Electrical System With Small Signal Suppression Circuitry

This invention relates to an electrical system and, in particular, to an electrical system including circuitry for suppressing small, noise-like, input signals.

A problem which exists in the art and which is solved by the present invention may be best explained with reference to FIG. 1 which shows a system designed to sense the "knock" of an engine and means for generating signals to correct for the knock. A knock sensor 10, which may generally be a microphone and which is located on or near an automobile engine, produces a knock signal (ek) in response to the "knocking" of the engine. The signal (ek) is then applied to an amplifier 12 whose output is applied to the input of an anti-aliasing filter 13 whose output is applied to the input of a programmable gain stage 14. In response to the knock signal ek the programmable gain stage 14 produces at its output an in-phase signal (ek1) and an out-of-phase signal (ek1b). The signal ek1 is applied to a bandpass filter 16a and the signal ek1b is applied to a bandpass filter 16b. The output of filter 16a produces a signal identified as V.sub.IN and the output of filter 16b produces a signal identified as V.sub.INB. Theoretically, in-phase signal V.sub.IN should be the exact complement (or inverse) of the out-of-phase signal V.sub.INB. Input signals V.sub.IN and V.sub.INB are then applied to a rectifier section 18 which controls the application of the signals V.sub.IN and V.sub.INB to an integrator 20. Integrator 20 includes a positive signal integrator 20a and a negative signal integrator 20b. The rectifying circuit 18 includes circuitry for comparing V.sub.IN and V.sub.INB and switches for enabling the positive going portion of signals V.sub.IN and V.sub.INB to be applied to integrator 20a and the negative going portion of signals V.sub.IN and V.sub.INB to be applied to integrator 20b. As a result, the outputs of integrators 20a and 20b function to increase the positive and negative amplitude of the knock signal over selected integrating intervals.

The outputs of integrators 20a and 20b are fed to a differential to single-ended amplifier whose output charges a storage capacitor C24 whose potential is used to drive a buffer 26 whose output is then used to control (reduce) the engine knock.

A problem exists when there is no knock signal (e.g., the knock sensor is disconnected) or when the knock signal is very small. It is virtually impossible to manufacture a programmable gain stage (e.g., 14) such that the circuit sections producing the in-phase and out-of-phase signals and their corresponding filters (e.g., 16a, 16b) are totally identical. No matter how carefully the circuit and its components are made, there are differences (albeit small) in the components and their lay out. Consequently, when the knock signal (ek) is very small or non-existent, a net difference or offset will nevertheless exist between V.sub.IN and V.sub.INB. Although the difference and offset may be small, a significant error may result. This is because the outputs of the bandpass filters (16a, 16b) are fed to a rectifier section 18 to control the application of V.sub.IN and V.sub.INB to integrators 20a and 20b. The difference between V.sub.IN and V.sub.INB due to noise and offsets produces a "false" signal which increases in time, magnifying the effect of the false signal. Therefore, it is an object of this invention to override and eliminate the effect of "low level" and "offset" signals.

In systems embodying the invention, circuit means responsive to first and second input signals controls the application of the input signals to a positive signal integrator and to a negative signal integrator. When the amplitude of the input signals is greater than a predetermined value, the one of the two input signals which is positive relative to the other is applied to the positive signal integrator and the one of the two input signals which is negative relative to the other is applied to the negative signal integrator. When the amplitude of the input signals is smaller than a predetermined level, the circuitry causes the periodic application, for a time period, of the first input signal to the positive signal integrator and the second input signal to the negative signal integrator, and the periodic application, for a like time period, of the first input signal to the negative signal integrator and the second input signal to the positive signal integrator.

In one embodiment, the first input signal is coupled via a first switch to the input of the positive signal integrator and via a second switch to the input of the negative signal integrator. The second input signal is coupled via a third switch to the input of the negative signal integrator and via a fourth switch to the input of the positive signal integrator. When the input signals are smaller than a predetermined amount, the switches are periodically and alternatively turned on and off such that the first and third switches are turned on and the second and fourth switches are turned off during a given time period and then, the first and third switches are turned off and the second and fourth switches are turned on during a like given time period.

In one embodiment of the invention the circuit means includes a comparator having first and second inputs and an output. First and second complementary input signals are respectively applied to the first and second comparator inputs and first and second complementary control signals are respectively applied to the first and second comparator inputs whereby, the net signal at the first input is a function of the first input signal and the first control signal and the net signal at the second input is a function of the second input signal and the second control signal. The first and second control signals are fixed amplitude square waves which vary at a first rate. The amplitude of the control signals is selected such that for values of input signals below a predetermined level, the control signals determine the nature and value of the signals at the comparator inputs. The comparator produces a signal at its output having a first value when the signal applied to its first input is more positive than the signal applied to its second input and produces a signal at its output having a second value when the signal applied to its second input is more positive than the signal applied to its first input. The output of the comparator is selectively supplied to the input of a latch circuit whose output controls the turn-on and turn-off of the four switches described above.

In the accompanying drawings like reference characters denote like components, and

FIG. 1 is a block diagram of a prior art system;

FIG. 2 is a schematic diagram of a circuit embodying the invention for use in the system of FIG.1;

FIG. 3 is a block diagram of a network for producing control signals for use in the circuit of FIG. 2;

FIG. 4 is a wave form diagram of control signals applied to the circuit and of the circuit response at selected nodes;

FIG. 5 is a more detailed schematic diagram of circuitry for applying control signals to a comparator, in accordance with the invention; and

FIG. 6 is a schematic diagram of a switched capacitor integrator suitable for use in the circuit of FIG. 2.

The circuit of FIG. 2 illustrates a portion of the rectifier section 18 embodying the invention. An in-phase signal V.sub.IN (See FIG. 1) is applied to an input terminal 51 and a signal V.sub.INB (See FIG. 1) which is the complement of V.sub.IN (180.degree. out-of-phase thereto) is applied to input terminal 52. The signal V.sub.IN is applied via a switch S9 to one input 53 of an amplifier A1 and the signal V.sub.INB is applied via a switch S10 to another input 54 of amplifier A1. A selectively enabled switch S1 is connected between the one input (53) of amplifier A1 and a reference potential (e.g., VDD/2), and a selectively enabled switch S2 is connected between the other input (54) of amplifier A1 and the reference potential (e.g., VDD/2). Amplifier A1 has an in-phase output e1 and an out-of-phase output e1b. The signal e1 is coupled via a capacitor C2 to a node 61 to which is connected the positive input terminal of an amplifier A2, which is designed to function as a comparator. The signal e1b is coupled via a capacitor C1 to a node 62 to which is connected the negative input terminal of amplifier A2. Normally, capacitors C1 and C2 are designed to have the same values.

The small (noise) signal suppression and cancelling circuit embodying the invention includes: (a) a capacitor C4 connected between a terminal 55 and node 61 and a capacitor C3 connected between a terminal 56 and node 62; (b) circuitry responsive to a clock signal CLx for generating a control voltage V1 and its complement (or inverse) V1B, as shown in FIG. 3; and (c) circuit means for applying the voltage V1 to terminal 55 and the voltage V1B to terminal 56. Normally, capacitors C3 and C4 are designed to have the same values. Capacitors C1 and C3 form a first capacitive divider network and capacitors C2 and C4 form a second capacitive divider network, which function as described below.

A selectively enabled switch S3 is connected between node 62 and the reference potential (e.g., VDD/2), and a selectively enabled switch S4 is connected between node 61 and the reference potential (e.g, VDD/2). Amplifier A2 compares the signal, e61, produced at node 61, with the signal, e62, produced at node 62, and produces an output signal e2 which is applied via switch S11 to the input of a latch circuit comprised of inverters A3 and A4 and switch S12.

Inverter A3 responds to the signal e2 and produces an inverted output e3 which is applied to the input of inverter A4. Inverter A4 responds to the signal e3 and produces an inverted output e4 which is fed back to the input of inverter A3 via switch S12, when switch S12 is closed. When S12 is closed, inverters A3 and A4 are cross coupled and function as a latch.

When switch S12 is open, switch S11 is closed coupling the output e2 of comparator amplifier A2 to the input of inverter A3. Subsequently, when S11 opens and S12 closes, the latch circuit (A3, A4) stores a state indicative of the condition of e2 at the time S12 closes (and S11 opens). By way of example, if the output e2 of amplifier A2 is above a certain level (e.g., above VDD/2), e3 goes low (ground) and e4 goes high (e.g., VDD). When S12 closes, these conditions will be maintained (stored) by cross-coupled inverters A3 and A4, so long as S12 remains closed. If, on the other hand, e2 of amplifier A2 is below the certain level (e.g., below VDD/2), then e3 goes high (e.g., VDD) and e4 goes low (e.g., ground). When S12 closes, these conditions will be maintained (stored) by the latch (A3 and A4), so long as S12 remains closed.

The latch has two outputs, e3 and e4, which are complementary to each other. The output e3 of A3 is used to control the turn-on and turn-off of switches S6 and S7 and the output e4 of A4 is used to control the turn-on and turn-off of switches S5 and S8. For purpose of illustration, when e3 is high and e4 is low, switches S6 and S7, which are controlled by e3, are closed, and switches S5 and S8, which are controlled by e4, are open. Alternatively, when e3 is low and e4 is high, switches S6 and S7 are opened and switches S5 and S8 are closed.

When switches S5 and S8 are turned-on (switches S6 and S7 are turned-off), the input signal V.sub.IN is applied to positive signal integrator 20a via S5, and input V.sub.INB is applied to negative signal integrator 20b via S8. When switches S6 and S7 are turned-on (switches S5 and S8 are turned-off), the input signal V.sub.IN is applied to integrator 20b via S6 and the input signal V.sub.INB is applied to integrator 20a via S7.

The operation of the circuit and, in particular, that portion pertaining to the invention will now be discussed in greater detail.

For ease of explanation in the discussion to follow, assume that switches S1 through S12 are MOS transistors of N conductivity type. That is, each one of these "switching" transistors is turned-on (i.e., switch-closed) when a "high" voltage is applied to its gate electrode and each one of the switching transistors is turned-off (i.e., switch-open) when a "low" voltage is applied to its gate electrode.

It should be understood that the switches could instead be a mixture of N and P type MOS transistors or that each switch could comprise a pair of complementary MOS transistors (i.e., an N and a P type MOS transistor with their conduction paths connected in parallel).

The circuit of FIG. 2 may be operated by a control clock signal (CL1) and its complement (CL2) and a control voltage (V1) and its complement (VIB) which are produced by clock shaping network 300, shown in FIG. 3. A clock signal CLx is applied to clock shaping network 300 which is designed to produce: (a) a first control clock signal, CL1 and a second control clock signal, CL2, which is the inverse or complement of CL1; and (b) a first control signal V1 and a second control signal V1B which is the complement, or inverse, of V1. V1 is a square wave whose rate (frequency) is related to the rate at which the switches S1, S2, S3 and S4 are opened and closed. V1 varies between fixed voltage levels (e.g., VDD and ground, though different levels may be used). For appropriate operation, the clocking and control signals need to be appropriately phased. One arrangement of CLx, CL1, CL2, V1 and V1B is shown in FIG. 4.

The first and second control signals (V1 and V1B) are designed to vary at one half the rate of clock signal CL1 (or CL2). For the particular embodiment of FIG. 2, V1 and V1B are designed to go from low-to-high and high-to-low when CL1 is low, i.e., when switches S3 and S4 are open). By way of example, note that V1 goes from low-to-high when CL1 is low and before CL1 goes from low-to-high, and that V1 goes from high-to-low after CL1 has gone from high-to-low and is low. Clock shaping network 300 may be any one of a number of circuits capable of producing the described signals and need not be detailed.

Referring back to FIG. 2, CL1 is applied to switches S1, S2, S3 and S4 and CL2 is applied so as to control the turn-on and turn-off of switches S9 and S10. V1 is applied via a capacitive divider network to an input of the comparator. V1B, which is the complement of V1, is also applied via a capacitive divider network to another input of the comparator. The amplitude of V1 (and V1B) is selected such that when it is applied to the capacitive network, it produces a desired voltage step at its corresponding input to the comparator A2. In practice, a clock signal CL1x, derived from CL1, is applied to and controls the turn-on and turn-off of S12 and the clock signal CL2x, derived from CL2, is applied to and controls the turn-on and turn-off of switch S11. The phasing of CL1X and CL2x is designed to ensure that the latch circuit (A3, A4) is in a latch condition (i.e., S12 closed) for most of the time. However, for ease of illustration in the discussion to follow, CL1x will be assumed to be like CL1 and CL2x will be assumed to be like CL2. Where the switches are N type MOS transistors, the clock signals are applied to the gate electrode of these transistors. Consequently, S1, S2, S3, S4 and S12 are closed (i.e., turned-on), at the same time, during the positive going portion of the CL1 signal. During this clock portion, S9, S10 and S11 are open. During the negative going portion of the CL1 signal, S1, S2, S3, S4 and S12 are open (i.e., turned-off) and S9, S10 and S11 are closed (i.e, turned-on).

When CL1 goes high, which defines the autozero phase of the comparator A2 and amplifier A1, switches S1, S2, S3 and S4 are closed causing the reference potential (e.g., VDD/2) to be applied to the inputs (53, 54) of amplifier A1 and to the inputs of amplifier A2. For ease of explanation, assume that when switches S1 and S2 are closed, the outputs e1 and e1b of amplifier A1 are at, or close to, VDD/2 volts, where VDD and ground are the operating potentials applied to amplifier A1.

When CL1 goes low (CL2 goes high), which defines the compare phase of comparator A2 and amplifier A1, V.sub.IN is coupled via S9 to one input (53) of amplifier A1 and V.sub.INB is coupled via S10 to the other input (54) of amplifier A1. If V.sub.IN is positive relative to VDD/2, e1 will go positive relative to VDD/2 and e1b will go negative relative to VDD/2. If V.sub.IN is negative relative to VDD/2, e1 will be, or go, negative relative to VDD/2 and e1b will be, or go, positive relative to VDD/2.

The outputs (e1 and e1b) of amplifier A1 are coupled via capacitors C1 and C2 to the input nodes (61, 62) of comparator amplifier A2. In the absence of C3 and C4 and the application of control signals V1 and V1B, the signals produced at nodes 61 and 62 are solely a function of e1 and e1b. With capacitors C3 and C4 in the circuit and with V1 and V1B applied to these capacitors, the signals produced at nodes 61 and 62 are a function of e1 and V1 for node 61 and e1b and V1B for node 62. If the signal at node 61 is more positive than the signal at node 62, the output e2 of A2 goes high (e.g., VDD volts). On the other hand, if the signal at node 61 is less positive than the signal at node 62, the output e2 goes low (e.g., ground). The output e2 is coupled via switch S11 (when S11 is on) to the input of inverter A3 of the latch circuit.

If the signal e2 is high, the output e3 of inverter A3 will be driven low and the output e4 of inverter A4 will be driven high. For this condition, switches S5 and S8 are turned-on and switches S6 and S7 are turned-off. The closure of switch S5 causes V.sub.IN to be applied to the input of positive signal integrator 20a and the closure of switch S8 causes V.sub.INB to be applied to the input of negative signal integrator 20b.

If e2 is low, e3 will be driven high and e4 will be driven low. For this condition, switches S6 and S7 are turned-on (closed) and switches S5 and S8 are open. The closure of switches S6 and S7 causes V.sub.IN to be applied to the input of integrator 20b and V.sub.INB to be applied to the input of integrator 20a. Thus, the outputs of the latch control which one of the input signals V.sub.IN and V.sub.INB are applied to the integrators 20a and 20b.

As discussed above with respect to FIG. 1, a problem arises if the knock sensor is disconnected and/or when the input signals are very low. For these conditions, there should be virtually no net output from the integrators and no signals should be applied via the buffer driver 26 to the engine control (See FIG. 1). However, when there is no knock signal and/or the input signal is very low, due to various offsets and/or differences in the components or their layouts, a condition may exist where e1 may be constantly greater than e1b or e1b may be constantly greater than e1. As a result, one pair of rectifier switches (e.g., S5, S8) may be closed for a long period of time and the other pair (e.g., S6, S7) may be opened for that period of time. Consequently, the outputs of integrators 20a and 20b will indicate the presence of an ever increasing knock signal when such signal does not actually exist.

Applicant's invention is designed to ensure that when the knock signal, as represented by V.sub.IN and V.sub.INB, is below a certain level, indicative of little or no knock error, the outputs of integrators 20a and 20b are, on the average, at or close to zero. As detailed below, this is accomplished by turning the rectifier switches (S5, S6, S7 and S8) on and off alternatively. That is, for one portion of a clock cycle, switches S5 and S8 are closed and switches S6 and S7 are opened and for the next portion of the clock cycle, switches S5 and S8 are opened and switches S6 and S7 are closed. As a result, for the one portion of each clock cycle, V.sub.IN is applied to integrator 20a and V.sub.INB is applied to integrator 20b and then for the next portion of each clock cycle, V.sub.IN is applied to integrator 20b and V.sub.INB is applied to integrator 20a. As a result, the average outputs of integrator 20a and integrator 20b will be at, or close to, zero volts.

This is best explained with reference to FIGS. 2, 3 and 4. As already noted, control signal V1 and its complement control signal V1B are generated by network 300. Signal V1 is applied to a terminal 55 and signal V1B is applied to terminal 56. The signal V1 is coupled via capacitor C4 to node 61 and signal V1B is coupled via capacitor C3 to node 62. The amplitude of signals V1 and V1B and the values of capacitors C3 and C4 are selected to ensure that V1 and V1B will determine the value of signals at nodes 61 and 62 when the input signals e1 and e1b are below a predetermined level. That is control signals V1 and V1B are selected such that for e1 and/or e1b below a predetermined level, the signal voltage at node 61 will be more positive than the signal voltage at node 62 for one portion of a clock cycle and that the signal voltage at node 61 will be more negative than the signal voltage at node 62 for a like portion of the next subsequent clock cycle. This alternation of the signals generated at nodes 61 and 62 continues until e1 and/or e1b is/are greater than a predetermined value.

The effect of V1 and V1B on the voltages e61 and e62 produced at nodes 61 and 62, respectively, is as follows.

For ease of description, refer to FIG. 4 starting at time t0 with CL1 low and CL2 high, which defines a compare phase. For this condition, switches S1, S2, S3 and S4 are open and switches S9 and S10 (and at some point, S1) are closed. The signal V.sub.IN is coupled via switch S9 to the input 53 of amplifier A1 which then produces a signal e1. Likewise, the signal V.sub.INB is coupled via switch S10 to the input 54 of A1 which then produces a corresponding output signal e1b. During this time (t0 to t1), CL1 is low causing switches S1, S2, S3 and S4 to be open and V1 is low, and V1B is high. From time t0 to t1, the signal (e61) produced at node 61 and the signal e62 produced at node 62 may be expressed as follows:

and

At time t1 (during a compare phase and after the application of input signals), the signal V1 makes a transition from zero volts to VDD volts, producing a positive going signal at node 61 which is approximately equal to (VDD).times.[c4/(c2+c4)]. In a complementary manner, at time t1, the signal V1B makes a transition from VDD volts to zero volts producing a negative going signal at node 62 which is approximately equal to (-VDD).times.[c3/(c1+c3)]. The net signal at node 61 is equal to the contribution of e1 and V1 and the net signal at node 62 is equal to the contribution of e1b and V1B. The signals at nodes 61 and 62 may then be expressed as:

and

where the values of capacitors c1 and c2 are substantially equal to each other and the values of capacitors c3 and c4 are substantially equal to each other. Equations 1 and 2 may be rewritten as:

and

where: kA=c2/(c2+c4)=c1/(c1+c3); and

kB=c3/(c1+c3)=c4/(c2+c4).

Note that VDD, kA and kB are constant. Hence, equations 3a and 3b may be rewritten as follows:

and

where: es=e1(kA); esb=e1b(kA); and

Vc=VDD[(c3/(c3+c4)]=VDD[c4/(c2+c4)]

Note:

a) Vc represents a fixed potential which is periodically added to, and subtracted from, nodes 61 and 62; b) e1 may be positive or negative and that e1b is normally the inverse of e1; c) e61 will be positive for all values of es which are more positive than -Vc; and d) e62 will be negative for all values of esb which are not more positive than -Vc.

The signals e61 and e62 are compared by comparator A2 which produces a "high" signal e2 for the condition where e61 is more positive than e62. Thus, so long as e1 and/or e1b is, or are, below a predetermined level, e61 will be driven positive relative to e62 when V1 goes high and V1B goes low.

At time t2, S11 closes (S1, S2, S3 and S4 open) transferring the high e2 signal to the input of A3 producing a low at e3 and a high at e4. The signal e4 going high causes S5 and S8 to be turned on whereby V.sub.IN is coupled to integrator 20a and V.sub.INB is coupled to integrator 20b for a period lasting from t2 to t6 [one cycle of the clock CL1 or CL2].

From time t2 to t4, which defines an autozero phase, switches S1, S2, S3 and S4 are closed clamping the inputs of A1 and A2 to ground. During this period, S12 is closed and S11 is open whereby the latch (A3, A4) remains in the condition to which it was set at time t2.

At time t4, CL1 goes low and CL2 goes high closing switches S9 and S10 (and also S11), defining a new compare phase. The closing of S9 and S10 causes V.sub.IN and V.sub.INB to be applied to the inputs of amplifier A1 to produce corresponding output signals e1 and e1b. Signals e1 and e1b are then coupled via capacitors C2 and C1, respectively, to node 61 and 62 to produce signals which, as above, may be expressed as:

and

When e1 and e1b are smaller than a predetermined level, they will cause the voltages at nodes 61 and 62 to vary about a reference level at those nodes. However, as discussed below, if they are less than a predetermined level, the control signals V1 and V1B, when applied, will override and mask the "input signals" e1 and e1b.

At time t5 (during a compare phase after the application of input signals), V1 makes a negative going transition from VDD to ground and V1B makes a positive going transition from ground to VDD. The se transitions superimpose a negative going step voltage of -(VDD)[c4/(c2+c4)] on node 61 and a positive going step voltage of (VDD)[c3/(c1+c3)] on node 62. The voltages at nodes 61 and 62 may then be expressed as follows:

and

As discussed above, these equations may be expressed more simply as follows :

and

Referring to equations 6a and 6b, note that es and esb are normally the inverse of each other and may be positive or negative. So long as +Vc is more positive than -esb, e62 will be positive relative to the reference level and so long as the value of es is such that it is not greater than -Vc, e61 will be negative relative to the reference level. If these conditions exist, e62 will then be positive with respect to e61. This will cause e2 to go "low". The "low" at e2 is coupled via switch S11 to the input of A3 causing e3 to go high and e4 to go low.

Then, at time t6, switch S11 is opened and switch S12 is closed causing the e3 high and the e4 low conditions to be stored in the latch (A3, A4). With e3 high and e4 low, V.sub.IN is applied to integrator 20b and V.sub.INB is applied to integrator 20a from time t6 to t10 which is equal to one clock cycle of CL1 or CL2.

Thus, for values of e1 and e1b below a predetermined level, V.sub.IN is applied to integrator 20a and V.sub.INB is applied to integrator 20b for one clock cycle and then V.sub.IN is applied to integrator 20b and V.sub.INB is applied to integrator 20a for the next clock cycle. This procedure is repeated so long as e1 and e1b remain below predetermined levels.

In general, an examination of equations 4a and 4b and 6a and 6b and the waveforms of FIG. 4 reveals that for one transition of the control signals (e.g., V1 going positive and V1B going negative--which defines a first compare phase) a positive going voltage step is applied to node 61 and a negative going voltage step is applied to node 62. For the next transition of the control signals (e.g., V1 going negative and V1B going positive--defining the second compare phase) a negative going voltage step is applied to node 61 and a positive going voltage step is applied to node 62. This causes V.sub.IN and V.sub.INB to be alternatively applied to the integrators; e.g., During the odd numbered compare phases, V.sub.IN will be applied to integrator 20a and V.sub.INB will be applied to integrator 20b. During the even numbered compare phases, V.sub.IN will be applied to integrator 20b and V.sub.INB will be applied to integrator 20a. By alternately applying V.sub.IN and V.sub.INB to the integrators, it is evident that (except for some offset errors) the average outputs of integrators 20a and 20b should be at, or close to, zero.

One aspect of the invention is best understood by reviewing the operation of the circuit for the condition when the input signals (V.sub.IN and V.sub.INB) are greater than a predetermined value and control the value (i.e., amplitude) of the signals applied to the inputs of comparator A2. That is, assume that the signals e1 and e1b produced at the outputs of amplifier A1 which are respectively coupled via capacitors C2 and C1 to the inputs of comparator amplifier A2 "override" or "mask" the effect of V1 and V1B coupled via C4 and C3. If, and when, V.sub.IN is more positive than V.sub.INB, the output e2 of A2 is driven high and V.sub.IN which is positive relative to V.sub.INB is coupled to positive integrator 20a and V.sub.INB which is negative relative to V.sub.IN is coupled to negative integrator 20b. If, and when, V.sub.IN is negative relative to V.sub.INB, the output e2 of A2 is driven low. For this condition, e3 will be high and cause V.sub.IN which is negative relative to V.sub.INB to be applied to integrator 20b and V.sub.INB which is positive relative to V.sub.IN to be applied to integrator 20a. Thus, when V.sub.IN and V.sub.INB are greater than a predetermined level, it is the relative amplitude of V.sub.IN and V.sub.INB which controls the closure of the rectifying switches to ensure that the more positive of the two input signals is always applied to the positive integrator circuit and the more negative of the two input signals is always applied to the negative integrator circuit. Thus, the signals produced at the outputs of integrators 20a and 20b increase as a function of time.

In accordance with the invention, whenever input signals V.sub.IN and V.sub.INB are below a predetermined level, the control signals V1 and V1B which may be described as square waves having a frequency f1 and a period T1 (where T1=1/f1) and which vary in amplitude between first and second levels override the "input signals" present at the inputs to the comparator. The control signals V1 and V1B cause the switches of the rectifier circuit to be opened and closed at a rate dependent on the frequency of the control signals. The switches of the rectifier circuit are then alternatively opened and closed for like (equal) periods of time as a function of the amplitude and frequency of the control signals. Thus, in accordance with the invention, the input signal V.sub.IN is applied to positive signal integrator 20a for a first period of time and to the negative signal integrator for a second period of time, subsequent to said first period, where the first and second periods are of like duration. Concurrently, V.sub.INB is applied to negative signal integrator 20b during the first period and to the positive signal integrator during the second period. The alternative application of V.sub.IN and V.sub.INB to integrators 20a and 20b results in the outputs of integrators 20a and 20b being equal to zero, on the average. That is, for small values of input signals (i.e., below the predetermined level), the input signals are applied to the integrators 20a and 20b such that the application of the input signals constitute a 50% duty cycle operation.

It has been shown that the comparator and amplifier circuit of FIG. 2 has alternate autozero and comparator phases. During the autozero phase, switches S1, S2, S3 and S4 are closed and switches S9 and S10 are open. During each comparator phase, switches S1, S2, S3 and S4 are open and switches S9 and S10 are closed. In addition, control signals (V1, V1B) are generated and applied to the comparator A2 during a portion of each compare phase, after the application of respective first and second input signals. If the input signals are greater than some predetermined level, they control the outputs of the comparator. On the other hand, if the input signals are below a predetermined level, set by the amplitude of the control signals, then the control signals determine the output conditions of the comparator.

The generation of input signals e1 and e1b and their application via capacitors C1 and C2 to the inputs of comparator A2 is as shown in FIG. 2. The generation of V1 and V1B and the coupling of V1 and V1B with e1 and e1b, respectively, to the inputs of comparator A2 may be as shown in FIG. 5. A square wave generator 99 which may be part of a clock and control signal generating network, such as network 300, produces a square wave VX1 which is applied to a divider network (R1, R2) whose output, identified as node 101, is applied to the positive input of an amplifier A5. The negative input of amplifier A5 is grounded. Amplifier A5 is preferably made to be similar to A1, whereby A5 has very similar characteristics to, and a similar response as, A1. Amplifiers A1 and A5 may be any general purpose differential amplifier having a differential input and a differential output. Amplifier A5 has a first output 103 at which is produced a square wave signal V1 of fixed amplitude and frequency, and a second output 104 at which is produced a square wave signal V1B which is the complement (or inverse) of V1. V1 is coupled via capacitor C4 to the positive input of A2 and V1B is coupled via capacitor C3 to the negative input of amplifier A2. Where, for example, the amplitude of VX1 is 5 volts (i.e., VX1 varies between zero and 5 volts) and the gain of A5 (and A1) is 100 and it is desired to have a "dead" zone of approximately 5 millivolts, the following ratios may be used. Making R1=999R2 yields a VX2 equal to VX1/1,000, whereby VX2 varies between zero and 5 millivolts. If A5 has a gain of 100, V1 will vary between zero and 500 millivolts. By making C1=C2=99C4=99C3, the contribution of V1 to node 61 (and V1B to node 62) is equal to: (500 mV)[c4/(c2+c4)]=(500 mV)[c4/(99c4+c4)]=5 millivolts.

As noted above, at the end of each compare phase period of comparator A2, the comparator output (e2) is coupled via switch S11 to the latch (A3, A4) and sets its output to either one of two binary conditions. The latch output, in turn, controls the application of V.sub.IN and V.sub.INB to integrators 20a and 20b. At the end of each compare phase, V.sub.IN is applied to one of the two integrators (20a, 20b) and V.sub.INB is applied to the other integrator. These integrators then provide an amplified signal at their output indicative of the input signal condition.

Each one of integrators 20a and 20b may be a switched capacitor integrator of the type shown in FIG. 6. Each integrator includes an output terminal 201 for the application thereto of an input signal (V.sub.IN or V.sub.INB) which is coupled via a switch S62 to one terminal 203 of a capacitor C.sub.A. A reference voltage V.sub.R (which may be equal to VDD/2) is applied to an input terminal 202 which is coupled via switch S61 to terminal 203. The other terminal of C.sub.A is connected to terminal 204 which is connected to the negative input terminal of an amplifier A6. V.sub.R is applied to the positive input terminal of A6. A selectively enabled switch S63 is connected between terminal 204 and output terminal 205 of A6. Selectively enabled switches S64 and S66 are connected between terminals 204 and 205 and the two terminals of a capacitor C.sub.B. A reset switch S65 is connected across capacitor C.sub.B. A clock A (CLA) signal controls the turn-on and turn-off of switches S62, S64 and S66 and a clock B (CLB) signal controls the turn-on and turn-off of switches S61 and S63.

The switched integrator has an autozero phase (CLB is high, CLA is low) and an integrate phase (CLA is high, CLB is low).

During autozero phase (CLB high, CLA low), switches S61 and S63 are closed and switches S62, S64 and S66 are open. Comparator C.sub.B then holds the charge accumulated during the previous cycle. During an integrate phase (CLB is low, CLA is high), switches S61 and S63 are open and switches S62, S64 and S66 are closed. During the integrate phase, a charge is added to C.sub.B. The additional charge may be expressed as follows:

The change in the voltage across C.sub.B may be expressed as follows:

where .DELTA.V=(C.sub.A /C.sub.B)(V.sub.IN -V.sub.R)

Therefore, the output of each integrator varies as follows:

FIG. 6 shows a switched capacitor integrator. However, it should be understood that any suitable integrator circuit could be used instead.