1. Field of the Invention
This invention relates to the field of current integration circuits.
2. Description of the Related Art
It is often necessary to know the magnitude of a particular current over time. This can be determined with a current integrator.
Current integrators are well-known; a basic implementation is shown in
This arrangement suffers a number of shortcomings, however. If Vmax is the maximum output voltage that A1 can produce, then the maximum charge Qmax that can be stored on integration capacitor C without causing A1's output to become saturated is given by Qmax=Vmax*C. The total charge to be integrated is given by
(where Tint is the integration period), or Qtotal=Iin×Tint if Iin is constant. Capacitor C needs to store Qtotal, SO Qmax≧Qtotal, or C≧Qtotal/Vmax. Thus, to achieve a high Qtotal requires a large C value to ensure that the amplifier does not saturate. The area used by C can dominate the area of an integrated circuit die.
Another shortcoming of the circuit in
A current integration circuit is presented which overcomes the problems noted above. The present circuit can integrate a large input charge, while keeping the integration capacitance small and improving the output resolution when compared with prior art integrators.
The invention includes an operational amplifier which receives an input current to be integrated. An integration capacitor is connected between the op amp's output and inverting input, which integrates the input current and causes the op amp's output voltage to increase or decrease, depending on the direction of the input current.
To prevent the op amp's output from becoming saturated due to a large input current, a charge dumping circuit is employed. The charge dumping circuit is arranged to dump a known charge to the junction of the integration capacitor and the op amp's inverting input. The dumped charge is of the opposite polarity to that stored on the integration capacitor, and thus acts to reduce the charge on the integration capacitor and thereby prevent the op amp's output from becoming saturated. The charge dumping circuit is controlled with a control circuit, which is arranged to trigger a charge dump whenever the op amp's output exceeds a predetermined trip voltage, but before it becomes saturated. Counting the number of charge dumps performed during a given integration cycle provides a coarse indication of the magnitude of the integrated input current, and the output of the op amp—when connected to an analog-to-digital converter (ADC), for example—provides a fine quantization of the integrated current. A high output resolution with large input currents is achieved by combining both coarse and fine quantizations, even if only a small integration capacitor is used.
Further features and advantages of the invention will be apparent to those skilled in the art from the following detailed description, taken together with the accompanying drawings.
The basic principles of a current integration circuit per the present invention are illustrated in
There is a maximum output voltage which A1 is capable of producing. If Iin is large and C1 is small, A1's output can become saturated; when this occurs, Vout no longer accurately represents the value of Iin over time. To prevent A1's output from becoming saturated, the present current integration circuit includes a charge dump circuit 10. Circuit 10 is arranged to dump a known charge Qdump to the junction 20 of A1's inverting input and C1, in response to a control signal 30. The known charge is of opposite polarity with respect to the charge stored on C1, such that when Qdump is dumped to junction 20, the charge stored on integration capacitor C1 is reduced, as is output voltage Vout.
Control signal 30 is provided by a control circuit 40, which receives the output Vout from A1 at one input, and a trip voltage Vtrip at a second input. Control circuit 40 is arranged to trigger charge dump circuit 10 as necessary to prevent A1's output from saturating. It does this by triggering a charge dump each time Vout exceeds trip voltage Vtrip, which is set to a voltage that is less than A1's saturation voltage. In this way, A1's output is kept out of saturation, and the integration of input current Iin is not interrupted.
The current integration circuit is arranged to integrate the input current over an integration period Tint. The number of times that Qdump is dumped during Tint provides a coarse indication of the magnitude of the integrated input current. The number of times that Qdump is dumped during Tint is tracked by a counting means 45, such as a digital counter which is incremented every time that control circuit 40 triggers charge dump circuit 10. This count, when multiplied by the magnitude of charge Qdump (using, for example, a microprocessor), provides a coarse quantization of the integrated input current.
The output of op amp A1 then provides a fine indication of the input current magnitude, and feeding A1's output to an A/D converter, for example, provides a fine quantization. By combining both coarse and fine quantizations, a high output resolution with large input currents is achieved—even if only a small integration capacitor (occupying a correspondingly small die area) is used.
In the prior art current integration circuits described above, the integration capacitance had to be equal to or greater than Qtotal/Vmax to store Qtotal (defined above). However, when arranged in accordance with the present invention, integration capacitor C1 can be less than Qtotal/Vmax with the smaller capacitor requiring a smaller die area. For example, if C1=Qtotal/(16*Vmax), then the integrator's output resolution is increased by 16 times (4 bits) for a given A/D converter, and the capacitance of C1 can be 16 times smaller than the capacitance in the circuits shown in
The present invention preferably includes a reset switch SR, which is connected between the op amp's output and inverting input and is controlled by control circuit 40. In operation, prior to integrating Iin, reset switch SR is closed and counting means 45 is reset, thereby resetting integration capacitor C1 and the charge dump count. Switch SR is then opened, allowing input current Iin to be integrated using C1. If Iin is such that A1's output approaches saturation, control circuit 40 triggers charge dump circuit 10 as needed, in the manner described above.
A basic embodiment of the invention is shown in
Charge dump circuit 10 comprises a first switch Scharge, a second switch Sdump, and a dump capacitor CD. Switch Scharge is arranged to connect a reference voltage Vref to a junction 50 when closed, and switch Sdump is arranged to connect junction 50 to junction 20 when closed. Dump capacitor CD is connected between junction 50 and ground. Switch Sdump receives control signal 30 from control circuit 40, and switch Scharge receives another control signal 60 from control circuit 40. Charge dump circuit 10 is operated by first closing Scharge to initialize CD by storing charge Qdump on capacitor CD, and then closing Sdump to dump charge Qdump to junction 20.
As configured in
The magnitude of Qdump is given by CD×Vref. To match the coarse quantization provided by the current integration circuit to the fine quantization provided by the A/D converter, Vref is preferably equal to the A/D converter's reference voltage. When Qdump is dumped to junction 20, the voltage across C1 is lowered by Vref×(CD/C1). CD is preferably smaller than C1, so that A1's output does not change polarity when Sdump closes.
As previously mentioned, Qdump must have the opposite polarity to Iin in order to discharge C1. Thus, the circuit of
Control circuit 40 preferably comprises a comparator A2 which receives Vout at one input and Vtrip at a second input; A2's output, for example, goes high (“toggles”) when Vout exceeds Vtrip, and goes low when Vout falls back below Vtrip. The output of A2 is control signal 30; when Vout exceeds Vtrip, A2's output toggles and closes Sdump, dumping Qdump to junction 20. Control circuit 40 also preferably includes an inverter 70, which inverts the output of comparator A2 to produce the control signal 60 which operates Scharge. Thus, for example, when Vout is less than Vtrip, A2's output is low and the output of inverter 70 is high, which closes Scharge and allows CD to be charged by Vref. Then, when Vout exceeds Vtrip, A2's output goes high and the output of inverter 70 goes low, which causes Qdump to be dumped to junction 20, reducing the charge stored on C1. Note that the output of A2 needs to be delayed to ensure enough time for Qdump to be fully transferred from CD to C1 before Sdump is opened and Scharge is closed. Means for achieving this are not shown, but are well-known to those of ordinary skill in the art of analog circuit design. This cycle is repeated as necessary throughout an integration period.
Control circuit 40 also preferably includes control logic 80, which provides control signals to operate reset switch SR and input switch Sin, and to reset counting means 45.
As noted above, the number of times that charge Qdump is dumped to junction 20 during an integration period Tint, multiplied by Qdump, gives a coarse quantization of the integrated current. The number of charge dumps can be tracked with a counting means 45 such as a digital counter, which counts the number of times that A2's output toggles.
A1's output is preferably provided to an A/D converter 46 to provide a fine quantization of the integrated current. To avoid saturating A/D converter 46, Vout should be less than the A/D's reference voltage Vref; therefore, Vtrip should be made less than or equal to Vref.
Although the circuit in
This problem is overcome in
Switches Sca and Scb are operated with control signal 60, and switches Sda and Sdb are operated with control signal 30. In operation, when Vout is (for example) less than Vtrip, Sca and Scb are closed by signal 60, initializing CD by discharging it. When Vout exceeds Vtrip, Sca and Scb are opened and Sda and Sdb are closed (by signal 30), such that a charge Qdump (given by CD×Vref) flows from C1 to CD. Vref is selected so that Qdump has the opposite polarity with respect to the charge stored on C1, so that dumping Qdump has the effect of reducing the charge stored on C1 by Qdump. This configuration is insensitive to parasitic capacitances from junctions 82 or 84 to ground. When so arranged, the control circuit 40 operates in the following sequence:
If Iin flows toward A1's inverting input and Vref is positive, or if Iin flows away from A1's inverting input and Vref is negative, the polarity of the charge dumped by the circuit of
The charge dump circuits of
Dumping Qdump into A1's inverting input causes its output to change quickly, which in turn causes transient voltages to appear back at A1's inverting input. It may be desirable to limit the magnitude of these transients at A1's input, because the linearity of the input current source (such as a photodiode) might be impaired.
One way in which such transients can be limited is to release Qdump slowly by limiting the slew rate of control signal 30 provided to dump switches Sdump, Sda and Sdb—assuming that the switches are such that the resistance between its signal terminals changes continuously with the magnitude of the control signal (as with a FET switch). This can be accomplished with an RC network (not shown), for example, which is interposed between the output of comparator A2 and the control input of switches Sdump, Sda and Sdb.
Alternatively, the rate at which Qdump is dumped could be limited by making switches Sdump, Sda and Sdb with a larger resistance, using a FET with a small width, for example, to limit the magnitude of voltage transients at A1's inverting input. This resistance could be further increased by interposing a series resistor between Sdump and Sdb, and junction 20.
Another way in which the effect of voltage transients on the input current source can be reduced is by opening switch Sin for a brief period when Qdump is dumped into C1 to isolate the current source from the integration circuit. When switch Sin is temporarily opened, the input current will be integrated across the capacitance of the input current source, but this charge will be transferred to C1 when switch Sin is closed again.
Note that the embodiments of control circuit 40 shown in
Switches Scharge, Sdump, Sca, Scb, Sda, Sdb, SR and Sin are preferably FET switches. The gate of each FET serves as the switch's control input, and its drain and source serve as the switch's signal terminals. The control signals provided by control circuit 40 are preferably arranged to turn on their respective FET switches such that the resistance between their signal terminals is reduced to near zero. Note that other types of switches, including electromechanical switches, could also be used—as long as they are switchable by means of a control signal and present a near-zero resistance between their signal terminals when closed.
While particular embodiments of the invention have been shown and described, numerous variations and alternate embodiments will occur to those skilled in the art. Accordingly, it is intended that the invention be limited only in terms of the appended claims.
This application claims the benefit of provisional patent application No. 60/297,960 to Tang, filed Jun. 12, 2001.
| Number | Name | Date | Kind |
|---|---|---|---|
| 6194703 | Kuderer | Feb 2001 | B1 |
| 6300613 | Kuderer | Oct 2001 | B1 |
| 6621441 | Haroun et al. | Sep 2003 | B2 |
| 6781434 | Jensen et al. | Aug 2004 | B2 |
| 20040004488 | Baxter | Jan 2004 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 60297960 | Jun 2001 | US |