1. Technical Field
The present invention relates generally to electrical circuit devices and specifically to signal propagation through electrical circuit devices. Still more particularly, the present invention relates to a method and circuit device for correcting duty-cycle distortion in signals propagating through electrical circuit devices.
2. Description of the Related Art
Duty-cycle in electrical circuit devices is a measure of the up pulse time versus the cycle period time for a clock signal propagating through the device. With many current device implementations, these clock signals are differential clock signals (rather then single ended clock signals), and tend to exhibit distortions in their duty-cycle as they propagate through the device.
Conventional circuit devices, such as ASICs (application-specific integrated circuits), for example, receive and propagate differential clock input signals with an up and down pulse via differential clock trees made up of a sequence of amplifiers (or clock buffers). These devices are bandwidth-limited and thus frequently experience a problem with duty-cycle distortion. This distortion is due to the fact that the clock trees are made with serial differential amplifier stages (or buffers) that have/exhibit bandwidths close to the clock frequency they are buffering. This distortion may also occur due to the large distance between amplifier stages (buffers) and the lowering of the amplifier bandwidth due to parasitic wiring capacitance. Each amplifier stage causes some amounts of distortion in the propagating clock signal, as illustrated by
As shown by
With these distortions adversely affecting the efficiency of these devices, two designs have been proposed to attempt to reduce the amount of duty-cycle distortions at each stage of propagation. Correction at each stage is implemented since if the duty-cycle is corrected at each stage, the drive distance (between stages) may be increased and the required bandwidth of the differential amplifiers is lowered. The first design involves changes or adjustments to traditional amplifier design while the second design involves adjustments to the output signal via single-ended feedback.
As illustrated by the accompanying chart 220, the addition of peaking allows the high frequency portion of the clock cycle to increase in magnitude while the lower frequency portion of the clock cycle decreases in magnitude. The chart 220 illustrates the gain over frequency utilizing the design of
The second design involves the addition of a single-ended feedback to the differential amplifier, by which the duty-cycle of the amplifier output is analyzed and then the amplifier circuit is adjusted (by feedback input) to correct the duty-cycle. The feedback method for duty-cycle correction is normally used on rail-to-rail, single-ended buffer circuits (i.e., complementary inputs) and involves a single-ended feedback.
The above single-ended feedback approach is utilized by several prior art references in a variety of applications. For example, U.S. Pat. No. 5,315,164 corrects a single-ended clock by adding in an error current to an incoming clock to change the switching threshold based on a measurement of averages of the single-ended circuit with a single-ended error current is single-ended. U.S. Pat. No. 5,896,053 utilizes a single-ended to complementary converter to create a true and complement clock signals, which are low-pass filtered to give an average DC level for each. The average signals are fed into an error amplifier that produces a single-ended error voltage, which is fed back to a voltage-controlled pulse-width modulator block that adjusts the duty-cycle.
As described above, conventional feedback approach utilizes single-ended circuit feedback. The above design, for example, changes the biasing of the inverter to make the output pulse width smaller or larger by passing a single plus or minus current to the inverter (i.e., adding or subtracting the single error current). These single-ended feedback circuits are, however, prone to noise and requires additional circuitry to convert between differential and single-ended signaling. Additionally, this design requires a large area due to the use of a replica and other feedback conversion mechanisms. Overall, these circuits have inherent problems with poor noise rejection, duty-cycle distortion, higher power, and increased area.
Disclosed is a circuit design and method for correcting duty-cycle distortions of a differential clock signal propagating through a differential amplifier (or clock buffer). A correction circuit is coupled to both (differential) output pulses/signals from the differential amplifier. The correction circuit comprises a differential low pass filter, which filters out the DC components of each output pulse/signal of the differential output, and a differential error amplifier, which compares the DC outputs from the low pass filter and generates a pair of differential error-adjustment DC currents. The differential error-adjustment DC currents are then fed back into the respective pulses of the differential output, where the duty-cycle of the differential output is corrected by adding the differential error-adjustment DC currents to respective pulses signals of the differential output. The duty-cycle distortion correction (DCDC) amplifier provides a completely differential approach to correction of duty-cycle distortions within the differential output.
The above as well as additional objectives, features, and advantages of the present invention will become apparent in the following detailed written description.
The invention itself, as well as a preferred mode of use, further objects, and advantages thereof, will best be understood by reference to the following detailed description of an illustrative embodiment when read in conjunction with the accompanying drawings, wherein:
The present invention provides a circuit design and method for correcting duty-cycle distortions of a differential clock signal propagating through a differential amplifier (or clock buffer). The circuit devices utilized include a differential amplifier, low-pass filter, and correction current source, which are combined into a simple two stage amplifier circuit with a correction output that is dotted to the differential output of a differential buffer.
A correction circuit is coupled to both (differential) output pulses/signals from the differential amplifier. The correction circuit comprises a differential low pass filter, which filters out the DC (direct current) components of each output pulse/signal of the differential output, and a differential error amplifier, which compares the DC outputs from the low pass filter and generates a pair of differential error-adjustment DC currents. The differential error-adjustment DC currents are then fed back into the respective pulses of the differential output, where the duty-cycle of the differential output is corrected by adding the differential error-adjustment DC currents to respective pulses signals of the differential output. The combination of the amplifier and correction circuit is referred to as a duty-cycle distortion correction (DCDC) differential amplifier to distinguish the DCDC differential amplifier configuration from convention amplifier correction circuits which provide only single-ended feedback correction currents.
Within the descriptions of the figures, similar elements are provided similar names and reference numerals throughout the figure(s). Where a later-described figure utilizes an element in a different context or with different functionality, the element is provided a different leading numeral representative of the figure number (e.g., 4xx for
Referring now to the figures,
Correction circuit 420 comprises low pass filter 430 and error correction amplifier 440, each receiving a differential input and producing a differential output. Input nodes of low pass filter 430 are respectively coupled to OUTP 415 and OUTN 417 of differential output clock wires 410. With these inputs, low pass filter 430 generates a pair of differential error outputs, ERROR_P 435 and ERROR_N 437. Low pass filter 430 senses the DC offset in the received differential output signals (415/417), filters these DC offsets out of the differential output signals (415/417), then forwards these DC offsets as respective error outputs (435/437) to error correction amplifier 440.
Error correction amplifier 440 is a differential current steering circuit that produces differential feedback outputs, FEEDBACK_P 445 and FEEDBACK_N 447, which are summed (dotted) into respective differential output signals (415/417) to produce a corrected differential output 415′/417′. As shown, correction circuit 420 (via error correction amplifier 440) provides DC correction voltages directly to differential outputs (415/417) rather than a feedback to amplifier itself.
As illustrated, error correction amplifier 440 comprises two N-channel transistors, coupled at their sources to a current source and at their drains to respective ones of differential output signal wires 410. N-channel transistors may be any type of transistor, e.g., field effect transistors (FETs) or CMOS FETS. Notably, also, one alternate embodiment of the invention may utilize P-channel transistors. With this alternate embodiment, the polarities of the error signals are inverted such that the P-channel transistors may turn on for relative sizes of negative DC voltages.
Each transistor is coupled at its gate input to one of the two error outputs (435/437) from low pass filter 430. The relative size of each error output (435/437) determines/influences the amount of current flow through each transistor (as the gate input turns the transistor on). This current flow in turn determines the amount of correction current provided to the particular output signal (415/417) connected to the transistor's source terminal.
Error correction amplifier 440 thus provides two correction currents, CORRECTION_P 445 and CORRECTION_N 447, each dotted into respective outputs, OUTP 415 and OUTN 417. Each correction current is the inverse of and proportional to the DC offset of the respective output signal (415/417). In the illustrative embodiment, the correction current is out of phase (180 degrees) with the DC offset from the differential output signal (415/417), so that the correction current operates to pull the DC offset (up or down) to a zero. The DC-level comparison by error correction amplifier 440 forces the average DC level of error signals 435/437 to be substantially equal (i.e., zero differential voltage). By bringing the DC offset in the error signals 435/437 to zero, the correction current boosts the small pulses (417) on the differential output wires 410 and shrinks the large pulses (415) on the differential output wires 410 so that the overall duty-cycle is improved.
The differential amplifier 405 may be any normal amplifier with normal differential tail current and resistor loads. Error correction amplifier 440 is a scaled down version of differential amplifier 405 having similar components (transistors and resistors). The error correction amplifier 440 is connected in such a way that the feedback signals 445/447 are 180 degrees out of phase with buffer output signals (415/417). Thus, OUTP 415 connects to inverted (negative) FEEDBACK_N 447 and OUTN 417 connects to inverted (positive) FEEDBACK_P 445. This pulls the output pulse that has the higher (relative) DC component down while the output pulse with the lower DC component is pulled up until both sides average approximately the same up time. The differential DC current from the output (445/447) of the correction amplifier is inverted to effectively remove the DC offset in buffer output signals (415/417) caused by duty-cycle distortion.
Notably, the feedback current provided is a differential pair of currents. The invention provides a “completely differential” approach to correcting the problems with duty-cycle distortions by introducing the DCDC amplifier, which includes a standard differential amplifier with output wires coupled to an input and an output of a correction circuit. All components of the correction circuit are differential and receive differential inputs and produce a differential output. No complementary signals are required. By implementing the completely differential approach, the differential feedback currents required are small. The voltage produced by the feedback currents is substantially smaller (e.g., 10's of millivolts) than would be required for an implementation utilizing complementary signals (100s of millivolts for switching). Also, by providing a completely differential approach, the resulting circuit requires a smaller number of stages, reduces power consumption, required area, and error.
As shown by
The addition of this differential feedback correction to a differential amplifier output enables the DCDC amplifier's correction circuit to dynamically analyze the duty-cycle of the amplifier's output and then adjust the output to correct the duty-cycle of the same and subsequent outputs. The design recognizes that a differential clock duty-cycle distortion has an average DC voltage offset between the positive and negative legs of the signal, and the design enables the average DC offset to be negated with the correction circuit, thus removing a substantial portion of the duty-cycle distortion.
The differential input signals, NP 515 and INN 517 (corresponding to output signals, OUTP 415 and OUTN 417 of
As shown, the error outputs are received at error correction amplifier 440, which generates correction/offset currents (FEEDBACK_P and FEEDBACK_N). These correction offset currents are generated by amplifying the filtered output signals (i.e., DC error outputs) and then passing these error outputs to error correction amplifier 440. The outputs of this correction amplifier stage are dotted (summed) into respective buffer output signals (415/417) propagating on amplifier output wires 410′ with negative polarity in order to provide negative correction currents (“feedback”). The negative correction currents reduces the high DC component while increasing the low DC component, resulting in correction of the duty-cycle within corrected output signals 515′/517′. As shown, the resulting relative pulse widths of both pulses (Tup and Tdown) are closer to being equal than when first inputted to low pass filter 430. Accordingly, the wider pulse (Tup) becomes shorter so that the up time is less, while the more narrow pulse (Tdown) becomes longer so that its up time is larger. Changes to the signal magnitudes may be on the order of 10's of millivolts.
All existing prior art approaches utilizes single-ended clocking, where the duty-cycle measurement is completed by comparing a DC average of a single ended clock to either a fixed reference voltage (DAC or voltage divider) or a complementary clock's DC average. Also, the duty-cycle correction is completed with single-ended signals. Such measurements have much more error than a truly-differential comparison of a differential signal, as described herein. Also, a majority of conventional feedback approaches to duty-cycle correction utilizes conversion circuits.
The present invention utilizes small-signal differential circuits to correct the duty-cycle. This differential approach is better at rejecting noise. Additionally, the present design has much better matching across product variations. Thus, unlike conventional feedback approaches, which utilize single-ended circuits (e.g., a single-ended reference and feedback voltage) and comparisons of complementary inputs to provide a single-ended feedback voltage, which are both prone to noise, the present invention provides a design by which a comparison is made of the differential outputs of the amplifier and a differential correction current is summed directly into the output of the amplifier.
One additional benefit of the present “completely differential” approach is that there is no requirement that the circuits convert between differential and single-ended signaling, since the present “completely differential” approach does not require the additional single-ended or conversion circuits. This reduces the total number of circuit stages required, reducing area and sources of noise. The completely differential approach also substantially eliminates the problems with such conventional circuits, which have inherent problems of poor noise rejection, duty-cycle distortion, higher power requirements, and increased area for added circuitry.
As a final matter, it is important that while an illustrative embodiment of the present invention has been, and will continue to be, described in the context of a fully functional computer system with installed management software, those skilled in the art will appreciate that the software aspects of an illustrative embodiment of the present invention are capable of being distributed as a program product in a variety of forms, and that an illustrative embodiment of the present invention applies equally regardless of the particular type of signal bearing media used to actually carry out the distribution. Examples of signal bearing media include recordable type media such as floppy disks, hard disk drives, CD ROMs, and transmission type media such as digital and analogue communication links.
While the invention has been particularly shown and described with reference to a preferred embodiment, it will be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the spirit and scope of the invention.