SWITCHED CAPACITOR BIASING CIRCUIT

Abstract

Bias circuit and a bias generator circuit comprising such a bias circuit. The bias circuit (10, 11) comprises a switched capacitor resistor circuitry (C1, C2, M12-M17), and an operational amplifier (M1-M4, M10) with an input differential transistor pair (M1, M2). The bias circuit further comprises additional source follower transistors (M5, M6) associated with the first and second input differential transistors (M1, M2). The bias generator circuit has a PMOS switched capacitor reference circuit (11) and a NMOS switched capacitor reference circuit (10) and a transconductor reference cell (15). The transconductor reference cell (15) is a replica of a basic reference cell used in a further circuit.

Description

FIELD OF THE INVENTION

The present invention relates to a bias circuit comprising a switched capacitor resistor circuitry, and an operational amplifier. In a further aspect, the present invention relates to a biasing generator circuit.

PRIOR ART

American patent publication U.S. Pat. No. 6,407,623 discloses a bias circuit for maintaining a constant value of a ratio of transconductance divided by load capacitance. A switched-capacitor biasing circuit is used to bias an operational transconductance amplifier to maintain a constant gm/C ratio over process and temperature variations. A fixed bandwidth is thereby obtained for the operational transconductance amplifier.

In European patent publication EP-A-0 766 385 a sampled data-biasing of continuous time integrated circuit scheme is disclosed. The gm/C ratio is made constant through replacing an external accurate resistor by a switched-capacitor circuit to compensate for process and temperature variations of an on-chip filter circuit.

SUMMARY OF THE INVENTION

The present invention seeks to provide an improved biasing circuit, specifically suitable for process and temperature stabilization of integrated circuits.

According to a first aspect of the present invention, a bias circuit is provided comprising a switched capacitor resistor circuitry, an operational amplifier with an input differential transistor pair, and an integrating capacitor connected between an output and an input of the operational amplifier, wherein an output of the switched capacitor resistor circuitry is connected to an input of the operational amplifier, a first input differential transistor being N times larger than a second input differential transistor, and the bias circuit further comprising additional source follower transistors associated with the first and second input differential transistors. The additional source follower transistors improve the compensation behavior of the bias circuit.

In a further aspect of the present invention, a biasing generator circuit is provided, comprising a PMOS switched capacitor reference circuit and a NMOS switched capacitor reference circuit, of which respective outputs are connected to a combiner element, and a transconductor reference cell receiving an output of the combiner element, wherein the transconductor reference cell is a replica of a basic reference cell used in a further circuit, and the biasing generator circuit is arranged to provide a bias output to the further circuit. As a result, the characteristics of the biasing generator circuit used to provide a constant transconductance to load capacitance ratio are improved.

SHORT DESCRIPTION OF DRAWINGS

The present invention will be discussed in more detail below, using a number of exemplary embodiments, with reference to the attached drawings, in which

FIG. 1 shows a schematic diagram of an embodiment of the present invention using generic blocks;

FIG. 2 shows a schematic diagram of a further embodiment of the present invention;

FIG. 3 shows a schematic diagram of a biasing generator circuit according to a further aspect of the present invention;

FIG. 4 shows a circuit diagram of a PMOS based bias generator; and

FIG. 5 shows a circuit diagram of a NMOS based bias generator.

DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS

The design and fabrication of signal processing integrated circuits is limited by device characteristics which cannot be precisely controlled and have chip to chip and wafer to wafer variations. Integrated circuit designers can take the advantage of matching multiple-like devices, then the circuits become independent of the absolute device characteristics.

Usually on-chip filters require a self-tuning or automatic-tuning circuitry. E.g. the bandwidth and corner frequency of on-chip gm-C filters are dependent on their devices gm/C ratio. Transconductance and capacitor values change ±20% over process and temperature extremes. Therefore on-chip compensation of filters is necessary. This necessitates additional tuning hardware to be implemented on the chip. An oscillator circuit is built with the filter transconductance components and the output frequency of the oscillator is compared with the stable crystal oscillator frequency and necessary calibrations are done with the results of this comparison.

Prior art solutions often require additional silicon area cost and circuit complexity penalty. E.g. in international patent publication WO2012/165941 a complementary constant-gm biasing circuitry is used to compensate the temperature variations within the chip. This approach still needs master-slave tuning with a reference oscillator to correct the center frequency of the filter over process variations.

In European patent publication EP-A-0 766 385 a sampled data-biasing of continuous time integrated circuit scheme is disclosed. The gm/C ratio is made constant through replacing an external accurate resistor by a switched-capacitor circuit to compensate for process and temperature variations of an on-chip filter circuit.

In a first aspect, the present invention aims to improve the characteristics of a biasing generator circuit used to provide a constant transconductance to load capacitance ratio. To that effect a biasing generator circuit is provided comprising a PMOS switched capacitor reference circuit 11 and a NMOS switched capacitor reference circuit 10, of which respective outputs are connected to a combiner element 12, and a transconductor reference cell 15 receiving an output of the combiner element 12. The transconductor reference cell 15 is a replica of a basic reference cell used in a further circuit, and the biasing generator circuit is arranged to provide a bias output to the further circuit.

FIG. 1 shows a schematic diagram of an embodiment of the present invention using generic blocks. NMOS and PMOS switched capacitor reference circuits 10, 11 generate proper biasing for the transconductance reference cell (or replica circuit) 15, using a combiner element 12. In this diagram, the output of the biasing generator circuit is the output Vref of the transconductance reference cell 15 (labelled replica circuit in FIG. 1). A clock signal clk is used to drive the switched capacitor circuitry. FIG. 2 illustrates how the disclosed invention can be used to properly bias a gm-C filter 17 to create an automatically tuned filter with low area and current consumption cost. The output Vref of the transconductance reference cell 15 (labelled Complementary gm-C cell in FIG. 2) is input to a regulator 16, the output of which is used to provide the bias for a gm-C filter 17 (using the Vdd auto-tune voltage). Thus, in an embodiment of the present invention, the further circuit is a continuous time gm-C filter circuit 17 (in combination with the regulator 16). As an alternative embodiment, the further circuit is a ring oscillator circuit, wherein the transconductance reference cell 15 is properly embodied to reflect the basic reference cell used in the ring oscillator.

With the present invention embodiments, no additional hardware is required such as a replica oscillator circuit, nor the production calibration. A further circuit 17 such as a filter is calibrated real-time with switched-capacitor biasing circuitry 10-12, 15 that uses the same capacitor type used inside the gm-C filter and a stable crystal oscillator clock input (clk). Transconductances inside the filter 17 are real time tuned to have a constant gm/C ratio over process and temperature changes. The biasing circuitry 10-12, 15 can be used with NMOS only, PMOS only or complementary MOS based further circuits, such as gm-C filter circuits.

A detailed implementation of an embodiment of the present invention is illustrated in FIG. 3 as a switched capacitor based biasing generator circuit. The biasing generator circuit comprises a NMOS switched capacitor reference circuit 10 and a PMOS switched capacitor reference circuit 11, which are described in more detail below with reference to FIG. 4 and FIG. 5. The biasing generator circuit is designed to stabilize the center frequency and bandwidth of an integrated gm-C bandpass filter which can be supplied by the biasing generator circuit of FIG. 3.

The biasing generator circuit uses the same capacitances C1 that is used in the further circuit connectable to the biasing generator circuit (in each of the NMOS switched capacitor reference circuit 10 and PMOS switched capacitor reference circuit 11). It also uses a transconductor reference cell 15 as a bias voltage reference, which transconductor reference cell 15 is based on the same technology as that used in the further circuit (i.e. a replica circuit). A crystal oscillator clock (not shown) is used to drive the switches with generated non-overlapping clocks Φ₁ and Φ₂. A generated reference voltage (V_BPF_BIAS) is applied to the further circuit through a low drop out (LDO) regulator circuit 18. In other words, the biasing generator circuit further comprises a low-drop out (LDO) regulator circuit 18 connected to an output of the transconductor reference cell 15 in a further embodiment. This allows a lower minimum operating voltage, higher efficiency operation and lower heat dissipation. Both stabilizing currents I_p/2 and I_n/2 generated in the output transistors MC1 and MC2 (as shown in FIG. 3) are applied to the transconductor reference cell 15 comprising reference cell transistors MR1, MR2) to compensate the transconductance (gm) and threshold (Vth) values over process and temperature.

In summary, the present invention embodiments provide for a switched capacitor reference circuit, which comprises of a NMOS based circuit 10 and a PMOS based circuit 11 providing current to a replica circuit 15 that needs to be stabilized over corners. By using NMOS and PMOS circuits 10, 11 with optimum ratio of output currents I_p/2 and I_n/2, a suitable reference bias is generated to compensate for process and temperature variations of the replica circuit 15. Using this generated replica circuit bias to supply a further circuit 17, e.g. a continuous-time filter (BPF) circuit, characteristics such as center frequency and bandwidth can be stabilized over process and temperature. This eliminates the area associated with additional hardware as needed in prior art solutions to stabilise the filter characteristics.

In this disclosure, the switched-capacitor based biasing circuitry includes NMOS and PMOS parts 10, 11, non-overlapping clock generator to drive the NMOS and PMOS part 10, 11 using a clock signal clk, and an LDO (low-dropout regulator) circuit 18. It also uses a crystal oscillator as a stable clock reference input and generates a voltage biasing output Vref to the further circuit 17 to be stabilized such as a gm-C based bandpass filter circuit. Transconductances in the further circuit 17 are tuned so that the gm/C ratio is constant, which provides reproducible filter characteristics. The advantages of the present invention embodiments are that a simple and cost-effective solution is given, using two switched-cap current reference circuits 10, 11 based on NMOS and PMOS, of which both currents I_p/2 and I_n/2 with an optimum ratio are fed into a replica circuit 15 of a further circuit 17 that needs to be stabilized over corners, in order to generate a suitable reference. No additional hardware required such as a replica oscillator circuit, and no production calibration is needed. The present invention embodiments require only limited additional chip area and power consumption.

A further aspect of the present invention relates to improvements in bias circuits which can be used as the PMOS and NMOS switched capacitor reference circuits 10, 11 of the biasing generator circuit described above. Implementations of the PMOS and NMOS switched capacitor reference circuits 10, 11 are shown in the circuit diagrams of FIGS. 4 and 5, respectively.

In general terms, a bias circuit is provided comprising a switched capacitor resistor circuitry (capacitors C1, C2, and transistors/switches M12-M17), an operational amplifier (transistors M1-M4, M10) with an input differential transistor pair M1, M2, and an integrating capacitor C3 connected between an output and an input of the operational amplifier M1-M4, M10, wherein an output of the switched capacitor resistor circuitry C1, C2, M12-M17 is connected to an input of the operational amplifier M1-M4, M10 (indicated as Vref_p and Vref_n). A first input differential transistor M1 is N times larger than a second input differential transistor M2, to ensure an offset voltage is generated at the input of the operational amplifier, while the currents are equal. The bias circuit further comprises additional source follower transistors M5, M6 associated with input differential transistors M1, M2 of the operational amplifier M1-M4, M10. Further additional transistors may be provided as shown in the embodiments of FIGS. 4 and 5 (transistor M11 in series connection with a first additional source follower transistor M5, and transistor M9 in series connection with a second additional source follower transistor M6). Compensation behavior of the bias circuit is improved as a result, as will be demonstrated below.

The circuit shown in FIG. 4 is a practical way to generate a current equivalent to that produced with a switched-capacitor resistor circuitry. The operation is as follows. There are two non-overlapping phases Φ₁ and Φ₂. During Φ₂, the capacitor C₁ is discharged completely. At the same time, C₂'s voltage is forced to V_REF by the virtual ground of the operational amplifier M1-M4, M10. During Φ₁, the output of the operational amplifier provides a constant voltage to the gate of transistor M7, which, in turn, provides a constant current to C₁ and C₂ via the mirror transistors M12 and M8. Accordingly, the voltage across and ramps linearly, and at the end of Φ₁, its value is

$V\left(C_2\right)=\frac{C_2V_{\mathrm{REF}}+I·\Delta T_1}{C_1+C_2}$

where ΔT₁ is the duration of Φ₁, C₂ is then discharged into the integrating capacitor C₃ during Φ₂ at the same time C₁ is discharged to ground. If the voltage across C₂ (V(C₂)) is larger than V_REF, the output of the operational amplifier (drain of transistor M1) will decrease when C₂ is discharged into the virtual ground. This will cause a decrease in current for the next cycle. Similarly, if V(C₂) is less than V_REF, the current will be increased for the next cycle.

Thus, if the loop is stable, the steady-state value of V(C₂) is equal to V_REF, and the steady-state current will be

$I=\frac{V_{\mathrm{REF}}C_1}{\Delta T_1}=2V_{\mathrm{REF}}C_1F_{\mathrm{CLK}}$

where a 50% duty cycle is assumed for simplicity (i.e., 2 ΔT₁=1/F_CLK).

Transistors M10 and M1-M4 thus act as an integrating operational amplifier that drives the gate of M7. The circuit generates its own reference because M1 and M2 are biased so that they will have the same current, but is M1 is N times as wide as M2. Therefore, the voltage difference across their gates will be (assuming strong inversion operation)

$V_{\mathrm{REF}}=V_{\mathrm{GS}1}-V_{\mathrm{GS}2}=\sqrt{\frac{2\mathrm{IL}}{\mu C_{\mathrm{ox}}W}}\left(1-\frac{1}{\sqrt{N}}\right)$

A MOS differential pair transistor of dimensions W_diff/L_diff biased with a current equal to the current in M2 will have a transconductance

$g_m=4F_{\mathrm{CLK}}C_1\sqrt{\frac{L}{W}\frac{W_{\mathrm{diff}}}{L_{\mathrm{diff}}}}\left(1-\frac{1}{\sqrt{N}}\right)$

where W and L are the dimensions of M2.

Finally, the bandwidth of an amplifier built with this differential pair and compensated with a capacitance C4 will be

$\frac{g_m}{C_4}=4F_{\mathrm{CLK}}\frac{C_1}{C_4}\sqrt{\frac{L}{W}\frac{W_{\mathrm{diff}}}{L_{\mathrm{diff}}}}\left(1-\frac{1}{\sqrt{N}}\right)$

which is independent of process and temperature.

It is noted that in an embodiment the bias circuit may comprise a stabilizing capacitor C4 connected between a drain of a first input differential transistor (M1) of the operational amplifier (M1-M4, M10) and a ground. Alternatively, a gate-source capacitance of the output transistor M7 may be used for this purpose.

A similar derivation assuming weak inversion for M1 and M2 as well as the operational amplifier differential pair M3, M4 results in

$\frac{g_m}{C_4}=2F_{\mathrm{CLK}}\frac{C_1}{C_4}\ln \left(N\right)$

which is also independent of process and temperature.

For best performance, the input differential pair of the further circuit 17 should not only be the same type as M1 and M2, but it should operate at a current density equal to the geometric mean of the current densities of M1 and M2. If this is done, then a constant—characteristic will occur regardless of the bias region (sub-threshold, strong inversion, etc) of the differential pair. As a practical matter, good results are obtained even if the bias points are not identical.

Of course there are countless ways of using feedback and transistors biased at different current densities to produce this same type of result. One advantage of the topology in FIG. 4 is that the differential pair can be biased to mimic the differential pair of the operational amplifier which it is to bias. In addition, small switches M13-M17 can be used in the switched capacitor resistor circuitry because most of the supply voltage can be used to drive them since one side of the switch is always at (or close to) the negative supply rail.

The circuitry as shown with reference to FIG. 4 can be used as a PMOS bias generator. The circuitry as shown with reference to FIG. 5 can be used as a NMOS bias generator, and further comprises additional NMOS compensating circuitry (M18-M21) associated with the additional source follower transistors (M5, M6).

As already mentioned above, an input differential pair of transistors of an integrator (C3, M1-M4) is dimensioned with a first input transistor M1 being N times larger than a second input transistor M2, so that there will be an offset voltage generated at the input of the operational amplifier M1-M4 which is related to V_ref=V_GS1−V_GS2 while their currents are equal to each other. M1 and M2 input differential transistors current densities are made equal to the reference transconductor cell's transistors MR1, MR2 (see FIG. 3 and description above). For the PMOS part 11, output current can be defined as:

$I_p=2{\mathrm{Vref}}_pC_1F_{\mathrm{CLK}}$ ${\mathrm{Vref}}_p=\sqrt{\frac{2I_p}{K_p}}\left(1-\frac{1}{\sqrt{N}}\right)$ $\sqrt{I_p}=2\sqrt{\frac{2}{K_p}}\left(1-\frac{1}{\sqrt{N}}\right)C_1F_{\mathrm{CLK}}$

Similarly, for the NMOS part 10, the current is:

$\sqrt{I_n}=2\sqrt{\frac{2}{K_n}}\left(1-\frac{1}{\sqrt{N}}\right)C_1F_{\mathrm{CLK}}$

where K_p and K_n are defined as below:

$K_p={\mu }_pC_{\mathrm{ox}}\frac{W_p}{L_p}$ $K_n={\mu }_nC_{\mathrm{ox}}\frac{W_n}{L_n}$

So the total effective gm of the reference transconductor cell 15 (and thus of the further circuit 17) can be derived as below:

$g_m=\sqrt{2K_p\frac{I_p+I_n}{2}}+\sqrt{2K_n\frac{I_p+I_n}{2}}=\sqrt{K_p\left(I_p+I_n\right)}+\sqrt{K_n\left(I_p+I_n\right)}=\sqrt{K_p+K_n}\sqrt{I_p+I_n}$

If the I_p and I_n equations are entered into the above gm equation the following equation results:

$g_m=\frac{K_p+K_n}{\sqrt{K_pK_n}}2\sqrt{2}\left(1-\frac{1}{\sqrt{N}}\right)C_1F_{\mathrm{CLK}}$

And hence the gm/C ratio will become:

$\frac{g_m}{C_1}=\frac{K_p+K_n}{\sqrt{K_pK_n}}2\sqrt{2}\left(1-\frac{1}{\sqrt{N}}\right)F_{\mathrm{CLK}}$

As can be seen from the above equation the filter center frequency and the bandwidth characteristics will become independent of the capacitor absolute value and its process change. When we make K_p=K_n through dimensioning the PMOS transistor width 2.5 times larger than the NMOS transistor, their mobilities (μ_n and μ_p) and their temperature dependency will also be cancelled first orderly. In other words, the PMOS transistor MR1 has a gain factor K_p and the NMOS transistor MR2 has a gain factor K_n, wherein K_p=K_n. Transistor threshold voltage (V_th) changes over process and its temperature dependency is already fully compensated automatically. Remaining not-cancelled parts are second order mobility variations and transistor W_p, L_p, W_n, and L_n, mismatches, which are relatively small effects. It is noted that in the prior art disclosure of U.S. Pat. No. 6,407,623 only PMOS transistor characteristics are compensated over process and temperature. If this bias is used for an inverter like transconductor circuit, then all NMOS K_n and V_th parameters will appear at the g_m/C ratio, so the ratio will become process and temperature dependent.

The present invention embodiments have been described above with reference to a number of exemplary embodiments as shown in the drawings. Modifications and alternative implementations of some parts or elements are possible, and are included in the scope of protection as defined in the appended claims.

Claims

1. Biasing generator circuit comprising: a PMOS switched capacitor reference circuit and a NMOS switched capacitor reference circuit, of which respective outputs are connected to a combiner element, and a transconductor reference cell receiving an output of the combiner element,wherein the transconductor reference cell is a replica of a basic reference cell used in a further circuit, andthe biasing generator circuit is arranged to provide a bias output to the further circuit.

2. Biasing generator circuit according to claim 1, wherein the transconductor reference cell comprises a PMOS transistor and a NMOS transistor, wherein the PMOS transistor has a gain factor Kp and the NMOS transistor has a gain factor Kn, wherein Kp=Kn.

3. Biasing generator circuit according to claim 1, further comprising a low-drop out (LDO) regulator circuit connected to an output of the transconductor reference cell.

4. Biasing generator circuit according to claim 1, wherein the further circuit is a continuous time gm-C filter circuit.

5. Biasing generator circuit according to claim 1, wherein the further circuit is a ring oscillator circuit.

6. Biasing generator circuit according to claim 2, further comprising a low-drop out (LDO) regulator circuit connected to an output of the transconductor reference cell.

7. Biasing generator circuit according to claim 6, wherein the further circuit is a continuous time gm-C filter circuit.

8. Biasing generator circuit according to claim 6, wherein the further circuit is a ring oscillator circuit.