Collective automatic gain control

[0002] This U.S. Nonprovisional Application for Letters Patent is a divisional application of co-pending U.S. Nonprovisional Application for Letters Patent assigned Application No. 09/891,577 and filed Jun. 25, 2001 now U.S. Pat. No. 6,600,374.

[0003] The invention relates to automatic gain control in signal receivers.

[0004] High-speed digital systems such as memory systems sometimes use a form of I/O in which data is defined by a differential voltage signal. A differential voltage signal comprises a pair of complementary signals. A high logic level is represented by setting a first of the signals to a relatively high voltage and the second of the signals to a relatively low voltage. A low logic level is represented by switching the two voltages, so that the first signal has a relatively lower voltage than the second signal. Differential signaling is advantageous because of its relative immunity to noise and other signal degradations. A disadvantage of differential signaling is that it requires two signal lines for every data bit.

[0005] In order to reduce the number of data signaling lines, single-ended signaling is often used in high speed circuits. This particular type of signaling is non-differential, although it is sometimes referred to as “pseudo differential” signaling. Pseudo differential signaling specifies logic levels as voltages relative to a common, intermediate reference voltage. For example, a signal might be defined to represent a high logic level whenever its voltage is above the reference voltage, and to represent a low logic level whenever its voltage is below the reference voltage. This type of signaling requires fewer conductors than differential signaling, because a single reference line can be used in conjunction with many data signal lines. Although this type of signaling is less immune to signal degradation than differential signaling, it represents a distinct improvement over signaling systems in which signal levels are specified in terms of absolute voltages, rather than in relation to a specified reference voltage. Typically, a reference voltage signal is transmitted alongside data signals so that the same sources of noise will affect both the reference signal and the data signals. This tends to cancel the effects of the noise and provides some degree of noise immunity.

[0006] Regardless of whether signals are differential or non-differential, it is frequently desirable to perform some sort of signal buffering and/or conditioning at the receiving device. This is typically accomplished by a data receiver corresponding to each incoming signal line.

[0007] FIG. 1 shows a data receiver 10 that buffers an incoming data signal D_INto form a buffered or amplified internal data signal D_OUT. This circuit uses automatic gain control to achieve a desired voltage amplitude at D_OUT.

[0008] Data receiver 10 comprises a variable gain amplifier 12 that receives D_INand produces D_OUT. The receiver also has an envelope detector 14 that detects the voltage amplitude of signal D_OUT. An envelope detector or peak detector is a well-known type of circuit whose output voltage tracks the peak or swing voltages of a modulating input voltage such as a data signal.

[0009] FIG. 2 shows a simplified example of an envelope detector 14, comprising an FET control transistor M, a tracking capacitance C, a charging current source I_CH, and a discharging current source I_DIS. Transistor M is controlled by data signal D_OUTto charge capacitance C whenever D_OUTis relatively high. When the voltage V_ENVon capacitance C approaches the voltage of D_OUT, the transistor shuts off because of the reduced gate-to-source voltage of the transistor in this condition. Thus, the transistor charges capacitance C to approximately the “high” logic level of D_OUT. The size of current source I_CHdetermines the “attack” rate of the envelope detector—the rate at which output voltage V_ENVwill climb in response to an increased voltage at D_OUT. Current source I_DISis connected to slowly discharge capacitance C, to account for situations in which the peak levels of D_INdecrease over time. The size of current source I_DISdetermines the “decay” rate of the envelope detector—the rate at which output voltage V_ENVwill fall in response to a decreased input voltage D_OUT. I_DISis chosen to be small enough so that the voltage at D_OUTwill remain close to its peak value between peaks that occur in D_IN.

[0010] If appropriate sizes are selected for current sources I_CHand I_DIS, an envelope detector can be configured to produce an output that closely tracks the peak voltages of a data signal. Although the envelope detector of FIG. 2 is configured to detect positive signal peaks, the circuit can be easily altered to detect negative peaks in an input signal.

[0011] Referring again to FIG. 1, envelope detector 14 is configured to determine and track the peak voltage of output signal D_OUT, and to produce a voltage signal V_ENVrepresenting this peak voltage. A feedback component 20 receives V_ENVand compares it to a supplied reference amplitude V_AMP. The output of feedback component 20 is connected to the gain control of amplifier 12, forming a feedback loop that operates to minimize any difference between the peak output V_ENVof signal D_OUTand the supplied amplitude reference V_AMP. In other words, this circuit sets the gain of amplifier 12 so that the peak voltage of output D_OUTis approximately equal to the voltage of V_AMP. In implementation, feedback component 20 is a g_mstage whose output increases or decreases depending on the relative values of its inputs.

[0012] FIG. 3 shows data receiver 10, with the generic representation of amplifier 12 of FIG. 1 being replaced by a more detailed implementation of a differential amplifier circuit. Such a differential amplifier is typically used in conjunction with an input an input data signal D_INthat is specified relative to an intermediate reference voltage V_REF, which is relatively constant. The differential amplifier produces a differential voltage output D_OUThaving + and − outputs.

[0013] The differential amplifier comprises a differential pair of FET transistors M₂and M₃, whose sources are connected in common. The drains of M₂and M₃form the high and low outputs of differential voltage output signal D_OUT, and are connected through respective loads R1_LOADand R2_LOADto a high supply voltage V_dd. The gates of M₂and M₃are connected respectively to D_INand V_REF. The sources of M₂and M₃are connected in common through a biasing current source I_BIASto a low supply voltage V_ss.

[0014] The input of envelope detector 14 is connected to the positive side of differential output D_OUT. The output of feedback component 20 controls current source I_BIAS, which in turn controls the gain of the amplifier circuit. Feedback component 20 receives the output of envelope detector 14 and the amplitude reference V_AMP, and therefore establishes the gain of the amplifier circuit so that the peak voltage of output signal D_OUTis approximately equal to V_AMP.

[0015] The circuits described above have been used with success in many situations. However, problems arise in certain situations. One problem arises from the use of automatic gain control and envelope detectors in situations where there are relatively long periods without a transition in the received data signal. For example, a relatively long period in which a data signal remains low results in a decaying envelope voltage, which in turn causes automatic gain control circuits to inappropriately increase circuit gains. Furthermore, in many cases it is challenging to determine the optimal amplitude of D_OUT. It is desirable to keep the amplitude as low as possible to reduce power consumption, but also to keep it as high as necessary to ensure accurate differentiation between high and low signals.

[0016] A further concern arises where a data signal is precisely timed relative to other signals. In cases such as this, it is important to maintain the relative timing between the two or more signals. However, variations in amplification can affect this timing. This problem arises, for example, with incoming signals that must exceed a certain threshold voltage in order for them to be resolved by receiving circuitry. In a situation like this, a signal having a relatively higher amplitude will be resolved more quickly than a signal having a relatively lower amplitude. Thus, differences, in amplification between two signals will be perceived as timing differences by receiving circuitry.

[0017] Differences in duty cycles between two signals can also affect their relative timings as perceived by receiving circuitry. This problem can arise in situations where one signal is a true differential voltage signal and the other is a non-differential or pseudo-differential voltage signal. A differential voltage signal typically achieves very close to a 50% duty cycle, even after amplification. Non-differential signals, however, are often subject to disproportionate amplification with respect to the positive and sides of their waveforms. This is especially true in circuits where reduced supply voltages are employed in order to increase circuit speeds. In such circuits, current sources sometimes begin to operate below saturation typically reducing the amount of amplification applied to negative portions of waveforms. This has the effect of changing the duty cycle of an amplified signal, which also changes the relative point in time at which the signal crosses specified voltage thresholds.

[0018] The circuits and techniques described below address these and other issues.

[0019] FIG. 1 is a schematic diagram of a prior art automatic gain control circuit.

[0020] FIG. 2 is a schematic diagram of an envelope detection circuit.

[0021] FIG. 3 is a schematic diagram of a prior art automatic gain control circuit.

[0022] FIGS. 4 and 5 are schematic diagrams of an automatic gain control circuit in which the envelope of an amplified data signal is adjusted to equal the envelope of a related clock signal.

[0023] FIG. 6 is a schematic diagram of a receiver circuit employing group envelope detection techniques.

[0024] FIGS. 7 and 8 are schematic diagrams showing examples of group envelope detection circuitry.

[0025] FIG. 9 is a schematic diagram of a receiver circuit employing group envelope detection techniques.

[0026] FIG. 10 is a schematic diagram illustrating group envelope detection applied to groups or bytes of data signals.

[0027] FIG. 11 is a schematic diagram illustrating techniques for correcting asymmetries in differential voltage output signals.

[0028] FIG. 12 is a graph illustrating asymmetries that can occur in prior art differential amplifier circuits.

[0029] FIG. 13 is a schematic diagram illustrating the use of a plurality of differential amplifiers such as the one shown in FIG. 11.

[0030] FIG. 14 is a schematic diagram illustrating circuits for operating the previously discussed circuits in situations where gains are calibrated in calibration procedures and stored for later use when receiving signals from different senders.

[0031] FIG. 15 is a schematic diagram of a g_mstage suitable for use in the embodiments described above.

[0032] The following description sets forth specific embodiments of circuits and techniques that incorporate elements recited in the appended claims. The embodiments are described with specificity in order to meet statutory requirements. However, the description itself is not intended to limit the scope of this patent. Rather, the inventors have contemplated that the claimed invention might also be embodied in other ways, to include different elements or combinations of storage elements similar to the ones described in this document, in conjunction with other present or future technologies.

[0033] FIG. 4 shows one embodiment of a receiver circuit 40. The signal receiver receives a data signal D_INand produces an amplified data output signal D_OUT. The input data signal D_INis timed relative to a clock signal CLOCK.

[0034] Receiver circuit 40 comprises an amplifier 42 having a variable gain. The amplifier receives data input signal D_INand produces amplified data output signal D_OUT.

[0035] Receiver circuit 40 implements automatic gain control with respect to its received data signal. The automatic gain control is implemented in part by signal envelope detection circuitry 44 that is responsive to the amplified data output signal D_OUTto indicate an amplified signal envelope voltage VO_ENV. VO_ENVtracks the voltage swing or peak voltage of D_OUT. FIG. 2 shows a simplified example of an appropriate envelope detector.

[0036] The automatic gain control of circuit 40 also comprises clock envelope detection circuitry 45 responsive to clock signal CLOCK to indicate a clock envelope voltage VC_ENV. VC_ENVtracks the voltage swing or peak voltage of clock signal CLOCK. Again, FIG. 2 shows a simplified example of an appropriate envelope detector.

[0037] Automatic gain control circuitry further includes a feedback component 46 configured to adjust the gain of amplifier 42 as a function of the difference between the amplified signal envelope voltage VO_ENVand the clock envelope voltage VC_ENV. More specifically, the feedback component has negative and positive inputs that receive VO_ENVand VC_ENV, respectively, and in response produces a signal V_Dthat changes depending on the relative values of VO_ENVand VC_ENV. A component such as this is typically referred to as a g_mstage. By connecting V_Dto the gain control input of amplifier 42, the amplifier gain is controlled in a way that reduces or minimizes any difference between VO_ENVand VC_ENV, effectively setting the envelope, swing, or amplitude of D_OUTto that of clock signal CLOCK.

[0038] This circuit is especially beneficial in systems where received data signals are single-ended, non-differential signals and that are timed relative to a differential voltage clock signal. Typically, the differential nature of the clock signal allows it to be conditioned more effectively than single-ended inputs and for its swing or amplitude to be set with precision to an optimum value. With the circuit of FIG. 4, the effort in conditioning the clock signal can be leveraged to determine and establish an optimum amplitude for buffered received data signals. Furthermore, setting the amplitude of received data signals to match that of the clock signal reduces timing or skew mismatches between the two types of signals, which might otherwise arise because of differing amounts of amplification being introduced in the different signals.

[0039] A further advantage of this circuit is that a reduced power mode can be implemented simply by reducing the amplitude of VC_ENV. Reducing the amplitude of VC_ENVdecreases the gain applied by amplifier 42, thereby reducing power consumption.

[0040] FIG. 5 shows an embodiment of data receiver 40 in which the generic representation of amplifier 42 in FIG. 4 has been replaced by a more detailed implementation of a differential amplifier circuit. Such a differential amplifier is appropriate for use in conjunction with an input data signal D_INthat is specified it relative to a relatively constant, intermediate reference voltage V_REF. The differential amplifier produces a differential voltage output D_OUThaving + and − outputs.

[0041] The differential amplifier comprises a differential pair of FET transistors M₄and M₅whose sources are connected in common. The drains of M₄and M₅form the high and low outputs of differential voltage output signal D_OUT, and are connected through respective loads R3_LOADand R4_LOADto a high supply voltage V_dd. The gates of M₄and M₅are connected respectively to D_INand V_REF. The sources of M₄and M₅are connected in common through a biasing current source I_BIASto a low supply voltage V_ss.

[0042] The input of envelope detector 44 is connected to the positive side of differential output D_OUT. The input of envelope detector 45 is connected to one end of differential voltage clock signal CLOCK. The outputs of the detectors 44 and 45 are connected to the positive and negative inputs, respectively, of a feedback component 46, such as a g_mstage as discussed above. The output of feedback component 46 controls current source I_BIAS, which in turn controls the gain of the amplifier circuit. This configuration establishes the gain of the amplifier circuit so that the peak voltage or swing of output signal D_OUTis approximately equal to the peak voltage or swing of clock signal CLOCK.

[0043] FIG. 6 shows another embodiment of a signal receiver, generally designated by reference numeral 50. In this example, signal receiver 50 receives a plurality of voltage data signals D_IN. For each data signal D_IN, signal receiver 50 has a corresponding amplifier 52. The amplifiers receive respective data signals D_INand produce corresponding amplified data signals D_OUT. The amplifiers have variable gains.

[0044] Receiver 50 has group envelope detection circuitry that is responsive to the collective amplified data signals D_OUTto indicate a group envelope voltage. In this example, the group envelope detection circuitry comprises a plurality of individual envelope detectors 54 whose capacitive outputs are connected in common to form a group envelope voltage signal VG_ENV. In this configuration, the detection circuitry is responsive to peaks in any of data output signals D_OUT, and VG_ENVgenerally tracks the most extreme peaks occurring at any moment on any of the data output signals.

[0045] The group envelope detection circuitry can be implemented in different ways. One implementation is shown in FIG. 7, where two envelope detectors 54aand 54bhave outputs connected in common. Although FIG. 7 shows only two individual detectors, any number of detectors can be configured together in this manner. As discussed with reference to FIG. 2, each detector includes a FET control transistor M, a tracking capacitance C, a charging current source I_CH, and a discharging current source I_DIS. Each circuit works as already described with reference to FIG. 2. The capacitive outputs of all such individual circuits are connected in parallel to form VG_ENV.

[0046] FIG. 8 shows another configuration, in which the tracking capacitance C and discharging current source I_DISare shared. There is a charging current source I_CHand control transistor M associated with each data signal D_OUT. However, these components are connected in common to charge a single capacitance C. Again, any number of detectors can be connected together in such a fashion.

[0047] In either of these embodiments, it is possible to choose current source and capacitance sizes so that the overall circuit works in a roughly OR'ed fashion, reflecting the most extreme peak occurring at any moment on any of the data output signals, subject to attack and decay times that depend on the actual circuit implementation. The circuit output can alternatively be viewed as presenting the average of the peak voltages of the data signals D_OUT. Although these circuits as shown are appropriate for tracking positive envelopes or peaks, inverse implementations can easily be designed for detecting negative envelopes or peaks.

[0048] Referring now again to FIG. 6, group envelope voltage VG_ENVis provided to the negative input of a feedback component 60, such as a g_mstage as discussed above. An amplitude reference signal is supplied to the positive input of feedback component 60. The output of feedback component 60 is connected in parallel to all of amplifiers 52 to control their gains. Feedback component 60 thus forms feedback that adjusts the amplifier gains as a function of the difference between group envelope voltage VG_ENVand supplied amplitude reference V_AMP. More specifically, the feedback adjusts the amplifier gains to reduce or minimize the difference between group envelope voltage VG_ENVand supplied amplitude reference V_AMP.

[0049] In accordance with the technique described with reference to FIG. 4, V_AMPcan be derived from a clock signal CLOCK that is used for timing transfer of information on data signals D_IN. A reference or clock envelope detector 62 receives the clock signal, which can be either a differential voltage signal or a non-differential voltage signal, and produces V_AMPas a reference or clock envelope voltage. As already described, V_AMPis used as feedback in combination with VG_ENV, so that the group envelope voltage is set to the approximate envelope voltage of clock signal CLOCK.

[0050] The multi-amplifier circuit of FIG. 6 is preferably duplicated for the data lines of each data group, data byte or data word, with the exception of clock envelope detector 62 whose output is used in common by all such circuits. Thus, within a single data byte or word, each signal line is subject to the same amplification. Furthermore, the gains applied to the respective data groups are adjusted so that their group envelopes match a common reference—in this case the envelope of the clock signal against which the data signals are timed.

[0051] This technique is a significant improvement over the prior art. Specifically, this circuit exhibits improved response in situations where a single data input might remain unchanged for a relatively long period. In prior art circuits, an individual envelope detector would have begun to decay during this period, and automatic gain control would have responded by inappropriately increasing the circuit gain. In the circuit of FIG. 6, however, this likelihood is reduced by a factor of N, where N is the number of data inputs that are collectively monitored by group envelope detector circuitry 54. Activity on any one of these inputs prevents the decay of the envelope voltage, and thus prevents the automatic gain control feedback from inappropriately increasing the circuit gain.

[0052] Furthermore, the techniques employed by the circuit of FIG. 6 minimize signal skew within data signals. In particular, the circuit of FIG. 6 results in the same gain being applied to all of the individual data lines of a data byte or data word. This minimizes line-to-line variations in signal skew, and accordingly increases the reliability of the signals. Byte-to-byte skew is also reduced by the technique of adjusting the gains of each byte to a common reference.

[0053] FIG. 9 shows the signal receiver 50 with specific implementations shown for individual amplifiers 52. These differential amplifiers are appropriate for use in conjunction with input data signals D_INthat are specified relative to a relatively constant, intermediate, common reference voltage V_REF. In response to D_INand V_REF, each differential amplifier produces a differential voltage output signal D_OUThaving + and − outputs.

[0054] Each differential amplifier 52 comprises a differential pair of FET transistors M₆and M₇whose sources are connected in common. The drains of M₆and M₇form the high and low outputs of differential voltage output signal D_OUT, and are connected through respective loads R5_LOADand R6_LOADto a high supply voltage V_dd. The gates of M₆and M₇are connected respectively to D_INand V_REF. The sources of M₆and M₇are connected in common through a biasing current source I_BIASto a low supply voltage V_ss.

[0055] The group envelope detection circuitry is depicted in FIG. 9 as a single group envelope detector 54 that receives the positive side of differential D_OUTfrom each amplifier 52. The output of group envelope detector 54 is connected to the positive input of feedback component 60. The circuit also includes clock envelope detector 62 whose output is connected to the negative input of feedback component 60. The output of feedback component 60 is connected in parallel to control the current sources I_BIASof the different amplifiers 52. Thus, the amplifiers are controlled to all have the same gain—a gain that produces a group envelope approximately equal to that of the clock signal CLOCK.

[0056] FIG. 10 shows yet another embodiment of a signal receiver, generally referenced by numeral 70. This embodiment is similar to that shown in FIG. 6, except that the data input signals D_INare segregated into groups such as data bytes or words. FIG. 10 shows two such groups: Byte A and Byte B.

[0057] Signal receiver 70 includes groups of amplifiers corresponding respectively to the groups of input signals D_IN. The amplifiers produce corresponding groups of amplified data signals D_OUT. The amplifiers have variable gains.

[0058] A group envelope detector 76 is associated with each group of amplified data signals D_OUT. Each group envelope detector is responsive to its associated group of amplified data signals D_OUTto indicate a group envelope voltage VG_ENVfor that group of signals D_OUT. Group envelope detectors are implemented in accordance with the techniques already described.

[0059] Receiver 70 comprises feedback associated with each group of amplifiers. The feedback comprises a g_mstage 78 that collectively adjusts the gains of the associated amplifiers as a function of the difference between the group envelope voltage VG_ENVof the amplifiers and a supplied amplitude reference V_AMP.

[0060] In this embodiment, V_AMPcomprises the envelope or peak voltage of a clock signal with which data signals D_INare timed. Specifically, the receiver comprises a clock envelope detector 79 that receives clock signal CLOCK and that in response produces clock envelope voltage signal V_AMP. Each of feedback components 78 receives the same clock envelope voltage signal V_AMP.

[0061] FIG. 11 illustrates a further technique that can be used to condition received data signals. The circuit shown in FIG. 11 addresses a problem that sometimes arises in amplifier circuits such as those shown in FIGS. 3, 5, and 9. These amplifiers accept differential inputs and produce differential outputs. However, the inputs D_INand V_REFare not truly differential, in that V_REFis actually a constant voltage while D_INvaries above and below V_REF. This can cause asymmetries in the amplified output signals. These asymmetries become acute when supply voltages are reduced to achieve higher operating speeds, creating situations in which certain transistors fail to reach saturation.

[0062] FIG. 12 illustrates this anomaly, showing the voltage outputs of D+_OUTand D−_OUTin response to an oscillating input D_IN. It is apparent that the negative peak V₁of D+_OUTis significantly lower than the negative peak V₂of D−_OUT. The circuit of FIG. 11 corrects this.

[0063] In the circuit of FIG. 11, a receiver 80 comprises a differential amplifier having two stages of amplification. A first amplification stage is similar to the double-ended differential amplifiers already discussed, being responsive to a data signal D_INand an intermediate reference voltage V_REFto produce positive and negative differential output signals D+_OUTand D−_OUT.

[0064] More specifically, the first amplification stage comprises a differential pair of transistors M₁₀and M₁₁, corresponding loads R7_LOADand R8_LOAD, and a biasing current source referred to in FIG. 11 as IA_BIAS. The sources of transistors M₁₀and M₁₁are connected in common and through IA_BIASto lower supply voltage V_ss. The drains of transistors M₁₀and M₁₁are connected respectively through loads R7_LOADand R8_LOADto upper supply voltage V_dd. The gate of transistor M₁₀is connected to data input signal D_IN, while the gate of transistor M₁₁is connected to intermediate voltage reference V_REF. Transistors M₁₀and M₁₁produce an amplified differential voltage output signal D_OUTcomprising first and second output signals referred to as D+_OUTand D−_OUT. The + component of D_OUTis generated at the drain of transistor M₁₀. The − component of D_OUTis generated at the drain of transistor M₁₁. The gain of the first amplification stage is adjustable by adjusting the magnitude of current produced by current source IA_BIAS.

[0065] Receiver 80 comprises a first envelope detector 82 whose input is connected receive one of the D_OUTsignals. Specifically, envelope detector is configured to detect the negative envelope or peaks of positive differential voltage signal D+_OUTand to create a first envelope voltage V1_ENVindicating the negative envelope of D+_OUT.

[0066] The receiver 80 comprises feedback, associated with the described first amplification stage, to adjust the gain of the first amplification stage to reduce the difference between the first envelope voltage V1_ENVand a supplied reference voltage. As discussed above, the supplied reference voltage is preferably the detected envelope voltage VC_ENVof a related clock signal. As shown in FIG. 11, clock signal CLOCK is received by an envelope detector 84 to produce the reference voltage VC_ENV.

[0067] The feedback associated with the first amplification stage comprises a g_mamplification stage or feedback component 86 whose output is connected to control current source IA_BIAS. The feedback adjusts the current through the first amplification stage so that the negative envelope of D+_OUTis approximately equal to the negative envelope of clock signal CLOCK. This adjustment affects both of the differential outputs D+_OUTand D−_OUT, although the negative envelope of the D−_OUTsignal potentially exhibits the anomaly described with reference to FIG. 12.

[0068] Receiver 80 further comprises a second, singled-ended amplification stage that affects the two differential outputs D+_OUTand D−_OUTunequally. Specifically, this second amplification stage affects only the negative, D−_OUToutput of the differential output pair.

[0069] The second amplification stage comprises a differential pair of transistors M₁₂and M₁₃. The gate of M₁₂is controlled by D_INand the gate of M₁₃is controlled by V_REF. The sources of these two transistors are tied in common and through a supplemental current source IB_BIASto the lower supply voltage V_ss. The drain of transistor M₁₃is connected to negative differential output D−_OUT. However, the drain of transistor M₁₂is connected directly to upper supply voltage V_dd. Thus, this stage of amplification affects only negative differential output D−_OUT. The gain of this amplification stage is adjustable, and is controlled by current source IB_BIAS.

[0070] The circuit further comprises a second envelope detector 88 that is responsive to the negative differential output signal D−_OUTto produce a second envelope voltage V2_ENVreflecting the negative envelope or peaks of negative differential output signal D−_OUT. A g_mstage or feedback component 90 is responsive to the V1_ENVand V2_ENVto adjust the gain of the second amplification stage. Specifically, this feedback is configured to adjust the gain of the second amplification stage to minimize the difference between the first and second envelope voltages V1_ENVand V2_ENV.

[0071] The two amplification stages, and the feedback associated with each amplification stage, effectively adjust both the positive and negative components of D_OUTto have amplitudes matching that of the clock signal CLOCK, and correct for the asymmetrical differential output illustrated in FIG. 12.

[0072] Although FIG. 12 shows a receiver that receives only a single data signal, the techniques shown in FIG. 13 can be used effectively in a receiver that receives a plurality of signals. FIG. 13 shows such a receiver 100.

[0073] In FIG. 13, receiver 100 includes a plurality or group of signal receivers 102 such as described above with reference to FIG. 11. Each signal receiver corresponds to and receives a single data input signal D_IN, and in response produces a differential voltage output data signal D_OUT. Each receiver has a first amplification stage that affects both the positive and negative components D+_OUTand D−_OUTof the differential output. Each receiver also has a second amplification stage that affects the first and second differential voltage data signals unequally. Specifically, the second amplification stage of each receiver affects only the negative, D−_OUTdifferential output signal.

[0074] Receiver 100 includes a clock envelope detector 103 that is responsive to a differential voltage clock signal to indicate a clock envelope voltage VC_ENV. In is this embodiment, the envelope detector is configured to detect the negative envelope or peaks of the clock signal CLOCK.

[0075] Receiver 100 also includes first and second group envelope detectors 104 and 106. These detectors are like the group detectors previously discussed with reference to FIGS. 7 and 8, in that each detector is responsive to a collection of signals to produce an envelope signal representing the most extreme envelope of the collection or to produce an average of the envelopes of the signals. In this embodiment, these envelope detectors are configured to detect the negative envelopes or peaks of the respective signals.

[0076] More specifically, first group envelope detector 104 is responsive to the collective positive differential voltage signals D+_OUTto indicate a first group envelope voltage V1G_ENV. Second envelope detector 106 is responsive to the collective negative differential voltage signals D−_OUTto indicate a second group envelope voltage V2G_ENV.

[0077] Receiver 100 includes two feedback components 108 and 109. A first feedback component or g_mstage 108 is configured to adjust the gain of the first amplification stages of the individual signal receivers 102 to reduce or minimize the difference between the first group envelope voltage V1G_ENVand the clock envelope voltage VC_ENV. A second feedback component or g_mstage 109 is configured to adjust the gain of the second amplification stages of the individual signal receivers 102 to reduce or minimize the difference between the first group envelope voltage V1G_ENVand the second group envelope voltage V2G_ENV.

[0078] The various exemplary embodiments described above illustrate the use of real-time feedback to match both individual and group envelope voltages. Specifically, the described envelope detection circuitry operates continuously to indicate group envelope voltages, and the AGC feedback similarly operates in a continuous mode in response to varying group envelope voltages. This is appropriate in many situations, and especially in situations where point-to-point data signaling is used. Such real-time feedback might not be appropriate in other situations, however, especially those in which data might be received from a plurality of different senders. In situations such as this, it might be more appropriate to determine gains during calibration procedures performed while receiving data from the different senders.

[0079] FIG. 14 shows pertinent components to implement an initialization procedure to determine appropriate gains for use when receiving signals from different senders. AGC feedback component 120 operates as described above with reference to received individual or group envelope signals and produces an output that adjusts the gain of an amplifier 122. A switching circuit 124 is configured during the initialization procedure to supply the output of the AGC feedback component 120 to the gain control input of amplifier 122. During initialization, an appropriate data pattern is transmitted from the sending device and the AGC feedback is allowed to reach a steady state. An analog-to-digital converter 124 samples the voltage output of feedback component 120 and converts it to a digital value. This value is stored in storage elements or registers 126. This procedure is repeated for different senders, and a gain value is stored for each sender.

[0080] During subsequent, normal operation, switching circuit 124 is configured to supply a voltage produced by a digital-to-analog converter 128. Digital-to-analog converter 128 receives its input from storage elements 126, which are controlled to output a value that corresponds to the particular sending device that is currently active.

[0081] Optionally, PVT (process, voltage, and temperature) correction can be performed on the data before it is provided to digital-to-analog converter 128, as shown by PVT correction component 130.

[0082] FIG. 15 shows an example of a g_mstage such as used in the circuits described above, generally designated in FIG. 15 by reference numeral 200. The g_mstage is preferably implemented digitally, so that PVT compensation and calibration can be more easily accomplished.

[0083] The g_mstage 200 includes a comparator 202 that receives a reference signal and a feedback signal. The comparator produces a positive or “true” output signal when the voltage of the feedback signal is lower than that of the reference signal, and a negative or “false” output signal when the voltage of the feedback signal is higher than that of the reference signal.

[0084] A digital counter 204 receives the output of comparator 202 at its increment/decrement input, labeled “+/−” in FIG. 15 Counter 204 is clocked at a chosen frequency, and produces a digital output that varies as a function of the output of comparator 202. If the output of comparator 202 is positive, the counter increments its output at every clock cycle. If the output of comparator is negative, the counter decrements its output at every clock cycle. In many embodiments, it may be desirable to filter the signal from comparator 202 to reduce jitter. For example, the circuit might be configured to require that the output of comparator 202 remain stable for a number of consecutive clock cycles before it is considered valid, or before counter 204 will respond to it. “Majority” filtering might alternatively be used, wherein a number of consecutive samples are collected and the majority value of the samples is used as the filtered output.

[0085] The digital output of counter 204 is received by a digital-to-analog converter (DAC) 206, where it is converted to an analog output signal.

[0086] This circuit produces an analog output voltage that varies in response to the reference and feedback signals. At equilibrium, the output voltage will be such that the reference and feedback signals are approximately equal, assuming that the feedback signal is related in some positive manner to the output voltage.

[0087] The circuits and techniques describe above provide improvements in automatic gain control. By collectively adjusting the gains of each data byte or word, I/O line-to-line skew is minimized. Mismatches between the signal lines of different bytes are minimized by adjusting different bytes or words to a common reference. Furthermore, the techniques can be generalized to match the group envelope of one type of data to the group envelope of some other type of data.

[0088] Although details of specific implementations and embodiments are described above, such details are intended to satisfy statutory disclosure obligations rather than to limit the scope of the following claims. Thus, the invention as defined by the claims is not limited to the specific features described above. Rather, the invention is claimed in any of its forms or modifications that fall within the proper scope of the appended claims, appropriately interpreted in accordance with the doctrine of equivalents.