invention

relates to transimpedance amplifiers, particularly to transimpedance amplifiers for measuring small current signals having bandwidths of a few hundreds of Hertz to milliHertz.

Relatively simple apparatus such as electrometers, vacuum ion gauges, and more complex systems such as mass spectrometers, optical spectrometry, semiconductor characterization and chromatography apparatus require accurate measurement of very small currents typically in the order of less than one nanoampere (l0 ampere), and more specifically between one picoampere (1x10'2 Ampere) to one femtoampere (1x10'5 Ampere). The mention of these specific amounts is provided by way of example, and should not be considered as limiting the scope of this application.

As the source of these small currents is usually of high impedance, the commonly used and preferred measurement circuitry is the transimpedance amplifier.

Referring to Figure 1, a known circuit configuration of a trans impedance amplifier comprises a current source 102, an operational amplifier 104 having inverting and non-inverting inputs 106 and 108, respectively, and an output 111 having a resistor Ri connected in series with R2 and capacitor Cl, forming a sampling network 118, a feedback network 114 having a resistor R3 (113) connected in parallel with a capacitor C3 (119) and a summing node 117. The non-inverting input 108 of the operational amplifier 104 is connected to common 116. The output 110 of the first operational amplifier is connected to the non- inverting input 109 of a second operational amplifier A2 (105), and the inverting input of said operational amplifier A2 is connected via a resistor R5 to common and to its output 111 by resistor R6.

This second operational amplifier A2 is provided to give an additional loop-gain to the signal from output 110.

A signal tap 112 is provided between a resistor Ri and resistor R2, and this arrangement, together with Cl, constitutes the sampling network 118. The feedback network 114 leads from the sampling network to the inverting input 106 of the operational amplifier 104, that is between the output 111 and the input 106. C2 represents the capacitance of the signal source, any connecting conductors and of the operational amplifier Al.

The voltage at the inverting input 106 is very nearly the same potential as it is at the non-inverting input 108. As the non-inverting input 108 is connected to common 116, the inverting input 106 is said to be at virtual common. Transimpedance amplifiers are commonly used as current measuring instruments since they are essentially an ammeter with substantially zero voltage across the meter. In order to detect such small signals it is necessary for the feedback resistor R3 to be large, such as, for example between 100 Gohm (10"ohm) and 100 T (l0'4ohm). The need to use resistors of such high value, in combination with incidental and stray capacitance, results in bandwidths of approximately one Hertz or less, typically 100mHz to 0. lmFIz. Consequently, this results in relatively slow response time to changes in signal.

In an effort to overcome these difficulties, applicant herefor has filed a UK Patent Application GB2393 865 (on which the abovementioned introduction is based).

In circumstances where the transimpedance amplifier is used in applications such as the elution of fractions from a chromatograph or the output signals from a static vacuum mass spectrometer, the changes in signals from such apparatus are not controllable in terms of time response and therefore the measured signal can be substantially misleading and unreliable. For example, in relation to the elution of fractions from a chromatograph, the peaks of the signal to be measured will have passed before the amplifier can properly respond to them, thereby giving erroneous readings. In another example, a mass spectrometer may be used to study isotopic ratios in noble gas such as, for example, Argon from geological or astrophysical materials. The relatively small samples of noble gas available in combination with the need to measure accurately the relatively large ratios between the isotopes, which also decay in intensity over time, means that a short response time and high sensitivity are necessary features of the measuring apparatus.

While the circuit configuration of UK Patent Application GB2393865 provides a fair representation of actual circuits, experience has shown that the feedback network consisting of only one capacitor C3 and one resistor R3 is, in terms of the response times and resistance values in use, too simplistic an approximation of the feedback circuit which should more accurately be represented by a known resistance- capacitance transmission line. An example of the modified feedback circuit including this feature is shown generally at 120 in Figure 2 provided herewith. This circuit includes additional modifications as compared with the simpler circuit of Figure 1.

Conventional circuit analysis of this modified circuit results in complex circuit equations which are not easily soluble, and which would lead to output response functions in which the various terms are not readily associated or capable of being easily correlated to the dynamic performance characteristics of the circuit. Applicant herefor has therefore analysed such a circuit using the commonly available SPICE suite of programs. Such analysis allows the prediction of the response for given circuit component values but does not allow parametrization of the responses.

However, analysing the circuit response using various different values of the capacitances and resistances in the transmission line part of the circuit does provide guidance on what values should be used for optimizing the response characteristics of the modified circuit.

In Figure 2, the transmission line portion of the circuit marked generally at 120 is divided into ten portions comprising resistances (R20- R29), said resistances having associated distributed capacitances (C2 1 - C29). Through experimentation, applicant herefor has found that the division of circuit portion 120 into this many segments works well as far as being a reasonable approximation of actual conditions, and while more or fewer segments may be used with differing values of resistance and capacitance, the overall results are generally equivalent and errors associated with this approximation are small and within the tolerances of the poorly known values of the capacitances in practice.

Further experimentation and analysis of the above circuit has shown that the distributed capacitance of the feedback resistance is very variable, depending on the external environment in which the system is placed. A further factor which can materially affect the results is the geometrical arrangement of the various components constituting the circuit. For transimpedance amplifiers to have good time response characteristics, the feedback capacitances required are very small (1O14 Farad) and at this level almost every conductor in the circuit can have an influence thereon.

Some feedback capacitance, corresponding to C3 in Figure 1, is required to ensure stability against oscillation, and the value of such is in part dependent on the value of C2 which represents the gross input capacity of the system. In existing realizations of transimpedance amplifiers C3 (119) is provided by a low value discrete physical component and its effective circuit value is often adjusted by means of feedback, as described in United States patent 4,069,459. This circuit described therein is however unsatisfactory as regards noise, and the use of a physical capacitor component introduces the possibility of leakage through its dielectric or across its surface thus compromising feedback resistance.

It is an object of the present invention to provide a transimpedance amplifier capable of measuring small currents and having a relatively short response time, and which is not as susceptible to noise and ambient stray capacitance of the system as a whole, and which furthermore allows the capacitance effects of the feedback resistance to be more accurately defined and limited.

It is a further object of the invention to provide a circuit technique for reducing the effect of distributed capacitance of a high value feedback resistance in a transimpedance amplifier while simultaneously compensating the system against oscillation.

It is a further object of the invention to provide a technique for removing the effect of the capacitative environment of the feedback resistor to provide an improved time response.

It is a further object of the invention to provide a technique for screening the feedback resistance from external interference without compromising bandwidth and time response.

It is a further object of the invention to provide a technique for providing the feedback capacitance necessary for stability without significant leakage resistance.

According to the present invention there is provided a transimpedance amplifier comprising a current source and an associated overall input capacitance, primary amplification means having an input to which said current source is connected and an output, said output being optionally connected to secondary amplification means having one or more inputs and an output, a first resistor, a second resistor and a first capacitor connected in series between the output of the primary or secondary amplification means if provided and a common voltage rail, a sampling network consisting of a signal tap provided between said first resistor and said second resistor which may optionally be connected to an input of tertiary amplification means having one or more inputs and an output, to deliver a feedback signal, a feedback network being both resistive and capacitive and which is connected between the sampling network or the output of said tertiary amplification means and the input of the primary amplification means at a summing node operable to sum the signal from the feedback network with an input signal from the current source, characterised in that the feedback network comprises at least one resistor at least partially disposed within a surrounding electrically conducting screen and which is charged with a screen drive signal which is a function of the feedback signal.

Preferably the screen completely surrounds said resistor, further preferably is substantially concentric therewith, and yet further preferably, said resistor is disposed completely within said screen in that the extremities of said resistor are within the ends of said screen.

Further, for convenience of computation, the screen is circular in crosssectional shape.

It is preferable that tertiary amplification means is connected to the voltage tap, said tertiary amplification means being in the form of a unity-gain buffer having an output which delivers the feedback signal to the capacitive feedback network.

It is most preferable that the current source is connected to the input of primary amplification means in the form of an input operational amplifier, the output of which is then connected to the input of secondary amplification means in the form of a loop-gain amplifier, the output of which is connected to the series connected first resistor, second resistor and first capacitor and second capacitor.

The provision of a conducting screen which is driven by a signal which is proportional to the feedback signal delivered to the feedback resistor which is concentrically disposed within said screen has a great number of advantages, and these will become apparent to the reader by virtue of the following specific embodiment of the invention, which is provided by way of example with reference to the following Figures:

Fig. 1 shows a circuit of prior art configuration,

Fig. 2 shows a common circuit representation of an RC transmission line, Fig. 3 shows a circuit diagram representing a transimpedance amplifier according to the invention, Fig. 4 shows a schematic circuit diagram of the feedback network according to the present invention, Fig. 5 shows a schematic end elevation of a cylindrically screened resistor used in the invention, Fig. 6 shows a schematic plan view representation of a printed circuit board (PCB) on which a transimpedance amplifier according to the invention may be provided, Fig. 7 shows a plot of electric potential contours which surround the resistor to which is applied a potential of lV and wherein the cylindrical screen surrounding the resistor is at OV potential, Fig. 8 shows a plot of electric field vectors surrounding the end of the resistor having a 1V potential applied thereto, with the screen surrounding said resistance at OV, Fig. 9 shows frequency response graphs for a transimpedance amplifier according to the invention having a feedback resistance of lT and distributed capacitance C20 - C30 (see Fig. 3) each of 60ff (60 x 1 015F), Fig. 10 shows the time response graphs corresponding to the frequency response graphs of Fig. 9, Fig. 11 shows the measured response for a I OTohm amplifier, The figures include a representation of a particular form of feedback resistor as a matter of illustration, but the system may be adjusted to suit other forms as required.

Referring firstly to Fig. 3, there is shown generally at 300 a circuit diagram representative of a transimpedance amplifier according to the present invention. This circuit consists initially of a current source 302 to be measured and which is fed to the inverting input 306 of an input operational amplifier Al, further identified in the Figure as 304. The non-inverting input 308 of operational amplifier Al, together with the one side of the current source, is coimected to a common rail voltage 316, which in this embodiment is at common potential. The output of operational amplifier Al is fed to the non-inverting input 309 of a secondary operational amplifier A2, designated also in the Figure at 305. A feedback signal from the output of A2 is fed back to the inverting input 307 of A2 through a resistor R6, and said non-inverting input 307 is further connected to common 316 through a further resistor R5.

The output of operational amplifier A2 is then fed to a resistor Ri, a further resistor R2, and a capacitance Cl connected in series, and in accordance with the invention specified in the above mentioned earlier filed patent application number GB 2393865. A voltage tap 312 is taken from a summing network 318 provided between the resistor Rl and the resistor R2. The output of operational amplifier A2 also serves to provide a signal output 311 which is the output signal for the whole of the transimpedance amplifier represented in Fig. 3. In contrast to the summing network of the prior published specification mentioned above, a further capacitance C4 is included in the summing network 318 and together the components Ri, R2, Cl, C4 act together as a high impedance phase shift network, the output of which is then fed to a unity gain buffer comprising a third operational amplifier A3 (323). The voltage tap 312 taken from the summing network 318 is fed to the non-inverting input of A3, and the inverting input is fed from a direct feedback from the output of A3.

The signal at 312 is then fed to the distributed resistance-capacitance transmission line circuit designated generally at 314, and consisting of ten distributed resistances R20 through R29, and nine parallel connected capacitances C2 1 through C29, and two further capacitances C20 and C30. This arrangement is representative of the actual ambient characteristics of a simple feedback resistor (represented by R20 through R29) placed inside an electrically conducting cylindrical screen, between which there will exist a capacitance (represented by C21 through C29). Additional capacitances C20 (322) and C30 (321) represent the capacitance between the feedback resistor leads and terminals and the cylindrical screen, and a further capacitance CS (301) represents the capacitance between the screen and the common rail, marked at 316 in the drawings. The impedance of the feedback network is very high so that it does not load the sampling network Ri, R2, Cl, C4 and connecting it at 312 avoids any offsets arising from A3. The screen drive requires a low impedance source and is therefore connected via the output (324) of A3 which drives R7 and R8.

The use of a concentric cylindrical screen as shown in Fig. 5 at 603 to concentrically surround a resistor element 601 allows for approximate calculation of the expected capacitances, notwithstanding deviations from these calculations resulting from end effects. The concentricity of screen and resistor element is clearly shown in Fig. 5, and designated generally by reference number 600. The resistor lead is shown at 604, whereas in the adjacent diagram, an end elevation is shown of an example of a closed end of a screen in which a resistor is clamped. In this diagram, 610 refers to clamping screws, 612 is a longitudinal split in the screen, 614 shows a hole for the resistor, and 616 represents the closed end of the screen.

Referring again to Fig. 3, an important element of the present invention is the fact that the screen as referenced at 303 is actively driven by a signal taken from between a pair of series connected resistors R7, R8 which are connected between the output 324 of the unity gain buffer A3 and the common rail 316. The active driving of the electrically conducting screen by this signal has a number of advantageous effects as hereinafter described.

In general, the driving of the electrically conducting screen by the signal VH (313) taken from between R7 and R8 may be termed screen drive attenuation, and, as described below, R7 will be chosen to equal R8 as a starting point. The relatively low impedance source of R7 in parallel with R8 effectively eliminates the influence of the external capacitive environment, represented by C5. If required, a further unity gain buffer may be inserted into the circuit but the effect is negligible for the low value load capacitors involved.

As an example calculation: The capacity of two coaxial cylinders of inner radius a and outer radius b is given by (per unit length i.e. im) (Figure 5): c - 2,rü rb ml - [a where c is the dielectric constant, o is the permittivity of free space 8.854x 10_12 F/rn.

For a glass-enclosed type resistor, (in which the actual resistor is rather smaller than the outer glass enclosure) the dimensions a = 1mm, length = 18mm, give a total capacity of 0.6pF for b = 5mm. For the resistor wires (diameter say 0.5mm) the capacity is 1 5.5pF/m for straight axial position. It is important to realise that a lower effective feedback capacity provides a faster response. The range of capacity available is not great since the relationship is logarithmic. The air-insulated capacitor (which is what placing the resistance inside the electric screen actually provides) now eliminates any possibility of leakage compared with say the polystyrene capacitors previously used in the feedback network.

There is a small capacitive effect from the glass dielectric, but since this fills only a small part of the interelectrode space its contribution may be ignored so E=l. For resistors with a ceramic former for the resistive element, the dielectric is internal to the cylindrical film resistance and so plays no part in the capacity between screen and resistor. At the very lowest current levels the screens will also provide some reduction in ionization currents generated by say cosmic rays since it restricts the volume of ion collection.

Further analysis and/or simulation of the circuit shown in Figure 3 shows important relationships between the capacitor (C20-C30) currents (i20 through i30) associated with the feedback resistor: 121 = 129, 22 = 128, 123127, 124 26, ]9 30 = 0, i16 = 0, and necessarily ij = i19 + i20 Also i20 = i18, i19 = i30 which would be expected at zero frequency but they are also equal during the transient interval.

These equalities are due to the choice of division into equal parts for the resistor and capacitance. The currents between screen and resistor in the capacitors C21 to C29 cancel out and hence these capacitances may to first order be disregarded. A similar effect will hold for the algebraic sum in the case of unequal division of the resistor (R20 - R29). The current i25 in C25 is zero so the resistor may be supported here, at the centre of the resistance, with negligible effect, and the effective feedback is through C30 and C20 in series. If these two capacitances are equal then their junction, i.e. the cylindrical screen 303, will be at half the potential of the feedback voltage at point G since the summing junction 317 is at zero potential. This is the reason for choosing the starting values of R7=R8. These may be replaced, in part or in full, by a potentiometer to provide control of the effective feedback capacity.

If C30 is not equal to C20 then the additional current i16 flows to make up the required feedback. Thus C30 does not have to be defined by C20 and the screen can be extended to the feedback end of the overall resistor body to provide extra screening of the feedback resistor from the large voltages at point B (31 1), F(324) or G (310). To keep the effective value of C20 to the low value required i.e. to limit feedback from the end of the cylindrical screen to the conductors associated with the input circuitry, an additional screen (404, see Figure 6), connected to common potential 316, is positioned as shown in Figure 6 and preferably encloses the input circuitry as well. Though the cylindrical screen 403 serves to isolate the feedback resistor from external influences it will in itself couple to adjacent circuitry, as for example a neighbouring amplifier, resulting in crosstalk. To eliminate this the cylindrical screen may be enclosed in a surrounding screen connected to common (316) i.e. an extension of screen 404, and this will also further reduce coupling to its own amplifier input.

A suitable hole in the screen at location W allows passage of the resistor body without contact with the screen so as to avoid leakage currents. The actual resistor element inside the glass body is typically placed so that its end at Q is about flush with the end of the cylinder. There is also then little capacity between the cylinder screen 403 and the resistor lead wire at P. The feedback end of the cylindrical screen 303 (and 403 in Fig. 6) at S can be closed (except for a hole to accommodate the glass body) since it will only be in the vicinity of the lead wire and the capacity to it is inconsequential as it coimects to the low impedance amplifier output 313. Thus by longitudinally splitting the cylinder as shown in Figure 5, end cap 412 may be used to clamp the glass body of the resistor to provide mechanical restriction of any possible vibration and hence noise from capacitive modulation, and to enhance thermal conduction to the resistor when the system is cooled or thermally stabilized. The presence of the common screen 404 and the end clamp 412 of the cylindrical screen does mean that the various capacitor values are slightly changed from those computed on the basis of concentric conductors. For the type of resistor not enclosed in a glass body the clamp support may alternatively be placed at the centre of the resistor since as indicated above the capacitive current is effectively zero there. Other forms of screen may be used to achieve the same objectives.

It is most important from the point of view of fast response that the effective feedback capacitance is as low as possible for short time response, but as high as necessary for stability of the circuit as a whole.

Although Fig. 5 shows a surrounding screen of circular cross-section other geometrical shapes can be used, but the capacitance of such arrangements is less readily computed. Since the feedback capacity now has only air (or a moderate vacuum) dielectric there is no leakage path. Compensation of a distributed circuit such as shown in Fig. 3 is more closely possible with distributed components.

In Fig. 6, there is shown a possible layout of feedback resistor and screen on a printed circuit board designated generally at 400. In this Figure, there is shown a signal input 411 which is fed through a protection resistor 406 to the input 407 of an operational amplifier 405, and simultaneously to a feedback resistor element 401 encased with a glass enclosure, both of which are concentrically disposed within an electrically conducting screen 403. The screen 403 is fed from a screen drive 408, and a terminal 409 allows for connection of the feedback drive signal for the resistor. A screen clamp or end cap 412 may also be provided to ensure that the resistor element 401 is substantially or exactly concentrically disposed within the screen 403.

As mentioned above and, there is shown in dotted line notation a secondary screen which may be connected to the common rail potential and which has the further advantageous effect of allowing for further adjustment of the overall feedback capacitance of the arrangement. The field effects of the common screen may be seen in Fig. 7, and additionally in Fig. 8.

Specifically, Fig. 7 shows a contour plot of electric potential for one volt on the feedback resistor and zero volts on the other screens, and indicates the containment of the field. In this figure, a ZX section is shown, and reference numerals 700 -712 represent respectively the common screen, hole edges, the cylindrical screen, the wire at its IC input end, the resistor, the closed end of the cylindrical screen, and the cylindrical screen itself.

Fig. 8 shows the electric field in the vicinity of the input where it can be seen that there is coupling between the resistor and both of the screens, namely the cylindrical screen surrounding the resistor (712), and the common screen provided at one end of the cylindrical screen. Moving the relative axial position of the resistor is an alternative means of adjusting the overall feedback capacitance, and therefore a useful way of making capacitance adjustments as required. The computed charge distribution on the wires and endcaps feeding the resistor allows calculation of the capacitance between each of said wires and endcaps and the ambient surroundings, or the cylindrical or common screens. From these simulations it is found that the closed end identified at S on Fig. 6 has very minor effects on overall capacitance, but that the proximity of the common screen marked as 404 in Fig. 6 to the conducting screen 403 has a significant effect: As an example for the particular dimensions used for the simulation, moving the common screen from 2mm separation to 6mm separation changes the capacitance between the common screen and the cylinder from about 140ff to 90ff, and the capacitance between the cylindrical screen and the resistor element from about 660ff to 700ff. For practical measurements, a distance of approximately 2mm has been used by the applicant, and found to give reasonable results. The effect of the capacitance between the cylindrical screen and the common screen (identified as Cc on Fig. 4) is minimal since it is driven from a relatively low impedance source. This capacity will also include that to any commoned screen surrounding the cylindrical screen 403.

Figure 4 shows the schematic layout for simulation and the corresponding capacities, and more specifically shows an example configuration for calculation of the overall capacities and for closer examination of the input end which allows for discrimination between the actual resistor element and the endcap terminals and lead wire. In this Figure, 350 represents a common electrode, 352, 354 represents the lead wire at its feedback end and IC amplifier end respectively, 356 represents the glass enclosure within into and from which the lead wire passes, 358 represents the resistor element, references 360 represent end caps, 362 represents the clamp end of the cylindrical electrode 364, and 366 represents a common ground. To define the capacities of this figure: C1 = inputcapacity of system CA = capacity of end cap and lead wire to common screen CB = capacity of end cap and lead wire to cylinder screen C = capacity of cylinder screen to rest of the world CD = distributed capacity of cylinder to resistor element CE = capacity of end cap and lead wire to cylinder screen The capacities involved are small (tens to hundreds of femtofarad) and such fine discrimination is necessary to determine the critical capacity C8 equivalent to C20 in Figure 3. The capacity CA is generally small relative to C1 and is effectively subsumed by it.

The radial position of the resistor inside the cylindrical screen is furthermore not critical, since for this coaxial configuration, the capacitance is a logarithmic function of the radii of the cylindrical screen and the resistor element, and thus deviations from absolute coaxiality have small effects. It is to be mentioned at this stage, that there is an advantage in using resistor elements with a small radius since for a given total capacitance, the radius of the cylindrical screen can be smaller.

Simulation of a circuit with a feedback resistance of 1 Tohm (i.e. the ten resistors R20-R29 of Fig. 3 each being of 100 Gohm) and capacitor values C20 to C30 each being of 60ff gave the frequency response curves shown in Fig. 9 and the corresponding time response curves in Fig. 10, which shows settling to within 1 Oppm in approximately 200ms for a 5V step. It is seen from these figures that the bandwidth is one or two orders of magnitude greater than would be found for the conventional circuit (preceding the prior published patent specification above- mentioned) and that therefore the rise time will be correspondingly shorter. The response of the phase shift network Ri, R2, Cl, C4 produces a minimum feedback transmission at the frequency corresponding to the - 3dB point of the overall response and the effect is to improve the linearity of the overall phase lag as a function of frequency.

This form of variation indicates a more constant group delay which in turn results in lower pulse distortion. The addition of C4 to the corresponding phase shift network specified in the prior art patent makes a significant improvement in phase correction and reduces the contribution of noise gain to the total noise output. The effects are illustrated in Fig. 9. The "standard transimpedance amplifier" response shown in Fig. 9 refers to amplifiers preceding the prior art of patent GB 2393 865A.

The response time of the system is adjusted by variation of the time constants of the phase control network Ri, R2, Cl, C4. The time constant R1CI is the primary control while R2 acts as a damping control to limit oscillatory ringing and C4 reduces phase shift at high frequency, but the various factors do interact. Since the magnitude of the input capacitance C2 varies with application, it is necessary to make the various time constants adjustable to enable optimum operation.

The standard measure of rise time (10% to 90%) is indicative but in the more common applications it is the settling time to within some tolerance that is significant, and settling time is controlled by minor variations of the parameters.

Simulation gives an initial indication of appropriate values but these may have to change significantly for an actual circuit since the various incidental capacities are only approximately known. With the layout outlined in the Figures, suitable starting values may be listed for guidance. Initially RI is adjusted to obtain the best general rise time as seen on an oscilloscope with R2 about one quarter of Ri. If there is an overshoot, Ri is decreased and vice versa. Any oscillatory ringing can be reduced by variation of R2 or adjustment of C4. An undershoot following the initial overshoot may be reduced by increase of R2. Variable capacitors of the values typically required are not available, so fine adjustment must be done using rheostats for Ri and R2, with fixed Cl and C4. Stable plastic film capacitors should be used to minimise drift in the compensation. If greater flexibility in compensation is required it is possible to drive the feedback resistor and the capacitive screen from separate networks equivalent to the phase control network Ri, R2, Cl, C4 and buffer A3..

Fixed high stability resistors, together with minimum value rheostats should be used for the same reasons. The transient response can be adjusted to be effectively deadbeat as illustrated in Fig. 10. Measurements on actual circuits show 10% to 90% risetimes of about 50msec for iT, 500msec for lOT and 5sec for lOOT amplifiers. As an example Figure 11 shows the measured response of a 1 OT amplifier. The present compensation technique extends effectively to lower value resistors down to at least lGohm where phase shifts arising from A2 (see Fig. 3) begin to effect the overall response, thus covering a gain range of at least 1 If the gain-bandwidth of A2 is increased then even lower value resistors may be accommodated but bandwidth restriction limits the rise time improvement.

With such high value resistors, it is essential to keep the surfaces clean so avoiding leakage. Such resistors also have significant temperature coefficients and it is therefore almost essential to operate such amplifiers in an evacuated and temperature controlled environment to achieve their ultimate performance. As feedback resistance is increased, there is an attendant rise in Johnson noise and for these high value resistors this is the major contribution to the overall system noise level. Since gain increases proportionally to resistance, but Johnson noise increases as the square root of resistance, then signal-to-noise should increase as the square root of resistance.

This assumes that the shot noise arising from the amplifier bias current is relatively negligible, but this will eventually be the limiting factor. Bias current in FET devices decreases with a decrease in temperature, so cooling the operational amplifier is beneficial. Cooling the integrated circuit amplifier produces a reduction in bias current, practically by a factor of 2 for 10 C reduction in temperature, and this allows higher feedback resistances to be used, so obtaining better signal-to-noise ratio and lower output offset. Thus limiting power dissipation in the system is of great importance.

The open-loop gain of operational amplifiers with the required low bias current have minimum specified gains of about 100dB (10) which is not sufficient to ensure a low enough voltage level at the summing junction (317, Fig. 3) to provide the required linearity of response. The loopgain is therefore increased by introduction of operational amplifier A2 connected for a gain of the order of 100 (40dB), inside the ioop to increase the loop-gain to 140dB minimum (10). This ensures that for a typical maximum output of 1OV the summing junction is at a maximum of imicrovolt. The amplifier A2 is required to have a wide bandwidth so that at a closed loop-gain of the above 40dB there is little contribution to phase lag in the region of the overall-loop unity-gain frequency of the complete system. The phase lag of the open-loop circuit can approach close to 3600 at low frequency but the compensated closed-loop circuit retains excellent response with good gain and phase margins.

The CMOS type operational amplifiers with bias currents of a few femtoamp typically have an overall supply voltage range of about 1 5V or less, limited output current rating and unsymmetrical gains for positive or negative going signals. The introduction of A2 overcomes these limitations since it can have higher voltage supplies, much greater output current, symmetrical swings and, in the non-inverting configuration shown in Fig. 3, presents a very high input impedance to the output of Al so making its limitations inconsequential.

A2 may also be located outside any evacuated and cooled enclosure for Al and so improve overall performance. The much reduced voltage swing at the output of Al also decreases local capacitive feedback between the output and input pins of the integrated circuit and allows the use of much lower voltage supplies. The lower supplies mean lower power dissipation and reduced possibility of current leakage from the supplies to the input pin.

For bipolar operations supplies can be as low as the minimum for the integrated circuit device (say +1- 3V). Appropriate integrated circuit amplifiers omit the offset adjustment controls of common standard integrated circuit devices and hence the pins on the amplifier package previously used for this purpose are no longer required and have become "no connection" and hence may be connected to the common rail. This avoids leakage from these previously high voltage pins, provides additional screening of the sensitive inverting input (usually pin 2) and assists in thermal contact between amplifier body and, for example, a cooled ground plane.

The following paragraphs detail the various advantages achieved by the inventions: - A design for the screen to allow both screening and thermal contact between resistor and cooling device to enhance thermal stabilization of the resistor without affecting the capacitive coupling between resistor and screen.

- An improved feedback sampling phase-control network driving the feedback resistance to allow adjustment of the system for optimum transient response.

- Techniques for actively driving the feedback resistor screens to allow in-circuit feedback capacity adjustment using the same phase-control network as for the resistor feedback but with separate amplitude control.

- Achievement of increased open-loop gain to ensure minimal deviation of the summing junction from common voltage to ensure improved linearity of response, to minimise stray leakage current at the input, to minimise the effect of input capacity and still retain circuit stability.

- As a result of increased open-loop gain by the use of dual gain amplifier topology the reduction of output signal swing from the amplifier Al so reducing power dissipation therein and local capacitive feedback from output to input and hence improved time response.

- The freedom to configure the amplifier A2 as non-inverting thus presenting a high input resistance to the amplifier Al and hence no loading of it, improving linearity and decreasing power dissipation in amplifier Al.

- As a consequence of the reduced demands on the amplifier Al the freedom to choose an input operational amplifier with lower input bias and quiescent current, and lower voltage supplies.

- As a consequence reduction of power dissipation in the operational amplifier Al reduced heating and hence improved ability to cool the system.

- As a result of improved cooling the achievement of reduced amplifier Al input bias current, consequential shot noise, and greater signal stability.

- As a result of reduced voltage supplies a decrease in possible leakage current from the supplies to the system input.

- A circuit arrangement to achieve optimum noise performance and to obtain a noise level determined primarily by Johnson noise generated by the feedback resistor.

- As a consequence of this the achievement of the theoretical improvement in signal-to-noise ratio from increase in the feedback resistor value up to at least 1 OOTohm.

- The freedom to choose the amplifier A2 to suit the output requirement in terms of voltage and current output, and bandwidth.

- The freedom to place the amplifiers A2 and A3 outside the evacuated and cooled enclosure thus again reducing power dissipation internally.

- The freedom to place all the circuit adjustments outside the evacuated and cooled enclosure allowing easy in situ adjustment in normal operating conditions.

Claims (10)

1. I. A transimpedance amplifier comprising a current source and an associated overall input capacitance, primary amplification means having an input to which said current source is connected and an output, said output being optionally connected to secondary amplification means having one or more inputs and an output, a first resistor, a second resistor and a first capacitor connected in series between the output of the primary or secondary amplification means if provided and a common voltage rail, a sampling network consisting of a signal tap provided between said first resistor and said second resistor which may optionally be connected to an input of tertiary amplification means having one or more inputs and an output, to deliver a feedback signal, a feedback network being both resistive and capacitive and which is connected between the sampling network or the output of said tertiary amplification means and the input of the primary amplification means at a summing node operable to sum the signal from the feedback network with an input signal from the current source, characterised in that the feedback network comprises at least one resistor at least partially disposed within a surrounding electrically conducting screen and which is charged with a screen drive signal which is a function of the feedback signal.
2. 2. An amplifier according to claim 1 wherein the screen completely surrounds said resistor.
3. 3. An amplifier according to claim 1 or 2 wherein the screen is substantially concentric with said resistor
4. 4. An amplifier according to any preceding claim wherein said resistor is disposed completely within said screen in that the extremities of said resistor are disposed inwardly of ends of said screen.
5. 5. An amplifier according to any preceding claim wherein said screen is circular in cross-sectional shape.
6. 6. An amplifier according to any preceding claim wherein said tertiary amplification means is connected to the voltage tap.
7. 7. An amplifier according to any preceding claim wherein said tertiary amplification means is in the form of a unity-gain buffer having an output which delivers the feedback signal to the capacitive feedback network.
8. 8. An amplifier according to any preceding claim wherein said current source is connected to the input of primary amplification means in the form of an input operational amplifier.
9. 9. An amplifier according to claim 8 wherein the output of the operation amplifier is connected to the input of said secondary amplification means in the form of a loop-gain amplifier.
10. 10. An amplifier according to claim 9 wherein the output of said loopgain amplifier is connected to the series connected first resistor, second resistor and first capacitor and second capacitor.