The field generally relates to differential amplifier circuits and, in particular, adaptive control mechanisms for controlling input and output common-mode voltages of differential amplifier circuits.
In many analog and mixed-signal electronic systems, differential signaling is used to electrically transmit information using two complementary voltage signals that are transmitted on two paired wires, referred to as a differential pair. With this method, the information transmitted is represented by the difference between the two complementary voltage signals. In these systems, a differential amplifier is used to amplify two complementary voltages transmitted on a differential pair. A differential amplifier includes inverting and non-inverting input terminals connected to a differential pair, and amplifies a difference between input voltage signals applied to the inverting and non-inverting input terminals, while rejecting an input common-mode voltage level. An input common-mode voltage is defined as the average voltage at the inverting and non-inverting input terminals of the differential amplifier. To maintain high gain and linearity of differential amplifier circuits, it is desirable to limit variations of both the input and output common-mode voltages of the differential input and output signals, which can occur across supply voltage variations, temperature variations, and global and local processing variations. The need to control the input and output common-mode voltage of differential signals is becoming increasingly critical as lower supply voltages are being utilized to bias differential amplifier circuits fabricated in new silicon technologies. Moreover, random device mismatches in a differential amplifier circuit can result in large variations in the offset and in the common-mode of the output signals, and adversely affect circuit performance.
In one embodiment of the invention, an amplifier circuit includes differential input nodes having first and second differential input nodes, a differential amplifier stage having differential input terminals and differential output terminals, and an input common-mode voltage adaptation circuit connected between the differential input nodes of the amplifier circuit and the differential input terminals of the differential amplifier stage. During an input common-mode adaptation phase, the input common-mode voltage adaptation circuit forces the differential input terminals of the differential amplifier stage to have a common-mode voltage equal to an adaptive reference voltage, independent of a common-mode voltage applied to the differential input nodes of the amplifier circuit during the input common-mode adaptation phase. During a normal period of operation of the amplifier circuit, the input common-mode voltage adaptation circuit maintains the common-mode voltage at the differential input terminals of the differential amplifier stage equal to the adaptive reference voltage, independent of an input differential voltage applied during the normal period of operation.
Other embodiments of the invention will become apparent.
The differential amplifier stage 110 comprises a differential transistor pair formed by transistors M1 and M2. In one embodiment, the transistors M1 and M2 are NMOS transistors having gate terminals connected to respective nodes N1 and N2 (first and second output nodes of the input common-mode voltage adaptation circuit 120), source terminals connected together, and drain terminals connected to respective first and second differential output nodes Voutp and Voutm. In one embodiment, the transistors M3 and M4 are PMOS transistors that serve as active load devices for the differential transistor pair M1 and M2, wherein transistor M3 is connected between the first power supply node P1 and the first differential output node Voutp, and wherein transistor M4 is connected between the first power supply node P1 and the second differential output node Voutm. In one embodiment, the transistor M5 is an NMOS tail transistor that is connected to the source terminals of transistors M1 and M2 and the second power supply node P2. In another embodiment of the invention, the PMOS and NMOS devices of the differential amplifier stage 110 are cascoded to improve output impedance.
The differential amplifier stage 110 differentially amplifies (subtracts and multiplies) voltages at nodes N1 and N2 (i.e., VN1−VN2) which are applied to the gate terminals of transistors M1 and M2, respectively. An input common-mode voltage VICM of the differential amplifier stage 110 is
In general, in one embodiment of the invention, the input common-mode voltage adaptation circuit 120 adaptively controls an input common-mode voltage that is applied to the differential inputs of the differential amplifier stage 110. Without the input common-mode voltage adaptation circuit 120, the input common-mode voltage at the input of the differential amplifier stage 110 would be
which could vary significantly depending on various factors such as, e.g., an output common-mode voltage of an upstream circuit, as well as supply voltage variations, temperature variations, and global and local processing variations.
A differential amplifier circuit it typically designed to operate over a defined range of input common-mode voltages, which depends on the supply rail voltages used in the circuit design. If the input common-mode voltage falls outside this range, the differential amplifier circuit will not operate properly. In fact, there is an optimal common-mode voltage at the gate terminals of the input differential transistor pair that maximizes the performance of the amplifier circuit. This optimal common-mode voltage is not a universal constant. Rather, the optimal voltage varies with conditions of temperature, supply voltage, and manufacturing process. The input common-mode voltage adaptation circuit 120 adapts to changing conditions to always generate a near optimal input common-mode voltage, even as the desired optimal voltage changes with the conditions. As explained in further detail below, in one mode of operation, the input common-mode voltage adaptation circuit 120 adaptively controls the input common-mode voltage at the gate terminals of transistors M1 and M2 independently of the input common-mode voltage of a differential signal received from an upstream circuit, to ensure optimal operation of the differential amplifier circuit 100 across supply voltage variations, temperature variations, and global and local processing variations.
The output common-mode feedback control circuit 140 has inputs connected to the first and second differential output nodes Voutp and Voutm of the differential amplifier circuit 100, and an output that drives a gate terminal of the transistor M5 of the differential amplifier stage 110. In general, the output common-mode feedback control circuit 140 operates by controlling the output common-mode voltage
of the first and second differential output nodes of the differential amplifier circuit 100. In one embodiment of the invention, the output common-mode feedback control circuit 140 operates by driving the gate terminal of the tail transistor M5 with a voltage that is adjusted over time to maintain the output common-mode voltage at the target level of a common-mode reference voltage Vcm_ref that is input to the output common-mode feedback control circuit 140.
Even when the input common-mode voltage is sufficiently controlled, the output common-mode voltage of the differential amplifier stage 110 can significantly vary over supply voltage variations, temperature variations, and global and local processing variations, which can adversely affect operation of downstream circuitry connected to the output of the differential amplifier circuit 100. In this regard, in one embodiment of the invention, the output common-mode feedback control circuit 140 is implemented to adaptively control the output common-mode voltage. The output common-mode feedback control circuit 140 may be implemented using feedback control circuits and frameworks known to those of ordinary skill in the art. In one embodiment of the invention, the output common-mode feedback control circuit 140 is implemented using the output common-mode feedback control circuit described below with reference to
Furthermore, as explained below, in another mode of operation, the input common-mode voltage adaptation circuit 120 operates in conjunction with the output common-mode feedback control circuit 140 during an adaptation phase to slightly adjust the input voltages VN1 and VN2 at the gate terminals of the differential transistor pair M1 and M2 to reduce or otherwise eliminate any offset voltage and accommodate any residual mismatch between the output common-mode voltage and the reference voltage Vcm_ref, while maintaining the common-mode voltage at the gates of the differential transistor pair M1 and M2 equal to the adaptive reference voltage, VN4. The circuit architecture and operational mode of the input common-mode voltage adaptation circuit 120 of
As shown in
As further shown in the embodiment of
In one embodiment of the invention, the input common-mode voltage adaptation circuit 120 is operated by a plurality of control signals, ph1 and ph2, which are generated by a clock circuit to switchably control switches S1, S2, S3, S4, S5, and S6. More specifically, in the embodiment of
During an adaption phase, the upstream circuitry, which is connected to the first and second differential input nodes Vinp and Vinm the differential amplifier circuit 100, is controlled to output a differential signal where the voltages on the first and second differential input nodes Vinp and Vinm are the same. In this regard, during an adaptation phase, the differential voltage Vinp−Vinm=0, and the input common-mode voltage VICM=Vinp=Vinm. During an adaptation phase, the input common-mode voltage adaptation circuit 120 generally operates in response to sequential assertion of control pulses ph1 and ph2 to force the input common-mode voltage at the input to the differential amplifier stage 110 to be equal to a target adaptive reference voltage VN4 on node N4, which is an adaptive cascode bias voltage generated by the diode-connected NMOS transistor M7.
After completion of an adaption phase, a “normal operating period” ensues in which neither control pulse ph1 and ph2 is asserted and a differential signal froth the upstream circuit is applied to the differential input nodes Vinm and Vinp of the differential amplifier circuit 100. In
At the start of a first portion of an adaptation phase (e.g., time t0), a control pulse ph1 is applied to activate (close) switches S1, S2, S5 and S6, while switches S3 and S4 are deactivated (open). When switches S1, S2, S5, and S6 are activated (closed), the common-mode reference voltage Vcm_ref is applied to the feedback nodes FN1 and FN2, and the output nodes N1 and N2 are connected to the reference voltage node N4. In this regard, during the first portion of the adaptation phase (as control signal ph1 remains asserted), an adaptive cascode bias voltage on node N4, which is generated by the diode-connected NMOS transistor M7, is applied to the gate terminals of differential transistor pair M1 and M2. The adaptive cascode bias voltage (adaptive reference voltage VN4) on node N4 is an optimal input common-mode voltage that is applied to the gate terminals of the different transistor pair M1 and M2. In this state, since the adaptive reference voltage VN4 is applied to both gate terminals of the differential transistor pair M1 and M2 (i.e., VN1 and VN2=VN4), a zero (0) differential voltage is applied at the input of the differential amplifier stage 110.
Furthermore, during the first portion of the adaptation phase, each of the first and second capacitors C1 and C2 are precharged to a steady state voltage level of Vcm_Ref−VN4. In addition, the third capacitor C3 is precharged to a steady state voltage level of Vinp−VN4, and the fourth capacitor C4 is precharged to a steady stage voltage level of Vinm−VN4.
The term “cascode bias voltage” refers to the fact that (a) the voltage on node N4 is independent of the input signals, Vinp and Vinm, and that (b) the sizes of transistors M6 and M7 are designed such that the voltage on node N4 is large enough to cause the voltage on the drain terminal of transistor M5 to be large enough to keep transistor M5 in the saturation region of operation. The term “adaptive” refers to the fact that the adaptive reference voltage on node N4 adapts with variations in voltage, temperature, and processing conditions. Due to matching of transistors M7, M1, and M2, and matching of transistors M6, M3, and M4, by design, the variations in the adaptive reference voltage VN4 are correlated with the variations in the differential amplifier circuit. As a result, the adaptive reference voltage VN4 is maintained at a near-optimal voltage for maximizing the differential circuit performance. In one embodiment of the invention, the transistors M7, M1, and M2 are matched in polarity (e.g., NMOS transistors) and sized accordingly (e.g., the width-to-length ratio of transistor M7 is ¼ the width-to-length ratio of transistors M1 and M2), and the transistors M6, M3, and M4 are matched in polarity (e.g. PMOS transistors) and sized appropriately (e.g. the width and length of transistor M6 are equal to the widths and lengths, respectively, of transistors M3 and M4) to provide a target adaptive cascade bias voltage.
In this regard, since transistor M7 is matched to transistors M1 and M2 by design, any affect that variation of temperature or processing has on transistor M7 will be matched to transistors M1 and M2. Similarly, any effect that variation of temperature or processing has on transistor M6 will be matched to transistors M3 and M4. While the adaptive reference voltage VN4 is fixed at an optimal level for a given adaptation phase, the adaptive reference voltage VN4 can change over time in subsequent adaptation phases to adapt to variations in process, voltage or temperature conditions. In practice, since supply voltage and temperature drift relatively slowly over normal operating periods of the circuit, the common-mode of the inputs of the differential transistor pair M1 and M2 can be fixed between successive adaption phases without adversely affecting circuit performance and the duration of the normal operating phase can be large relative to the duration of the adaptation phases, ph1 and ph2.
Next, at the start of the second portion of the adaptation phase (e.g., time t1), the control pulse ph1 is de-asserted to deactivate (open) switches S1, S2, S5 and S6, and a control pulse ph2 is asserted to activate (close) switches S3 and S4. With switches S1, S2, S5, and S6 deactivated, the common mode reference voltage Vcm_ref at node N3 is disconnected from the feedback nodes FN1 and FN2, and the adaptive reference voltage VN4 on node N4 is disconnected from nodes N1 and N2. Moreover, with switches S3 and S4 activated (closed), the feedback nodes FN1 and FN2 are connected to the second and first differential output nodes Voutm and Voutp, respectively, of the differential amplifier circuit 100. In this instance, the capacitors C1 and C2 are connected across the drain and gate terminals of respective differential input transistors M2 and M1, respectively, thereby providing negative feedback by capacitive coupling of the output (drain) terminals to the input (gate) terminal.
In this state, during the second portion of the adaptation phase, the input common-mode voltage adaptation circuit 120 operates to slightly adjust the voltages VN1 and VN2 at the inputs of the differential amplifier stage 110 to generate a small differential input voltage to perform offset cancellation, while preserving the average input adaptive common-mode voltage applied to the inputs of the differential amplifier stage 110. At the end of the first portion of the adaptation phase, the voltages VN1=VN2=VN4 such that a zero (0) differential voltage is applied to the inputs of the differential amplifier stage 110. In an ideal state where it is assumed that there is no mismatch between transistors M1 and M2, and no mismatch between load transistors M3 and M4, the differential amplifier stage 110 would generate a zero (0) differential output signal with Voutp=Voutm (i.e., Voutp−Voutm=0) when a zero differential input signal is applied to the input terminals of the differential transistor pair M1 and M2. However, in practice, a slight mismatch between transistors M1 and M2 and/or a slight mismatch between transistors M3 and M4 would result in an “offset voltage” at the differential output nodes Voutp and Voutm. As is known in the art, the term “offset voltage” is a non-zero differential output voltage that is generated by the differential amplifier stage 110 when the differential input voltage is zero (0). While the output common-mode feedback control circuit 140 operates during the second portion of the adaptation phase to force the output common-mode voltage
to be equal to the common-mode reference voltage Vcm_ref, the input common-mode voltage adaptation circuit 120 operates to reduce or otherwise eliminate any offset voltage at the output of the differential amplifier stage 110.
In particular, during the second portion of the adaptation phase, the output common-mode feedback control circuit 140 drives the gate of the tail transistor M5 to adjust the output common-mode voltage
to be equal to the common-mode reference voltage Vcm_ref As the differential output voltages Voutp and Voutm are adjusted to fix the output common-mode voltage at Vcm_ref the output differential voltage Voutp−Voutm is applied to the feedback nodes FN1 and FN2 by virtue of the negative feedback resulting from the activation of switches S3 and S4. The voltages on the feedback nodes FN1 and FN2 will slightly adjust (increase and decrease) from Vcm_ref to match the difference between the differential output voltages Voutp and Voutm. This slight offset voltage at the feedback nodes FN1 and FN2 will couple through the capacitors C1 and C2 to the gate terminals of transistor pair M1 and M2, thereby creating a small offset between the gate voltages VN1 and VN2, which were both initially set at the adaptive reference voltage VN4 at the end of the first portion of the adaptation phase. The negative feedback connection drives the differential voltage at the output to be equal to the differential voltage at the input of the differential amplifier stage (with negative polarity) and the high-gain of the differential amplifier stage then causes both differential voltages to be driven to almost zero. While the gate voltages VN1 and VN2 may slightly increase and decrease from the initial value of the adaptive reference voltage VN4 to achieve offset voltage cancellation, the input common-mode voltage adaptation circuit 120 operates to maintain the input common-mode voltage
at the input of the differential transistor pair M1 and M2 equal to the adaptive reference voltage VN4.
Next, at the end of the second portion of the adaptation phase (e.g., time t3), the control pulse ph2 is de-asserted to deactivate (open) switches S3 and S4, thereby disconnecting the capacitors C1 and C2 from the differential output nodes Voutp and Voutm in the feedback configuration. When both ph1 and ph2 are disabled, the differential amplifier circuit 100 is available for use and enters into a normal operating period, with the output offset error largely eliminated and the differential amplifier stage 110 adaptively biased for optimal performance. In particular, at the start of the normal operating period (e.g., time t2), upstream circuitry connected to the differential amplifier circuit 100 will generate and output a differential voltage that drives the differential input nodes Vinp and Vinm of the differential amplifier circuit 100.
During the normal operating period, a voltage stored across the capacitors C3 and C4 act as a level-shifter circuit to level-shift the input common-mode voltage at the input nodes Vinp and Vinm (either up or down) to the level of the adaptive input common-mode voltage level at the nodes N1 and N2. In this regard, when a differential signal is applied to the input nodes Vinp and Vinm, the differential signal VN1−VN2 applied to the input terminals of the differential amplifier stage 110 will have an input common-mode voltage fixed at the level of the adaptive reference voltage VN4 generated during the previous adaptation phase. As such, the embodiment of
Moreover, as noted above, the input common-mode voltage at the input of the differential transistor pair M1 and M2 adapts to the process parameters of the individual instance of the circuit and adapts periodically to changing conditions of power supply voltage and temperature. This adaptation maintains optimal biasing for the differential amplifier stage 110. As shown in
In the embodiment of
In this embodiment, the effective capacitance series capacitors C3/C1 between nodes Vinp and N2 is less than the capacitance C3 alone (as in the embodiment of
In an embodiment where the values of the capacitors C1, C2, C3, and C4 are on the order of about 0.1 pf, there can be significant attenuation of the input signal, which reduces the gain of the differential amplifier stage 110. However, as explained above, the embodiment of
In the embodiment of
The output common-mode feedback control circuit 440 shown in
The switch S10 is connected between the reference voltage node Vcm_ref and a first feedback node FN10. The switch S11 is connected between the reference voltage node Vcm_ref and a second feedback node FN11. The switch S12 is connected between the first feedback node FN10 and the first differential output node Voutp (which is the first input node of the output common-mode feedback control circuit 440). The switch S13 is connected between the second feedback node FN11 and the second differential output node Voutm (which is the second input node of the output common-mode feedback control circuit 440). The switch S14 is connected between the first differential output node Voutp and the output node FN12. The switch S15 is connected between the second differential output node Voutm and the output node FN12. A first capacitor C10 is connected between the first feedback node FN10 and the output node FN12, and a second capacitor C11 is connected between the second feedback node FN11 and the output node FN12. A gate terminal of the tail transistor M5 is connected to the output node FN12.
The output common-mode feedback control circuit 440 generates a control voltage Vdio on the output node FN12 to drive the tail transistor M5 of the differential amplifier stage 110 in order to adjust an output common-mode voltage
to be equal to the reference voltage Vcm_ref. In one embodiment of the invention, the output common-mode feedback control circuit 440 is operated by a plurality of control signals, ck and ckb, which are generated by a clock circuit to switchably control switches S10, S11, S12, S13, S14, and S15. More specifically, in the embodiment of
As shown in
At the end of the control phase, the control signal ck is de-asserted, and the complementary control signal ckb is asserted, which begins a normal operating period. In the normal operation period, switches S10, S11, S14, and S15 are deactivated (opened) and switches S12 and S13 are activated (closed). In the normal operating period, the first capacitor C10 is connected between the first differential output node Voutp and the output node FN12 (gate terminal of transistor M5), and the second capacitor C11 is connected between the second differential output node Voutm and the output node FN12. In this state, the capacitors C10 and C11 form an averaging circuit whereby the charge of the capacitors C10 and C11 is redistributed between them, but the average charge stored in C10 and C11 does not change. The current in tail transistor M5 must be equal to the sum of the currents in PMOS transistors M3 and M4. Therefore, a small change in the gate voltage Vdio of tail transistor M5 causes a large change of the opposite polarity in the common-mode voltage of Voutp and Voutm. During the normal operating phase, the capacitive coupling of Voutp and Voutm to the gate of tail transistor M5 through capacitors C10 and C11 forms a negative feedback path and a high-gain negative feedback loop. When the control signal ckb is asserted, because of this high-gain negative feedback loop, the voltage Vdio adjusts slightly at the start of the normal operating phase until the common-mode voltage of Voutp and Voutm is equal to Vcm_ref.
While the control signal ck remains de-asserted and the control signal ckb remains asserted during the normal operating period, the gate voltage Vdio of tail transistor M5 is adjusted so that the output common-mode voltage is maintained substantially equal to the reference voltage Vcm_ref. As the voltages of the differential input signals Vinp and Vinm vary in differential and/or input common-mode voltage, the gate voltage Vdio of tail transistor M5 is continuously adjusted to maintain the output common-mode voltage equal to Vcm_ref.
The embodiment of
Furthermore, by eliminating the need to match the tail transistor M5 to a current mirror transistor, the size/area of the tail transistor M5 can be significantly reduced. This reduction in size/area of the tail transistor M5 results in reduced parasitic capacitance at the gate node (output node FN12) of the tail transistor M5. Moreover, this reduced parasitic capacitance results in increase gain of the common mode feedback loop.
Furthermore, during a control phase when the control signal ck is active, the output common-mode voltage is well controlled because the output voltages Voutp and Voutm are equal to the gate voltage Vdio. As noted above, the voltage Vdio is a well-controlled voltage because it is the diode voltage of the NMOS tail transistor M5. This is in contrast to conventional circuits in which the output common mode voltage can vary a lot before adjustment, which is not desirable for certain applications.
Moreover, since there are no matching considerations for the NMOS tail transistor M5, the tail transistor M5 can have a shorter channel length, higher transconductance and smaller device parasitic capacitance. As a result, the common-mode control accuracy and bandwidth are improved. Moreover, small capacitors can be implemented for the first and second capacitors C10 and C11. In one embodiment, the differential amplifier stage 110 is a moderate to high-speed circuit with bias currents on the order of hundreds of μ amperes. For example, assume the NMOS tail transistor M5 has a bias current of 160 μA with 80 μA flowing through each PMOS load transistors M3 and M4. With this amount of bias current, during the control phase, the effective resistance for charging the capacitors C10 and C11 (1/gm of the NMOS tail transistor M5) is relatively small. This small effective charging resistance, together with small capacitance values of capacitors C10 and C11, results in a short RC time constant for charging the capacitors C10 and C11 during the control phase. With a faster charge time for the capacitors C10 and C11, the duration of the control phase (i.e., the pulse width of the asserted control signal ck) can be short, resulting in reduced time that the differential amplifier circuit is not operating in the normal operating mode.
Embodiments of the invention as shown in
Moreover, embodiments of the invention as shown in
In one embodiment of the invention, the current mirror circuit 530 can be implemented using any suitable current mirror circuit framework, such as a single diode-connected transistor 130 as shown in
In one embodiment of the invention, the output common-mode voltage feedback control circuit 540 can be implemented using the output common-mode feedback control circuit 440 shown in
A system-on-chip having embodiments of differential amplifier circuits with input and/or output common mode adaptation and control circuitry, such as shown in
In this regard, although embodiments of the invention have been described herein with reference to the accompanying drawings, it is to be understood that embodiments of the invention are not limited to the described embodiments, and that various changes and modifications may be made by one skilled in the art resulting in other embodiments of the invention within the scope of the following claims.