SWARTZ RONALD (US)
TU WEN-LUNG (US)
SRIKANTH ADHIVEERARAGHAVAN (US)
SWARTZ RONALD (US)
TU WEN-LUNG (US)
US6384653B1 | 2002-05-07 | |||
EP1184971A1 | 2002-03-06 |
1. | What is claimed is: A phase interpolator, comprising: an alternating current (AC) coupling capacitor; and a common mode bias keeper circuit coupled to the AC coupling capacitor, wherein the AC coupling capacitor and the common bias keeper circuit operate together to force common modes of phase interpolator input signals to be the same. |
2. | The phase interpolator of claim 1, further comprising: a preconditioner circuit; a mixer circuit; and an amplifier. |
3. | The phase interpolator of claim 2, further comprising the AC coupling capacitor and the common mode bias keeper circuit coupled between the mixer circuit and the amplifier. |
4. | The phase interpolator of claim 3, further comprising: a second AC coupling capacitor coupled between the pre conditioner circuit and the mixer circuit; and a second common mode bias keeper circuit coupled to the second AC coupling capacitor. |
5. | The phase interpolator of claim 2, further comprising the AC coupling capacitor and the common mode bias keeper circuit coupled between the preconditioner circuit and the mixer circuit. |
6. | The phase interpolator of claim 2, wherein the preconditioner circuit further comprises circuitry to generate overlap between a plurality of phase interpolator input signals. |
7. | The phase interpolator of claim 2, wherein the mixer circuit further comprises circuitry to mix signals from a preconditioner circuit, the mixing based on proportionate current weighting controlled by a digitaltoanalog converter. |
8. | The phase interpolator of claim 7, wherein the digitaltoanalog converter generates controls for the proportionate current weighting based on edge and data samples, and on the current position of a sampling clock. |
9. | The phase interpolator of claim 2, wherein the amplifier further comprises circuitry to receive signals from the mixer circuit in order to produce railtorail sampling clocks. |
10. | The phase interpolator of claim 9, wherein the amplifier is a Complementary Metal Oxide Semiconductor (CMOS) level converter. |
11. | A method, comprising: receiving a plurality of reference clock signals; maintaining a common mode of the plurality of reference clock signals by a common mode circuit, the common mode circuit including an alternating current (AC) capacitor and a common mode bias keeper circuit; generating interrelated control signals based on comparing the plurality of reference clocks to a received data signal; and outputting amplitude contributions from phases of the interrelated control signals. The method of claim 11, wherein the maintaining a common mode further comprises forcing the common mode of the plurality of reference clock signals to be the same. |
12. | The method of claim 11, wherein the maintaining a common mode further comprises: receiving the plurality of reference clock signals from a mixer circuit; forcing the common mode of the plurality of reference clock signals to be the same; and sending the reference clock signals to an amplifier. |
13. | The method of claim 13, further comprising: receiving the plurality of reference clock signals from a pre conditioner circuit; forcing the common mode of the plurality of reference clock signals to be the same; and sending the reference clock signals to the mixer circuit. |
14. | The method of claim 11, wherein the maintaining a common mode further comprises: receiving the plurality of reference clock signals from a pre conditioner circuit; forcing the common mode of the plurality of reference clock signals to be the same; and sending the reference clock signals to a mixer circuit. |
15. | A receiver, comprising: a local reference clock providing a plurality of clock phases; a clock and data recovery unit to receive a serial data signal from a transmitter; and a phase interpolator within the clock and data recovery unit, the phase interpolator including: an alternating current (AC) coupling capacitor; and a common mode bias keeper circuit coupled to the AC coupling capacitor, wherein the AC coupling capacitor and the common bias keeper circuit operate together to force the common mode of phase interpolator input signals to be the same. |
16. | The receiver of claim 16, the phase interpolator further comprising: a preconditioner circuit; a mixer circuit; and an amplifier. |
17. | The receiver of claim 17, the phase interpolator further comprising the AC coupling capacitor and the common mode bias keeper circuit coupled between the mixer circuit and the amplifier. |
18. | The receiver of claim 18, the phase interpolator further comprising: a second AC coupling capacitor coupled between the pre conditioner circuit and the mixer circuit; and a second common mode bias keeper circuit coupled to the second AC coupling capacitor. |
19. | The receiver of claim 17, the phase interpolator further comprising the AC coupling capacitor and the common mode bias keeper circuit coupled between the preconditioner circuit and the mixer circuit. |
FIELD OF THE INVENTION
[00011 The present embodiments of the invention relate generally to phase
interpolators and, more specifically, relate to such interpolators useful in
recovering a clock from serial data sent to a receiver.
BACKGROUND
[0002] In many data communication arrangements, separate clock signals
are not transmitted with the data. This requires recovering the clock from the
data at the receiving end in order to then recover the data. When transmitting the
clocked data across a transmission medium, noise in the data signal, such as jitter
and phase skew, reduces the sampling window for the data. Duty cycle
distortion, for instance, is caused by non-symmetric positive and negative duty
cycles of a data symbol and can show up either as a high frequency correlated
jitter or as a phase step. The jitter, phase skew, and duty cycle distortion reduce
the perceived sampling window by the receiver.
|0003| Phase interpolator circuits are increasingly used in embedded clock
data recovery systems to position the sampling clock at the center of the data eye.
Phase interpolators typically use fixed phase clocks, generated from a Phase
Locked Loop (PLL), and mix them appropriately to generate interpolated clocks
that can be adjusted to be at the center of the data bit. Some implementations of
phase interpolator circuitry contain pre-conditioner circuitry, mixer circuitry, and
an amplifier. With the phase interpolator operating at multiple Gb/S speeds, any
duty cycle corruption can result in improperly sampled data.
[0004] The fixed phase clocks at the input of the phase interpolator may
have some cycle-cycle jitter from the source and power supply noise. There is
also a possibility of phase skew between the adjacent clocks that are mixed in the
phase interpolator. This jitter and phase skew lead to duty cycle distortion in
phase interpolator outputs.
[0005] A conventional phase interpolator circuit is very sensitive to input
jitter and layout mismatches which may result in poor quality clocks. With jitter
and phase skew in the input clocks, the outputs of the pre-conditioner circuitry in
a phase interpolator may have different common mode voltages. These different
common mode voltages cause differing operating points in the differential mixer
circuitry of the phase interpolator. This may result in duty cycle distortion in the
phase interpolator output clocks and, consequently, improperly sampled data.
With higher speeds and jitter associated with clocks increasing, a phase
interpolator that can provide cleaner clocks, even with significant phase skew and
power supply noise induced jitter on the input clocks, will help reduce the
occurrence of improperly sampled data.
BRIEF DESCRIPTION OF THE DRAWINGS
[0006| The present invention will be understood more fully from the
detailed description given below and from the accompanying drawings of various
embodiments of the invention. The drawings, however, should not be taken to
limit the invention to the specific embodiments, but are for explanation and
understanding only.
[0007| Figure 1 illustrates a block diagram of a simple communication
system;
[0008] Figure 2 illustrates a block diagram of a receiver of the type in which
a phase interpolator may be used;
|0009] Figure 3 illustrates a block diagram of one embodiment of a phase
interpolator;
fOOlO] Figure 4 illustrates a graphical representation of ideal phase spacing
of outputs of the phase interpolator;
10011] Figure 5 illustrates a graphical representation of non-ideal phase
spacing of outputs of the phase interpolator;
[0012] Figure 6 illustrates a circuit/block diagram of one embodiment of a
phase interpolator; and
[0013] Figure 7 is a flow diagram of one embodiment of a method for
recovering an embedded clock from a data stream.
DETAILED DESCRIPTION
[0014J A method and apparatus to reduce duty cycle distortion in a phase
interpolator circuit is described. Reference in the specification to "one
embodiment" or "an embodiment" means that a particular feature, structure, or
characteristic described in connection with the embodiment is included in at least
one embodiment of the invention. The appearances of the phrase "in one
embodiment" in various places in the specification are not necessarily all referring
to the same embodiment.
[0015] In the following description, numerous details are set forth. It will
be apparent, however, to one skilled in the art, that the embodiments of the
invention may be practiced without these specific details. In other instances, well-
known structures and devices are shown in block diagram form, rather than in
detail, in order to avoid obscuring the present invention.
[00161 Figure 1 is a block diagram of one embodiment of a simple
communication system 100 that may be used to reduce duty cycle distortion in a
phase interpolator circuit. The system 100 includes a transmitter (Tx) 110, a serial
data signal 130, and a receiver (Rx) 120. Receiver unit 120 further includes a clock
and data recovery (CDR) unit 125.
[0017] Transmitter 102 transmits the serial data signal 130 including, for
example, a series of data symbols, to receiver 120. CDR unit 125 of receiver 120
samples serial data signal 130 (for example, symbols included in the serial data
signal) to recover data from the serial data signal. CDR unit 125 samples serial
data signal 130 at sample times established by a sampling signal 140 generated
locally at receiver 120. In some embodiments, sampling signal 140 may be
generated by a Phase Locked Loop (PLL).
[0018] In recovering data from the serial data signal 130, sampling signal
140 causes CDR unit 125 to sample the serial data signal at sample times
coinciding with occurrences of a maximum Signal-to-Noise (S/N) level of the
serial data signal. Often, however, there is a phase offset between the serial data
signal and the sample signal, causing CDR unit 125 to sample serial data signal
130 at sub-optimal sample times, which can cause errors in recovering the data
from serial data signal 130.
[0019| Figure 2 is a block diagram of one embodiment of a CDR unit 125 of
receiver 120 in which embodiments of a phase interpolator may be used. Remote
serial data on line 210 is input to a phase detect circuit 220, which has as a second
input the recovered remote clock signal on line 290. A control signal from block
220, which represents the difference in phase, is an input to a remote clock
recovery mechanism 230, having as a second input a local reference clock on line
260. Local clock reference on line 260 may be the sampling signal 140 of Figure 1.
[0020] The control signal from block 220 is used to vary the phase of the
recovered remote clock until it is in a desired frequency and phase relationship
with the incoming data. The recovered clock on line 290 is provided as an output
and is used as a clock input to flip-flop 240 that is used to recover the data.
[0021] Figure 3 is a block diagram of one embodiment of a phase
interpolator. The phase interpolator 300 can be used as the remote clock recovery
mechanism 230 in a system such as that of Figure 2. However, the phase
interpolator implementation in Figure 3 is useful in any arrangement where
interpolation between different clock phases is desired.
100221 In the illustrated embodiment, the phase interpolator 300 includes
three circuitry units: a pre-conditioner 310, a mixer 320, and an amplifier 330.
Fixed phase clocks from a Phase Locked Loop (PLL) 350 are used as input clocks
to the pre-conditioner 310. These fixed phase clocks are converted to triangle
waves by the pre-conditioner 310. The pre-conditioner 310 is a square-to-triangle
waveform shaper, which generates good overlap between the two phases. The
output of the pre-conditioner 310 is a resistor-capacitor (RC) type of waveform
rather than a triangle. This basically provides a good region of overlap between
any two adjacent pre-conditioner output phases. The two output phases of the
pre-conditioner are coupled to a common mode circuit 360.
[0023] Common mode circuit 360 contains circuitry to keep the pre-
conditioner outputs biased at the same voltage. It forces the common mode of the
pre-conditioner outputs to be the same, and consequently helps the pre-
conditioner output signals swing around a fixed common mode to produce good
differential signal crossovers and ideal clock outputs. Implementing the common
mode circuit 360 may help produce proper phase spacing of the phase
interpolator outputs after amplification. The output of the common mode circuit
360 is coupled to a mixer circuit 320.
[0024] At mixer circuit 320, the signal phases outputted from the pre-
conditioner 310 are mixed proportionately based on proportionate current
weighing. The mixer 320 takes in any of the plurality of adjacent triangular waves
and mixes them to produce a resultant output that is close to sinusoidal in shape.
The output of the mixer is controlled to produce an output whose phase is
somewhere between the mixer input phases.
[0025] The proportionate current weighting implemented in the mixer
circuit is controlled by a Digital-to- Analog Converter (DAC) 340. The DAC 240
controls are generated based on edge and data samples, as well as the current
position of the sampling clock. Proportional weighting of currents between IQl
and IQ2, illustrated in Figure 3, determines the output phase. The sum of the
currents IQl and IQ2 is constant, K.
[0026] The output of the mixer 320 is analog and is fed to another common
mode circuit 370. Common mode circuit 370 operates analogously to common
mode circuit 360 to facilitate proper phase spacing of the outputs of mixer 320.
Embodiments of the present invention may not require both common mode
circuits 360 and 370. Some embodiments may implement only one of these
common mode circuits, while other embodiments may implement both common
mode circuits.
[0027] The output of the common mode circuit 370 is fed to an amplifier
330 that produces rail to rail (OV to VDD voltage swing) sampling clocks.
Amplifier 330 may in some embodiments be a CMOS (Complementary Metal
Oxide Semiconductor) level converter.
[0028] Figures 4 and 5 are graphical illustrations of phase spacing within
the phase interpolator at each of the pre-conditioner 310 input, pre-conditioner
310 output, mixer 320 output, and phase interpolator output (i.e., amplifier 330
output). The pre-conditioner 310 input clocks that come out of the PLL 350 are
routed several thousand microns and buffered periodically before they go to each
phase interpolator in a multi-lane configuration. Generally, these fixed phase
clocks at the input of pre-conditioner will have some cycle-cycle jitter from the
source and power supply noise. There is also a possibility of phase skew between
the adjacent clocks that are mixed.
[0029] Figure 4 illustrates ideal phase spacing within the phase interpolator
when there is no jitter or phase skew present in the input clocks to the phase
interpolator. Phase spacing at each of the pre-conditioner input, pre-conditioner
output, mixer output, and the phase interpolator output (amplifier output) is
shown. Ideally, the phase spacing between two adjacent output phases is T/n pS,
where T is the period in Pico seconds and n is the number of input phases within
each period. For example, the phase spacing between Phase 1 and Phase 2, and
Phase 2 and Phase 3, and Phase 3 and Phase 4, and Phase 4 and Phase 1, in an
ideal 4-phase system will be T/4 pS.
[0030] In Figure 4, line (a) depicts ideal pre-conditioner input phase
spacing. Line (b) depicts the resulting phase spacing at the pre-conditioner
output. Line (c) depicts the resulting mixer output phase spacing. Line (d)
depicts the overall phase spacing at the phase interpolator output (at the
amplifier).
|0031] Figure 5 graphically depicts non-ideal phase spacing within the
phase interpolator. Such a non-ideal scenario may occur when there is jitter
and/or phase skew present in the input clocks to the phase interpolator. In
general, the common mode and signal swing of the pre-conditioner outputs
determine the bias point of the mixer outputs. The bias point of the mixer outputs
is the pseudo common mode of the differential swings at the output of the mixer.
As the pre-conditioner clocks are displaced in time due to jitter and skew, the
outputs of the pre-conditioner can have different common modes. The different
common modes at the pre-conditioner outputs may bias the mixer outputs to
different voltage levels.
[0032] Furthermore, different common modes may cause mixer crossovers
to be significantly different from expected values and result in long pulse-short
pulse scenarios, thereby leading to duty cycle distortion in the phase interpolator
outputs. With phase skews in the pre-conditioner input clocks, the pre-
conditioner outputs swing around different common modes causing shifting of
mixer output bias voltages. With phase skews (time) present, mixer output bias
skews (voltage) result, leaving bad duty cycle clocks. In other words, poor phase
spacing of the output clocks results. Figure 5 illustrates the effect of non-ideal
phase spacing of the pre-conditioner input clocks on the phase interpolator
output.
10033] In Figure 5, line (a) depicts non-ideal pre-conditioner input phase
spacing. Line (b) depicts the resulting phase spacing at the pre-conditioner
output. Line (c) depicts the resulting mixer output phase spacing with skewed
bias points. Line (d) depicts the overall phase spacing at the phase interpolator
output (at the amplifier) with duty cycle distortion.
[0034J Figure 6 illustrates a circuit/block diagram of one embodiment of a
phase interpolator, such as phase interpolator 300 of Figure 3, used to alleviate the
duty cycle distortion in the phase interpolator circuit. The diagram includes
circuit level implementation of the pre-conditiĻ ner 310, the mixer 320, the
amplifier 330, and the common mode circuits 360, 370, as depicted in Figure 3.
[0035] The common mode circuits 360, 370 include an alternating current
(AC) coupling capacitor 610, 630, together with a common mode bias keeper
circuit 620, 640. AC coupling capacitor 610, 630 and common mode bias keeper
circuit 620, 640 operate together to maintain the common mode of the input
signals to the common mode circuit 360, 370.
[0036] A common mode circuit 360 may be located between the pre-
conditioner circuitry 310 and the mixer circuitry 320. Alternatively, in other
embodiments, common mode circuit 370 may be located between the mixer
circuitry 320 and the amplifier 330. One embodiment of the phase interpolator
may include both common mode circuit 370 between the mixer circuitry 320 and
the amplifier 330, and common mode circuit 360 between the pre-conditioner
circuitry 310 and the mixer circuitry 320.
[0037] The embodiment of phase interpolator 300 presented in Figure 6
may lessen differences in common mode bias voltages at the differential outputs
of the mixer 32O 7 keeping both outputs biased at the same voltage and thereby
producing proper phase spacing after amplification. The embodiment of phase
interpolator 300 also forces the common mode of the differential outputs of the
pre-conditioner 310 to be one value and the common mode at the mixer 320
outputs to be the same. Consequently, the signals swing around a fixed common
mode, thereby producing good differential signal crossovers and ideal clock
outputs of the phase interpolator 300.
[0038] Figure 7 is a flow diagram illustrating one embodiment of a method
700 for recovering an embedded clock from a data stream. Recovering the
embedded clock from the data stream includes the phase interpolator maintaining
the common mode of the signals it is processing. The flow diagram includes, at
process block 710, receiving a plurality of reference clock signals. Process block
720 includes maintaining a common mode of the plurality of reference clock
signals by a common mode circuit, the common mode circuit including an AC
capacitor and a common mode bias keeper circuit. Next, process block 730
includes generating interrelated control signals based on comparing the plurality
of reference clocks to a received data signal. Lastly, process block 740 includes
outputting amplitude contributions from phases of the interrelated control
signals.
[0039] Method 700 may be implemented in the embodiments of phase
interpolator illustrated in Figures 3 and 6. More specifically, maintaining a
common mode of the plurality of reference clock signals at process block 720, may
be implemented with common mode circuits 360 and 370 as depicted in Figures 3
and 6. Furthermore, the maintaining of a common mode of the plurality of
reference clock signals further includes forcing the common mode of the plurality
of reference clock signals to be the same.
[0040] Embodiments of the phase interpolator and its accompanying data
recovery scheme may be used in serial interfaces such as PCI Express. However,
the embodiments of the phase interpolator implementation presented here are
useful in any arrangement where serial data transfer over a networks system is
desired.
[0041] Whereas many alterations and modifications of the present
invention will no doubt become apparent to a person of ordinary skill in the art
after having read the foregoing description, it is to be understood that any
particular embodiment shown and described by way of illustration is in no way
intended to be considered limiting. Therefore, references to details of various
embodiments are not intended to limit the scope of the claims, which in
themselves recite only those features regarded as the invention.