Integrator topology for continuous integration

[0002] This application is a divisional of copending U.S. application Ser. No. 09/502,134, filed Feb. 11, 2000, now issued as U.S. Pat. No. 6,380,790.

[0004] This invention relates to integrators.

[0005] Integrators have high linearity, wide bandwidth, and low noise characteristics. Integrators, however, require a reset interval to discharge the capacitor in the integrator's feedback loop which results in significant “dead” times in measurements and harmful transients on the integrator's input. Additionally, the rapid discharge interval aggravates the problem of dielectric absorption, thereby undermining the lower limit of instrument precision.

[0006] Referring to FIG. 1, an integrator 10 includes a feedback loop having a switch 12 in parallel with a feedback capacitor 14. The switch 12 allows the feedback capacitor 14 to discharge when the switch 12 is closed. Placing one or more strings of series resistors and capacitors in parallel with the feedback capacitor 14 with or without the switch 12 reduces at least some of the harmful effects of this discharge. However, even in some arrangements having multiple capacitors, dielectric absorption is still a problem since the charge in the series capacitors is redistributed with the feedback capacitor 14.

[0007] The invention overcomes these unwanted effects of integrator reset by providing integrator circuit topologies that enable continuous integration, without the need for reset of the integrator circuit. One such integrator circuit includes a first integrator and a second integrator, each of the two integrators having a non-inverting terminal. Each of the non-inverting terminals is connected to an input node to alternately receive an input current for continuous integrator circuit integration without integrator circuit reset.

[0008] In further configurations, the inverting terminal of the second integrator can be connected to an inverting terminal of the first integrator. The non-inverting terminal of the second integrator can be connected to an output of the first integrator through a first capacitor, and the output of the second integrator can be connected to the non-inverting terminal of the first integrator through a second capacitor. In operation, the first integrator and the second integrator have voltages on respective ones of the inverting and non-inverting terminals that are substantially equal, and the two integrators produce output voltages that are complementary.

[0009] In a further integrator circuit provided by the invention, at least one integrator is provided, having an input for receiving an input current. A plurality of integrator feedback capacitors are provided, with each capacitor being connected to alternately charge and discharge, based on the integrator input current. This cooperative charging and discharging enables continuous integrator circuit integration without integrator circuit reset.

[0010] These integrator circuit topologies can be employed in a wide variety of applications in which low signal level, precise measurements are required. For example, in one biological application, the first integrator and the second integrator can be operated to each introduce an output voltage into a chemical bath on either side of a biological membrane. In this application, the integrator circuit is configured to detect fluctuations of ion channels. In another application, e.g., the integrator circuit can be configured for charge detection.

[0011] These applications are particularly well-served by the integrator circuit of the invention in its elimination of a need for rapid discharging of feedback capacitors during operation. The integrator circuit of the invention can perpetually integrate incoming current signals, such as low-level transducer signals, to produce an output of a continuous flow of two complementary voltages. This perpetual integration eliminates “dead time” and input transients, compensates for charge injection at the integrator input, and reduces the harmful effects of dielectric absorption. At the same time, the integrator circuit maintains a high degree of operational linearity, produces a low level of noise, and can accommodate a wide bandwidth of input signals.

[0012] Other features and advantages of the invention are provided in the following detailed description and the accompanying drawings, and in the claims.

[0013] FIG. 1 is a schematic diagram of a conventional integrator.

[0014] FIG. 2 is a block diagram of a chopper stabilizing circuit.

[0015] FIG. 3 is a schematic diagram of the block diagram of FIG. 2.

[0016] FIG. 4 is a graph showing the output of the integrator circuit of FIG. 3.

[0017] FIGS. 5-6 are graphs showing the chopper stabilization of the integrator circuit of FIG. 3.

[0018] FIG. 7 is an unfolded view of the integrator circuit of FIG. 3.

[0019] FIGS. 8-9 are graphs showing the output of the integrator circuit of FIG. 3.

[0020] FIGS. 10-11 are graphs showing the response of the integrator of FIG. 3 to input current.

[0021] FIGS. 12-13 are graphs showing the charge injection compensation of the integrator circuit of FIG. 3.

[0022] FIG. 14 is a graph showing currents detectable by the integrator circuit of FIG. 3.

[0023] FIG. 15 is a graph showing charge detection by a conventional integrator circuit.

[0024] FIG. 16 is a graph showing charge detection by the integrator circuit of FIG. 3.

[0025] FIG. 17 is a schematic diagram of the differentiator circuit of FIG. 2.

[0026] Referring to FIG. 2, a chopper stabilizing circuit 20 includes a switching circuit 22, an integrator circuit 24, a sensing circuit 26, a control circuit 28, and a differentiator circuit 30. In general, the chopper stabilizing circuit 20 has a topology and is controlled in a manner that eliminates the need for rapid discharging of feedback capacitors in the integrator circuit 24. In particular, and as will be discussed in greater detail below, this advantage is accomplished by alternating the signal current from the switching circuit 22 to the integrator circuit 24. In this way, the integrator circuit 24 can perpetually integrate these incoming current signals (low-level transducer signals) and output a continuous flow of two complementary voltages. The sensing circuit 26 detects when one of the complementary voltages reaches a threshold value and notifies the control circuit 28. The control circuit 28 responds by sending a signal to the switching circuit 22. This signal changes the position of switches in the switching circuit 22, thereby alternating the signal current to the integrator circuit 24. The differentiator circuit 30 receives the complementary voltages output by the integrator circuit 24 and provides a demodulated differentiation bit stream representing the slope of the complementary voltages. As will be described in more detail below, this chopper stabilizing circuit 20 eliminates dead time and input transients, compensates for charge injection at the input, and reduces the harmful effects of dielectric absorption. At the same time, the chopper stabilizing circuit 20 maintains high linearity, low noise, and wide bandwidth.

[0027] In the layout of the chopper stabilizing circuit 20, the switching circuit 22 has an input at a first node 32 for receiving an input signal. The input signal includes the driving current/voltage for the chopper stabilizing circuit 20 from a load, a current source, and/or a voltage source. The switching circuit 22 has an output at a second node 34 that is determined by the position of the switch(es) included in the switching circuit 22. The integrator circuit 24 has an input at the second node 34 for receiving an input signal from the switching circuit 22 and an output at a third node 36. The sensing circuit 26 has an input at the third node 36 for receiving an input signal from the integrator circuit 24 and an output at a fourth node 38. The control circuit 28 has an input at the fourth node 38 for receiving an input signal from the sensing circuit 26 and output at a fifth node 40 and a sixth node 42. The switching circuit 22 has an input for receiving an input signal from the control circuit 28 at the fifth node 40. This input signal controls the position of the switch(es) in the switching circuit 22.

[0028] The differentiator 30 is shown in FIG. 2, though its presence is not necessary to ensure proper functioning of the chopper stabilizing circuit 20. If it is not present, the integrator circuit 24 and the control circuit 28 may not necessarily have outputs at the third node 36 and the sixth node 42, respectively. The differentiator circuit 30 has an input at the third node 36 for receiving an input signal from the integrator circuit 24 and at the sixth node 42 for receiving an input signal from the control circuit 28. The input signal at the sixth node 42 controls the switch(es) included in the differentiator circuit 30. The differentiator also has an output at a seventh node 44.

[0029] Referring to FIG. 3, one particular embodiment of a chopper stabilizing circuit 20 includes a switching circuit 22, an integrator circuit 24, a sensing circuit 26, and a control circuit 28. The chopper stabilizing circuit 20 eliminates the need for rapid discharging of feedback capacitors 60a-b(preferably Teflon®) in the integrator circuit 24 by alternating the signal current from the switching circuit 22 to two integrators 62a-bincluded in the integrator circuit 24. In this way, one feedback capacitor discharges while the other charges, thereby providing two inversely related output voltages (Vout+, Vout−) at Vout nodes 36a-b. Once either of the output voltages reaches a predetermined threshold value (Vth), a regenerative comparator 76a-bincluded in the sensing circuit 26 and connected to this output voltage is tripped. Hysteresis prevents the sensing circuit 26 from causing false resets. The comparator 76a-btriggers a D-type flip-flop 78 through a NAND gate 79, both included in the control circuit 28. As the flip-flop 78 changes state, the outputs Q and Q-bar connected to the switches 66a-b, 68a-bcause them to reverse position. This reversal preserves the same orientation with respect to the load 72, maintaining a uniform bias, while alternating the signal current to the integrator circuit 24.

[0030] More specifically, the switching circuit 22 includes two pairs of two symmetric switches 66a-b, 68a-b. The switches 66a-b, 68a-bmay be any type of standard MOS (metal oxide semiconductor) switch, e.g., MAXIM 326. Only one set of switches 66a-b, 68a-bis closed at a time, each closed switch providing a path for a signal to the non-inverting input terminal of an operational amplifier (opamp) 70a-b, e.g., Burr-Brown OP627, included in the integrators 62a-b. When the phase one (>1) switches 66a-bare closed, a load 72 provides the input current (Io) to the first opamp 70awhile a voltage source 74 provides the bias voltage (Vb) to the second opamp 70b. When the phase two (>2) switches are closed, the load 72 and the voltage source 74 provide current/voltage to the other opamp 70a-b. The values of Vout+ at the Vout node 36aand Vout− at the Vout node 36bdepend on the position of these switches 66a-b, 68a-b.

[0031] FIG. 4 shows the inverse relationship between Vout+ (Vcf2) and Vout− (Vcf1). In this scenario, the >2 switches 68a-bbegin closed and the feedback capacitors 60a-binitially are discharged, so Vout+ and Vout− begin at Vb. When Io flows through the load 72, Vout+ and Vout− alternately and inversely ramp up and down in accordance with: $\frac{dV}{dt}=\frac{\mathrm{Io}}{\mathrm{Cf}}.$

[0032] When Io decreases at a time t1, this relationship ceases.

[0033] The integrator circuit 24 can effectively integrate forever (constantly flowing Io), with negligible glitching during phase switching. This lack of glitch is helped by the symmetry of input stage of the integrator circuit 24. Every input stage node 80a-csees one switch 66a-b, 68a-bturn on and another turn off during a phase transition. The already low charge injection of the switches 66a-b, 68a-bis then effectively reduced to tens of femtoCoulombs (fC). Additionally, the symmetric pair requires no voltage drop across a switch 66a-b, 68a-b, aiding in keeping leakage currents below a picoAmp (pA). The voltages at the input stage nodes 80a-care substantially the same.

[0034] Referring to FIGS. 5 and 6, it is appreciated that offset may be a problem as in FIG. 5, but techniques exist to alleviate this problem, e.g., a stabilizing circuit. FIG. 5 shows the chop before stabilization, and FIG. 6 shows the chopper stabilization of the integrator circuit 24.

[0035] Referring to FIG. 7, an unfolded view of the integrator circuit 24 helps demonstrate the manner in which the circuit functions. The compensation of the integrator circuit 24 may be broken down into two sections: minor and major loops. The minor loop concerns the stability of each opamp 70a-b; the major loop comprises the total feedback loop around the integrator. The major loop encompasses a unity gain inverter with a voltage divider formed by the first feedback capacitor 60areacting with the capacitance off the input stage of the first opamp 70a. The input capacitance is dominated by the opamp input capacitance and the parasitics of the switches 66a-b, 68a-b. The ratio of the capacitive voltage divider in this embodiment is approximately ten, which will keep the major loop crossover well below that of the minor loops. The minor loops are stabilized with the addition of shunt capacitances 82a-b, which help compensate for phase lag due to shunt resistors 84a-b(preferably metal film) reacting with the input capacitance of the opamps 70a-b. With the bandwidth of the opamps 70a-bon the order of 10 MHz in this embodiment, the chopper stabilizing circuit 20 should be able to track currents with a bandwidth of approximately 1 MHz.

[0036] FIGS. 8-13 further demonstrate the functioning of the integrator circuit 24. FIG. 8 shows Vout+ and Vout− with 50 μs per horizontal division, the typical reset duration in standard integrators, e.g., Axopatch 200B and nuclear physics instrumentation. FIG. 9 shows a zoom in on the reset transient, with the switching occurring of the order of 500 ns, e.g., 700 ns. The 2 pF feedback capacitor 60a-band a residual voltage jump of 20 mV signifies under 40 fC of charge injection. FIG. 10 shows the response of the integrator circuit 24 (top trace) to input current (bottom trace), a 2 nA peak-to-peak triangle wave. Because of this response, the integrator circuit 24 could be used for direct digitization of input current via single-slope integration by measuring the period between resets. FIG. 11 shows the response of the integrator circuit 24 in FIG. 10 superimposed with a 100 kHz sinusoid supplied by a 2 pF capacitor at the input. FIG. 12 shows the charge injection before compensation, and FIG. 13 shows the charge injection after compensation by the integrator circuit 24.

[0037] Now referring to FIG. 14, the integrator circuit 24 can be used to detect the fluctuations of ion channels important in cell signaling and biological transport. These currents range from 0.1 pA to 100 pA, with bandwidths of 10 kHz. The integrator circuit 24 allows for measuring these currents without glitches from resetting.

[0038] Now referring to FIGS. 15 and 16, the integrator circuit 24 can also be used for charge detection. For example, x-ray and particle detectors output charge pulses that are usually integrated. Whenever a conventional integrator hits a limit value as in FIG. 15, it must reset and data can be lost. Using the integrator circuit 24, the dead-time (lost data) is greatly reduced by the absence of capacitor resets as shown in FIG. 16.

[0039] A differentiator circuit 30, shown in FIG. 17, may be part of a chopper stabilizing circuit. The differentiator circuit 30 includes two switches 92a-b. The switches 92a-bmay be any type of standard MOS (metal oxide semiconductor) switch, e.g., MAXIM 326. Each switch 92a-bis either in a horizontal (>1) position, e.g., switch 92afrom a top start node 94ato a top end node 96a, or a diagonal (>2) position, e.g., switch 92afrom the top start node 94ato a bottom end node 96b, at any given time. Each closed switch 92a-bprovides a path for a signal at entering nodes 36a-bto travel to the inverting terminal or to the non-inverting terminal of an opamp 100. Input from a control circuit (not shown) determines the position of the switches 92a-b. If the differentiator circuit is connected to the chopper stabilizing circuit 20 (see FIG. 2), the output from the control circuit 78 provides the phase information for the switches 92a-b.

[0040] A number of embodiments of the invention have been described. Nevertheless, it will be understood that various modifications may be made without departing from the spirit and scope of the invention. Accordingly, other embodiments are within the scope of the following claims.