TUNABLE TRANSCONDUCTANCE-CAPACITANCE FILTER WITH COEFFICIENTS INDEPENDENT OF VARIATIONS IN PROCESS CORNER, TEMPERATURE, AND INPUT SUPPLY VOLTAGE

Abstract

A transconductance-capacitance (Gm-C) filter of arbitrary order is provided that is biased by a bias circuit such that the Gm-C filter is robust to variations in process corner and temperature as well as input supply noise. The bias circuit includes a biased transistor that has a width-to-length ratio that is a factor X times larger than a corresponding transistor in the Gm-C filter. The biased transistor couples to ground through a switched capacitor circuit.

Description

TECHNICAL FIELD

The present invention relates generally to filters, and more particularly to tunable transconductance-capacitance (G_m-C) filters.

BACKGROUND

Transconductance-capacitance (G_m-C) filters offer attractive performance characteristics. Thus, the use of G_m-C filters is widespread and pervasive in radio communications and signal processing. Analog G_m-C filters are constructed using operational transconductance amplifiers (OTAs). OTAs operate to translate a voltage input signal into a current output signal. An example balanced (differential output) OTA is shown in FIG. 1a. The transconductance G_m for the OTA determines the I⁺ and I⁻ currents based upon the input voltages V⁺ and V⁻ according to the following equations:

I ⁺ =G _m(V⁺−V⁻)

I−=G _m(V⁻−V⁺)

Various approaches are known to construct OTAs such as using cascodes or differential architectures. A simple analog transconductance-capacitance (G_m-C) filter may be constructed using a single-ended OTA as shown in FIG. 1b. If a time constant τ is defined as C_l/G_m, then it can be shown that V_in for this filter equals V_out+τd(V_out(t)/dt. The cutoff frequency for the resulting G_mC filter will thus rely on both G_m and C_l. But process corner variations will typically be in the range of 20% for a desired capacitance whereas a desired transconductance will have process corner variations in the range of 10%. It follows that the resulting time constant τ for such a filter will be accurate to just 30% across all the process corner variations. Moreover, transconductance values will vary significantly with temperature and the supply voltage level. In addition, input noise will introduce variations in the filter coefficients. Accordingly, it is conventional to provide some sort of tuning circuitry on G_mC filters. In this fashion, a tunable G_m-C filter has its time constant set to some desired value with some isolation from variations in the power supply voltage, process corner, and temperature.

Although such independence is desirable, conventional tunable G_m-C filters are still sensitive to power supply variations and suffer from non-idealities. Accordingly, there is a need in the art for improved tunable G_m-C filters that are more robust to variations in process corner, power supply, and temperature.

SUMMARY

In accordance with one aspect of the invention, a transconductance-capacitance (G_m-C) filter is provided that includes: a plurality of operational transconductance amplifiers (OTAs), wherein a first one of the OTAs has a first transconductance and the remaining ones of the OTAs have transconductances that are proportional to the first transconductance, and a bias circuit for biasing the first transconductance to a desired value responsive to a clock frequency, the bias circuit including a switched capacitor circuit generating a resistance inversely proportional to the clock frequency, wherein the desired transconductance value is proportional to the clock frequency.

In accordance with another aspect of the invention, a transconductance-capacitance (G_m-C) filter is provided that includes: a plurality of operational transconductance amplifiers (OTAs), wherein each OTA includes a differential pair of transistors providing a tail current to a third transistor having a transconductance g_m, and a bias circuit for biasing a gate of a given one of the third transistors with a control voltage, the bias circuit including a switched capacitor circuit such that a transfer function for the G_m-C filter is proportional to a ratio of capacitances and is independent of process corner variations.

In accordance with yet another aspect of the invention, a bias circuit to bias the transconductance g_m of a first transistor within a G_m-C filter is provided that includes: a second transistor having a width-to-length ratio that is a factor X larger than a width-to-length ratio of the first transistor, the second transistor coupling to ground through a switched capacitor circuit such that g_m is proportional to (1−1/X^1/2).

The scope of the invention is defined by the claims, which are incorporated into this section by reference. A more complete understanding of embodiments of the present invention will be afforded to those skilled in the art, as well as a realization of additional advantages thereof, by a consideration of the following detailed description of one or more embodiments. Reference will be made to the appended sheets of drawings that will first be described briefly.

BRIEF DESCRIPTION OF DRAWINGS

FIG. 1 a is a schematic diagram of an operational transconductance amplifier (OTA).

FIG. 1 b is a schematic diagram of a conventional transconductance-capacitance (G_m-C) filter.

FIG. 2 is a schematic diagram of a biquad G_m-C filter tuned using a switched-capacitor bias circuit.

FIG. 3 is a circuit diagram of a differential amplifier within an OTA in the filter of FIG. 2.

FIG. 4 a is a circuit diagram illustrating the equivalence of a switched capacitor circuit to a resistor.

FIG. 4 b is a circuit diagram of a switched capacitor circuit adapted for greater robustness to parasitic effects.

FIG. 5 is a circuit diagram for a bias circuit to provide the control voltage applied in the circuit of FIG. 3.

Embodiments of the present invention and their advantages are best understood by referring to the detailed description that follows. It should be appreciated that like reference numerals are used to identify like elements illustrated in one or more of the figures.

DETAILED DESCRIPTION

Reference will now be made in detail to one or more embodiments of the invention. While the invention will be described with respect to these embodiments, it should be understood that the invention is not limited to any particular embodiment. On the contrary, the invention includes alternatives, modifications, and equivalents as may come within the spirit and scope of the appended claims. Furthermore, in the following description, numerous specific details are set forth to provide a thorough understanding of the invention. The invention may be practiced without some or all of these specific details. In other instances, well-known structures and principles of operation have not been described in detail to avoid obscuring the invention.

To provide a tunable G_m-C filter that self-compensates with regard to process corner variations, power supply variations, and temperature variations, a switched capacitor circuit is used to tune the transconductance G_m of one of more of the OTAs included within the G_m-C filter. In that regard, a biquad second order G_m-C filter such as filter 100 shown in FIG. 2 includes five different OTAs, each having their own independent transconductance (denoted as g_m1 through g_m5). In addition, filter 100 includes 3 classes of capacitors (having corresponding capacitances C₁ through C₃). The transfer function H(s) for filter 100 thus depends on the various transconductances and capacitances as given by the following equation:

$\begin{array}{cc}H\left(s\right)=\frac{s^2\left[\frac{C_3}{C_3+C_2}\right]+s\left[\frac{G_{m5}}{C_3+C_2}\right]+\left[\frac{G_{m2}G_{m4}}{C_1\left(C_3+C_2\right)}\right]}{s^2+s\left[\frac{G_{m3}}{C_3+C_2}\right]+\left[\frac{G_{m1}G_{m2}}{C_1\left(C_3+C_2\right)}\right]}.& \left(1\right)\end{array}$

This relatively complex behavior can be simplified as follows. Although transconductances have large variations in their absolute values, relative transconductance values can be set quite accurately by the ratio of the OTA transistor widths (provided the same channel lengths are used for all devices). Thus, an arbitrary OTA such as OTA 105 having a transconductance g_m1 may be used to define the transconductance of the remaining OTAs. For example, the transconductance g_m2 for OTA 110 may be defined as K_m2-g_m1, the transconductance g_m3 for OTA 115 may be defined as K_m3g_m1, and so on. In general, the ith transconductance can be expressed in terms of the first OTA as

G _mi =K _mi G _mi  (2)

Similarly, the sum of the capacitances C₃ and C₂ can be defined in terms of C1 using a constant K as

C ₃ +C ₂ =KC ₁  (3)

Using equations (2) and (3), equation (1) can be simplified as follows

$\begin{array}{cc}H\left(s\right)=\frac{s^2\left[\frac{C_3}{C_3+C_2}\right]+s\left(\frac{K_{m5}}{K}\right)\left(\frac{G_{m1}}{C_1}\right)+\left(\frac{K_{m2}K_{m4}}{K}\right){\left(\frac{G_{m1}}{C_1}\right)}^2}{s^2+s\left(\frac{K_{m3}}{K}\right)\left(\frac{G_{m1}}{C_1}\right)+\left(\frac{K_{m1}K_{m2}}{K}\right){\left(\frac{G_{m1}}{C_1}\right)}^2}.& \left(4\right)\end{array}$

From equation (4), it can be seen that if just the ratio of Gm1/C1 is tuned to be self-compensating with regard to variations in power supply, process corner, and temperature, then the remaining transconductance/capacitance ratios need no tuning since these ratios can be conventionally manufactured to an accuracy of approximately one percent.

Although the above simplification was described with regard to the biquad filter 100 of FIG. 1, it can be shown that any order (nth order) of G_m-C filters can be tuned in this fashion. In other words, a single one of the OTAs may be self-compensated as discussed further herein yet the entire G_m-C filter will be self-compensated. This self-compensation may be better understood with reference to the transistor differential pair within each OTA. An example differential pair of matched transistors M1 and M2 is shown in FIG. 3. The tail current from transistor M1 and M3 is biased by a control voltage V_cntl driving the gate of a transistor M3. Matched transistors M1 and M2 each have a transconductance of G_m whereas M3 is sized to have a transconductance of AG_m.

The following discussion will show how to generate the bias voltage V_cntl such that the ratio of G_m/C_L for the OTA is self-compensating. This self compensation will rely on the use of a switched capacitor circuit to produce a desired resistance. As known from Ohm's law, a voltage potential V_A−V_B applied across a resistor of resistance R will produce a current I equaling (V_A−V_B)/R. However, as seen in FIG. 4a, the same amount of charge can be moved between these voltage potentials using a switched capacitor circuit 400 that couples voltage V_A to a capacitor having a capacitance C_ck through a switch S₁. Similarly, the capacitor couples to voltage V_B through a switch S₂. If switched S₁ is driven on and off by a clock of frequency f_ck while switch S₂ is driven by the complement of this clock, it can be shown that a current I flowing through the capacitor will equal f_ckC_ck(V_A−V_B). Thus, the switched capacitor circuit functions as a resistor having a resistance R_m of

R _m=1/f_ckC_ck.  (5)

The equivalence of a switched capacitor circuit to provide a desired resistance is made more precise by using the additional switches S₃ and S₄ as shown in FIG. 4b for a switched capacitor circuit 405 in that the additional switches make the circuit parasitic insensitive. S₁ and S₄ are driven by the clock whereas S₂ and S₃ are driven by the complement of the clock.

The incorporation of a switched capacitor circuit into a bias circuit 500 as shown in FIG. 5 for the generation of the control voltage V_cntl will now be discussed. A pair of PMOS transistors P₁ and P₂ form a current mirror. Thus, if P₁ and P₂ are matched (same width W and length L and thus the same W/L ratio), they will each source the same current I. Thus, a current I flows through an NMOS transistor M₄ and an NMOS transistor M₅. M4 is diode connected between the drain of a PMOS transistor P₁ and ground so as to be in saturation mode. Transistor M₄ is matched to M₃ of FIG. 3. The gate of M₄ is tied to the gate of transistor M₅, where M₅ is larger than M₄. If M₅ has the same length L as does M₄, then the width of M₅ is a factor X times larger than a width W for M₄. The sources of both P₁ and P₂ are driven by a power supply voltage node V_CC. The source of M₅ couples to ground through a switched capacitor circuit 505 that functions to provide a resistance of R_m. It can be shown that the transconductance g_m4 for M₄ can be expressed as

g _m4=2/R_m(1−1/Sqrt(X))  (6)

Substitution of equation (5) into equation (6) allows the transconductance to be expressed as

g _m4=2(1−1/Sqrt(X))f_ckC_ck  (7)

It will be appreciated that the switched capacitor circuit 505 may be made more robust as discussed with regard to FIG. 4b. Examination of equation (7) shows that the transconductance dependence on the width X is such that by making X sufficiently large, the necessary clock frequency for driving the switched capacitor circuit is reduced. This is a substantial advantage over other techniques used to make G_m-C filters more robust to variations in power supply voltage (input noise), process corners, and temperature.

Referring back to FIG. 3, one can see that if the transconductance of M₃ is controlled by the control voltage V_cntl generated as discussed with regard to FIG. 5, the ratio of G_m/C_L for the OTA/capacitor combination including such a differential pair can be expressed as

G _m /C _L=2A(1−1/Sqrt(X))f_ckC_ck/C_L  (8)

Referring again to FIG. 2, suppose that the OTAs were all matched in the sense of having matched differential pairs of transistors as discussed with regard to FIG. 3. It has already been shown with regard to equation (4) that if just one of the transconductances is tuned, then the overall G_m-C for the filter is established. As seen by equation (8), the filter coefficients will depend only on the ratio of device parameters and capacitances. This is quite advantageous as the resulting filter coefficients will be independent of process and temperature variations as well as power supply noise. For example, a fast process corner will affect C_ck equally as it does affect C_L. Thus, the ratio of capacitances cancels out process corner variations. The same argument applies to temperature and power supply noise. Moreover, although this self compensation of a G_m-C filter has been discussed with regard to the biquad filter 100 of FIG. 2, the same self-compensation can be applied to any nth order G_m-C filter.

It will be obvious to those skilled in the art that various changes and modifications may be made without departing from this invention in its broader aspects. The appended claims encompass all such changes and modifications as fall within the true spirit and scope of this invention.

Claims

1. A transconductance-capacitance (Gm-C) filter, comprising: a plurality of operational transconductance amplifiers (OTAs), wherein each OTA includes a differential pair of transistors providing a tail current to a third transistor having a transconductance gm, anda bias circuit for biasing a gate of a given one of the third transistors with a control voltage, the bias circuit including a switched capacitor circuit such that a transfer function for the Gm-C filter is proportional to a ratio of capacitances and is independent of process corner variations.

2. The Gm-C filter of claim 1, wherein the transfer function for the Gm-C filter is independent of temperature variations.

3. The Gm-C filter of claim 1, wherein the transfer function for the Gm-C filter is independent of power supply noise.

4. The Gm-C filter of claim 1, wherein the bias circuit includes a fourth transistor matched to the given third transistor and a fifth transistor that has a width-to-length ratio that is a factor X times larger than a width-to-length ratio for the fourth transistor, and wherein the switched capacitor circuit is coupled between a source for the fifth transistor and ground.

5. The Gm-C filter of claim 4, wherein the bias circuit further comprises a first PMOS transistor, wherein the fourth transistor is a diode-connected NMOS transistor coupled between a drain of the first PMOS transistor and ground, and wherein the fourth transistor has a gate coupled to a gate of the fifth transistor.

6. The Gm-C filter of claim 5, wherein the bias circuit further comprises a second diode-connected PMOS transistor, wherein the a drain of the second PMOS transistor couples to a drain of the fifth transistor and wherein a gate of the first PMOS transistor couples to a gate of the second PMOS transistor.

7. The Gm-C filter of claim 6, wherein a source for each of the first and second PMOS transistors couples to a power supply voltage node.

8. The Gm-C filter of claim 7, wherein the control voltage equals a gate voltage for the fourth transistor.

9. A bias circuit to bias the transconductance gm of a first transistor within a Gm-C filter, comprising: a second transistor having a width-to-length ratio that is a factor X larger than a width-to-length ratio for the first transistor, the second transistor coupling to ground through a switched capacitor circuit such that gm is proportional to (1−1/X1/2).

10. The bias circuit of claim 9, wherein a gate of the first transistor is biased by a control voltage that equals a gate voltage for the second transistor.

11. The bias circuit of claim 10, wherein the switched capacitor circuit is driven by a clock and a complement clock of frequency fck and has a capacitor of capacitance Cck, whereby the switched capacitor circuit provides a resistance of 1/fckCck.

12. The bias circuit of claim 11, wherein the control voltage drives a gate of a third transistor that matches the first transistor.

13. The bias circuit of claim 12, wherein the third transistor is diode connected.

14. The bias circuit of claim 13, wherein a drain of the first transistor couples to a source of a diode-connected fourth transistor, a drain of the fourth transistor coupling to a power supply voltage node.

15. A transconductance-capacitance (Gm-C) filter, comprising a plurality of operational transconductance amplifiers (OTAs), wherein a first one of the OTAs has a first transconductance and the remaining ones of the OTAs have transconductances that are proportional to the first transconductance, anda bias circuit for biasing the first transconductance to a desired value responsive to a clock frequency, the bias circuit including a switched capacitor circuit generating a resistance inversely proportional to the clock frequency, wherein the desired transconductance value is proportional to the clock frequency.

16. The Gm-C filter of claim 15, wherein the first OTA includes a first transistor having a gate driven by a bias voltage produced by the bias circuit.

17. The Gm-C filter of claim 15, wherein the Gm-C filter is a biquad filter.

18. The Gm-C filter of claim 16, wherein bias circuit includes a second transistor having a width-to-length ratio that is a factor X times larger than a width-to-length ratio of the first transistor, and wherein the desired transconductance value is proportional to a factor of (1−1/X1/2).