Variable offset amplifier circuit

[0002] The invention is in general related to amplifier circuits and in particular to those having variable offset capability.

[0003] Amplifier circuits are used to amplify an input electrical signal to provide current and/or voltage gains or reductions. They may be used to amplify a single ended or a differential signal. In addition, when used in conjunction with an output regenerative latch stage, they can provide a digital output signal (having one of two stable states) that is an indication of a comparison between two single ended input signals or a determination of the magnitude of a differential signal. A basic component of many amplifier circuits is the differential transistor pair used as the input stage of the amplifier.

[0004] Most practical implementations of amplifier circuits suffer from manufacturing process-induced variations in the structure of the circuit elements, which cause an offset in the amplifier's operation. This offset may be explained by, for instance, considering an amplifier that is designed to amplify a differential input signal. In the ideal case, the output of the amplifier would be zero volts if the input differential signal was zero volts. However, in practice, a zero voltage differential input signal often yields a small but nevertheless non-negligible output offset voltage.

[0005] Output offset may be corrected using a wide range of techniques known as offset cancellation techniques. In one such technique, the value of the input differential signal that actually yields a zero output voltage is measured and stored, and then is subsequently subtracted from each new input signal to thus cancel the offset of the amplifier.

[0006] To help reduce the inherent offset of the amplifier, the input differential transistor pair that receives the input signal is constructed using transistor matching techniques such that the pair is said to be balanced. The transistors of a balanced differential pair are typically replicates of each other in terms of size and as such are designed to mimic each other's electrical characteristics. This also allows the amplifier to respond in the same manner to variations in the input signal that have opposite polarity but the same magnitude.

[0007] The invention is illustrated by way of example and not by way of limitation in the figures of the accompanying drawings in which like references indicate similar elements. It should be noted that references to “an” embodiment in this disclosure are not necessarily to the same embodiment, and they mean at least one.

[0008] FIG. 1 shows a circuit schematic of an embodiment of a variable offset amplifier circuit.

[0009] FIG. 2 depicts a circuit schematic of another embodiment of the variable offset amplifier circuit.

[0010] FIG. 3 illustrates a circuit schematic of an embodiment of a digitally controllable current source.

[0011] FIG. 4 shows a circuit schematic of a digitally controllable current source cell.

[0012] FIG. 5 depicts a circuit schematic of an embodiment of the variable offset amplifier that also includes an embodiment of a common mode feedback circuit.

[0013] FIG. 6 illustrates a circuit schematic of an embodiment of a digitally controllable variable offset comparator circuit.

[0014] FIG. 6B shows a flow diagram of an embodiment of an offset trimming procedure.

[0015] FIG. 7 a flow diagram of an embodiment of a process for designing the variable offset amplifier circuit.

[0016] A variable offset amplifier circuit is described. According to an embodiment, first and second differential transistor pairs are provided that are intentionally unbalanced. Each pair has first and second output nodes that are “inter-coupled”, i.e. the first output node of the first pair is coupled to the second output node of the second pair, and the second output node of the first pair is coupled to the first output node of the second pair. Variable current generators are coupled to control the respective tail currents of the differential pairs. Such an amplifier circuit allows a variable non-zero offset to be introduced, by varying the respective tail currents of the differential pairs. This variable offset capability also allows the inherent offset of the amplifier to be cancelled so that zero offset is achieved. In addition, the non-zero offset may be beneficial in certain applications such as adders; rather than requiring a separate adder circuit, the amplifier can be used to inherently add a value to an input signal by setting a non-zero offset to include the value to be added. The amplifier circuit also permits the polarity of the offset to be changed as a function of changing tail current. This means that there may be a positive or a negative offset that is permissible about a center or nominal offset value. Yet another advantage of the amplifier circuit is that relatively large offset changes may be introduced using relatively small changes in the tail current. This feature helps keep the speed of operation of the amplifier circuit as a whole from varying too much across a wide offset range.

[0017] Referring now to FIG. 1, an embodiment of the amplifier circuit is shown as implemented using p-channel metal oxide semiconductor field effect transistors (MOSFETs). This embodiment includes a first differential transistor pair 102, 104, and a second differential transistor pair 106, 108. Each pair is intentionally unbalanced. For instance, in this embodiment, the unbalanced characteristic is obtained by the transistor 102 being larger than the transistor 104 by a factor N>1 in the width of the transistor channel. Also, transistor 108 has a greater channel width than that of transistor 106, by, in this embodiment, the same factor N. To help keep manufacturing costs low, the length of the transistor channels may be the same across all transistors in the amplifier circuit.

[0018] The first and second transistor pairs may be referred to as being inter-coupled to each other because output node 120 of the first pair is coupled to output node 124 of the second pair, while output node 104 of the first pair is coupled to output node 126 of the second pair. Output nodes 120 and 126 are respectively coupled to load devices 110 and 112. These load devices may include passive and/or active circuits, depending on the application of the amplifier circuit. The load devices may alternatively represent a separate output amplification state that provides an output signal at further output nodes of that stage.

[0019] The respective tail currents of the differential pairs are controlled by variable current generators 114 and 116 as shown. In this embodiment, variable current generators 114 and 116 are variable current sources that also pass the tail currents of the respective differential pairs. Other types of variable current generators to control the tail currents may alternatively be used. See, e.g., the current sinks in FIG. 2 described below.

[0020] Continuing to refer to FIG. 1, the amplifier circuit provides output voltages V_out^band V_out^ain response to the input voltages V_in^a, V_in^b, V_in^c, and V_in^d. In an embodiment of the amplifier circuit, a difference output voltage V_out^b−V_in^ais generated in response to the difference V_in^a−V_in^band V_in^c−V_in^d. The gain of the amplifier circuit may be determined by a variety of factors, including the impedance R_loadof the load devices 110 and 112, and the transconductance g_mof each transistor in the first and second differential pairs. The provision of variable non-zero offset by an embodiment of the amplifier circuit in FIG. 1 may be explained using the following example.

[0021] Consider the situation where the input voltages are equal, i.e. V_in^a=V_in^b=V_in^c=V_in^d. Also assume that the tail currents I₁and I₂are equal and the load impedances are equal. In such a configuration, the amplifier circuit provides a nominal offset that will appear at the output as V_out^b−V_out^a=V_nominal

[0022] ={[N/(N+1)]I₁+[1/(N+1)]I₂} R_{load −{[}N/(N+1)]I₂+[1/(N+1)]I₁} R_load=0.  (1)

[0023] Next, keeping the input voltages the same, if I₁is increased and I₂is decreased both by the same amount, namely I₁+ΔI and I₂−ΔI, then V_out^achanges to the following:

[0024] V_out^a={[N/(N+1)][I₁+Δ1]+[1/(N+1)] [I₂−ΔI]} R_load={I+[(N−1)/(N+1)]ΔI}R_load  (2)

[0025] Similarly, the new value of V_out^bis given by:

[0026] V_out^b={[N/(N+1)][I₂−ΔI]+[1/(N+1)][I₁−ΔI]}R_load={I−[(N−1)/(N+1)]ΔI}R_load  (3)  (3)

[0027] Finally, the difference output voltage V_out^b−V_out^ais given by:

[0028] V_out^b−V_out^a=[2(N−1)/(N+1)]ΔI R_load  (4)

[0029] Thus, increasing I₁and decreasing I₂resulted in a decrease in the difference output voltage as given in the expression above. This decrease is the offset forced by the change in the tail currents. Now, if the tail currents are changed in the reverse direction, that is if I₁is decreased and I₂is increased by the same amount, then following an analysis similar to that above gives the following expression:

[0030] V_out^b−V_out^a=[2(N−1)/(N+1)]ΔI R_load  (5)

[0031] which is an offset in the output voltage that is opposite in polarity to that given by equation (4). Thus, this example illustrates how opposite polarity offsets may be obtained in proportion to a differential change in the tail currents.

[0032] It should be noted that although in the example above the change in tail current was exactly differential in that one tail current was increased by exactly the same amount as the other tail current was decreased, the amplifier circuit may still be useful if the tail currents are changed in unbalanced ways. Also, unequal load impedances may alternatively be provided to force a nonzero nominal offset.

[0033] Continuing to refer to FIG. 1, to improve the common mode rejection of the amplifier circuit, a common mode feedback circuit (CMFB) 130 may be coupled to adjust the variable current generators 114, 116 in response to voltages of the first and second differential pairs. In the embodiment shown, the CMFB 130 detects the voltage at the output of the variable current generators 114 and 116 and in response adjusts the tail currents of the differential pairs by adjusting the variable current generators 114, 116. The CMFB 130 does not add any significant capacitance to the inputs of the differential pairs, making the CMFB 130 compatible with applications that require low input capacitance.

[0034] Referring now to FIG. 2, what's shown is a circuit schematic of another embodiment of the variable offset amplifier, this time using n-channel MOSFETs 202, 204 for the first differential pair and n-channel MOSFETs 206, 208 for the second differential pair. The variable current generators in this embodiment are current sinks 214 and 216 which control the tail currents of the differential amplifiers, respectively. The output nodes 220 and 224 are coupled to a load 210 which in turn is coupled to a power supply node, whereas output nodes 222 and 226 are coupled to a load 212 which is also coupled to the power supply node. Thus, in contrast to the embodiment of FIG. 1, the amplifier circuit in FIG. 2 has its load devices 210, 212 referenced to a power supply node rather than to a power return (e.g. ground) node. Despite this difference, the same ability to adjust the offset is present and the example used to explain the operation of the embodiment of FIG. 1 also applies albeit with a slight difference in nomenclature to the n-channel embodiment in FIG. 2. In addition, a CFMB 230 may also be provided for the n-channel embodiment in FIG. 2, to further improve the common mode rejection of this amplifier circuit.

[0035] Referring now to FIG. 3, this figure depicts a circuit schematic of an embodiment of a digitally controllable current source that can be used in place of the variable current generator 114 or 116 of FIG. 1. This current source has a number of digitally variable current cells 302₋1, 302₋2 . . . 302₋M that are coupled in parallel to provide their individual currents which are summed to yield I_out. This output current I_outmay be the tail current of a differential transistor pair. Each individual cell current may be adjusted by varying a bias level in each cell in response to an input signal. The individual cell current may be turned on or off in response to a signal at a control input CTL as shown. This control signal may be digital in nature, that is have one of two stable states corresponding to two levels of current at each individual cell's output. The total current may thus be adjusted by setting a digital value at the control inputs of the individual cells 302.

[0036] It should be noted that the output currents provided by the individual cells 302 may be unbalanced. For instance, some of the cells may provide larger currents (for coarse granularity control of the output current) while others may provide smaller currents (for fine granularity control of the output current I_out.) Use of such digitally controllable current sources allows the offset of the amplifier circuit in FIG. 1 to be trimmed digitally, by selecting the desired offset according to a multi bit digital value.

[0037] FIG. 4 shows a circuit schematic of an embodiment of a digitally controllable current source cell 302. When the input signal at CTL is at a relatively high voltage, such that the p-channel transistor 402 is cut off, the output drive p-channel transistor 404 is biased according to an input bias signal at BIAS via p-channel transistor 408. In the other state, i.e., when the input signal at CTL drops to a relatively low voltage such that p-channel transistor 402 has its channel inverted, the gate of the p-channel transistor 404 is pulled to a relatively high voltage that is sufficient to, in this embodiment, place the transistor 404 in cutoff. Thus, two different levels of cell current I_cell, e.g. “on” and “off”, are obtained in response to the input control signal at CTL. As mentioned in the previous paragraph, the various cells may be designed to provide different levels of “on” currents so that some may be used for fine granularity control of the total current I_out(see FIG. 3) while others may be used for relatively coarse granularity control of this output current. Referring back to FIG. 4, the varying levels of “on” currents may be achieved by sizing the transistors 408 and 404 as known to those of ordinary skill in the art.

[0038] Turning now to FIG. 5, this figure depicts a circuit schematic of an embodiment of the CMFB 130 used to improve common mode rejection, in an embodiment of the amplifier circuit. This embodiment of the amplifier circuit is similar to the p-channel embodiment shown in FIG. 1 that includes p-channel transistors 102, 104 for the first differential pair and 106, 108 for the second differential pair. Variable current generators 114 and 116 are coupled to source tail currents into a common source node for each pair as shown.

[0039] The voltages at two common source nodes are provided to separate n-channel transistors 504 and 508. If the two common source node voltages are equal, then the transistors 504 and 508, which in this embodiment are assumed to be matched, will present the same impedance to a bias circuit that includes the n-channel transistor 512 coupled in series with a diode connected p-channel transistor 516. A predetermined bias voltage appears at the node 520 which controls both bias levels of the two variable current generators 114, 116. The CMFB 130 operates as follows. Any differential variation in the common source nodes of the differential pairs should not cause a significant change in the total impedance between the node 520 and power supply return (ground), because of the way transistors 504 and 508 are connected to the same output node 520. However, if there is a common mode voltage change such that the voltages at both common source nodes of the differential pairs either increase or decrease simultaneously, then the impedance presented to node 520 will change, thereby causing the voltage at that node 520 to also change in a predetermined manner. In the embodiment shown in FIG. 5, if the voltages at the common source nodes of the differential pairs were to increase simultaneously, then the voltage at node 520 decreases in response, which in turn decreases the bias level and therefore the output current of each variable current generator 114, 116. On the other hand, if the common source node voltages of the differential pairs were to decrease simultaneously, then the voltage at the node 520 would increase in response, thereby increasing the tail currents I₁and I₂. Such a feedback mechanism helps improve the common mode rejection of the amplifier circuit when it is used as a difference amplifier.

[0040] Turning now to FIG. 6, this figure illustrates a circuit schematic of an embodiment of a digitally controllable variable offset comparator circuit. The comparator circuit includes an amplifier circuit substantially as shown in the schematic of FIG. 1, including the first and second differential pairs which are defined by transistors 102, 104 and 106, 108, respectively. The variable current generators 114 and 116 are also coupled to control the tail currents I₁and I₂to the respective differential pairs. These current generators 114, 116 are controlled by a digital value that is received on multiple, offset select lines as shown. In this embodiment, each digital value of the offset corresponds to two oppositely varying tail currents I₁and I₂that are equidistant from a nominal tail current. This configuration is similar to the example given above in connection with FIG. 1 which helped explain the availability of opposite polarity offset using the amplifier circuit.

[0041] A single ended output voltage for this comparator may be available as either V_outor V_out#. To drive these output signals into one of two possible stable states, a regenerative load circuit 610 is provided as shown. After being reset by an input signal, this regenerative load circuit 610 will quickly amplify any difference between V_outand V_out#, where such amplification occurs at a relatively high gain due to the cross coupled n-channel pair 620 and p-channel pair 624, thereby ensuring that the output signals V_outand V_out# only assume one of two possible stable states. Thus, if V_in⁺ is greater than V_in⁻ by at least the amount of offset that has been selected (as referred back to the input of the differential pairs), then the regenerative latch circuit 610 forcefully drives V_outto a low voltage level and simultaneously drives V_out# to a high voltage level. Other types of regenerative latch circuits may be used to provide the digital type output signal typically associated with a sense amplifier or a comparator application.

[0042] The variable offset comparator shown in FIG. 6 may behave as a sense amplifier which detects small differences between two analog signals V_in⁺ and V_in⁻. In this embodiment, a pair of differential signals are sensed, where the first differential signal is applied to the first differential pair 102, 104, while an inverted version of the differential signal is applied to the second differential pair 108, 106. This signal definition assumes that V_in⁺ is fed to the gate of the larger transistor 102 as well as the gate of the smaller transistor 106. Similarly, V_in⁻ is fed to the gate of the smaller transistor 104 and to the gate of the larger transistor 108.

[0043] Turning now to FIG. 6B, a flow diagram for an offset trimming process is depicted. In operation 660, the comparator inputs are shorted to each other and, if necessary, to some reasonable voltage. The offset adjustment is then sweeped from the maximum negative to positive offset values (operation 664). Assuming the gain of the comparator is very large, the output of the comparator will change from low to high when the offset adjustment value is at the lowest offset setting (operation 668). At this setting, the offset of the comparator may be less than 1 least significant bit (LSB) of offset resolution. Note that changing the offset of the comparator may be viewed as changing the reference level to which a differential input signal V_in⁺−V_in⁻, as connected in the embodiment of FIG. 6, will be compared by the comparator. Thus, the reference level of the comparator may be set by the offset adjustment.

[0044] The various embodiments of the amplifier circuit described above may be designed using conventional computer aided design/electronic design automation tools. FIG. 7 depicts a flow diagram of an embodiment of a computer-implemented method for designing the amplifier circuit. The method involves the creation of a representation of the differential transistor pairs where each pair is intentionally unbalanced (Operation 704). This representation may, for instance, include representing the first output node as being coupled to be driven by a larger transistor of the pair, and a second output node as being coupled to be driven by a smaller transistor of the pair. The size of the transistors that constitute the pair may be selected according to the width of the channel of each transistor, while keeping the length of the channels the same. Other techniques for sizing the transistors may alternatively be used. As was described above, the representation of the differential transistor pairs should show each pair as having first and second output nodes, where the first output node of the first pair is coupled to the second output node of the second pair, and the second output node of the first pair is coupled to the first output node of the second pair. This is also referred to here as an inter-coupled configuration of the dual differential transistor pair circuit.

[0045] The method also includes creating a representation of first and second variable current generators that are coupled to control respective tail currents of the first and second differential pairs (Operation 708). This representation of the variable current generators may include representing each generator as being digitally controllable as, for instance, shown in FIG. 3 (operation 712).

[0046] As mentioned above, there are various different scenarios for applying input signals to the amplifier circuit. For instance, a differential input signal may be applied to the larger and smaller transistors of the first pair, while an inverted version of the input signal may be applied to the larger and smaller transistors of the second pair. In a comparator embodiment, a magnitude of this differential input signal may be compared to a reference value that has been defined by a digital offset selection value. In a sense amplifier embodiment, this reference value may be set to a very small value. A representation of these input signals may be created using a computer-implemented method according to various well known techniques.

[0047] In addition, an output of the amplifier circuit may be defined in different ways as well. For instance, a differential output voltage may be defined as the difference between V_out^band V_out^a(see FIG. 1). Other types of outputs, such as the single ended output as defined for the sense amplifier/comparator application of FIG. 6, may also be represented in the computer-implemented method. In general, the representations of all the embodiments described above may be created using computer aided design/electronic design automation tools that are well know to those of ordinary skill in the art.

[0048] To summarize, various embodiments of an amplifier circuit and process for designing such a circuit have been described. In the foregoing specification, the invention has been described with reference to specific exemplary embodiments thereof. It will, however, be evident that various modifications and changes may be made thereto without departing from the broader spirit and scope of the invention as set forth in the appended claims. For instance, the particular circuit schematic shown in FIG. 4 for the digitally controllable current cell is just an example of a wide range of different designs that may be used to provide digital control of cell current. The specification and drawings are, accordingly, to be regarded in an illustrative rather than a restrictive sense.