This invention relates to a high-speed digital circuit, and more particularly, to a method and apparatus for adjusting for skew in digital data signals transmitted between devices in a digital interface circuit.
A high speed digital interface circuit typically has two independent devices, called a “source” device and a “sink” device, that communicate with each other across a bus having multiple data signal lines or channels. Individual data bits forming a digital data signal, when transmitted along the bus, take a different amount of time to reach the other end. The time difference between the “fastest” bit and the “slowest” bit is the bus's maximum delay skew. Excessive skew is problematic because it results in timing problems for signals intended for virtually simultaneous arrival.
Data rates of 2.5 Gb/s and higher are used between the source and the sink devices in high-speed optical interfaces. Since skew is a significant problem in a high-speed optical interface, a special skew correction circuit is often implemented in the sink device in order to compensate. Currently available technologies, which are capable of processing data at a speed up to 40 Gb/s, do not allow complex circuits (or multiplexing devices) to be constructed, thereby limiting the types of circuits that can be used to compensate for skew. In some instances, the source device and the sink device are implemented in different technologies, further limiting the opportunities for skew correction. The sink device can even be implemented in a technology that does not allow high complexity, even further limiting the circuits that can be used to compensate for skew.
In accord with the present invention, the skew in the bits of a digital data signal transmitted through an interface with a source device and a sink device is detected in the sink device and compensated for in the source device. The phase detection data for the individual channels of the digital data signal are fed back from the sink device to the source device as a multiplexed signal. A circuit in accord with the invention allows a larger number of data transfer approaches to be used in the sink device, as the circuit can be relatively simpler than in the prior art. This flexibility allows the system to be optimized for cost and performance. A circuit of the present invention is particularly suited for use in a high-speed optical transmission system such as a synchronous optical network (SONET), a key packet router or a dense wave division multiplexing (DWDM) system.
A circuit in accord with the invention allows an interface to be constructed for use with a high complexity source device and a low complexity sink device. Such a circuit allows an interface to be constructed even where the technology of the source and sink devices would otherwise generate too high an amount of skew among the bits of the data signal.
In accord with the present invention, a high speed digital circuit for use with an N bit digital data signal includes a source device that initially receives the N bit digital data signal, and a sink device that receives the N bit digital data signal from the source device, with the N bit digital data signal having a skew when received by the sink device. A skew detection circuit in the sink device detects the skew in the N bit digital data signal and generates a skew detection signal. A line supplies the skew detection signal to the source device. A compensation circuit in the source device receives the skew detection signal and compensates for the skew in the N bit digital data signal.
Further in accord with the present invention, a method of correcting the skew in an N bit digital data signal transmitted between a source device and a sink device of an interface comprises the steps of measuring the skew in the sink device, generating a skew detection signal in the sink device, transmitting the skew detection signal from the sink device to the source device, and compensating for the skew in the source device in response to the skew detection signal.
Still further in accord with the present invention, a high speed digital interface circuit for use with an N bit digital data signal comprises a source device, a sink device, and an N bit bus connecting the source device and the sink device, where the N bit digital data signal is received at the sink device with a skew. The source device comprises N variable delay units, each of the variable delay units individually controlled for each of the N bits, and a compensation circuit that compensates for the skew. The compensation circuit includes a counter circuit that supplies a reference signal to the sink device, an AND gate coupled to the counter circuit, and N low pass filters. Each of the low pass filters is coupled to one of the N variable delay units and to the AND gate. The sink device comprises N phase detectors, each of the phase detectors individually controlled for each of the N bits by the reference signal, N analog to digital converters coupled to the N phase detectors, and a skew detection circuit that detects the skew in the N bit digital data signal and generates a skew detection signal. The skew detection circuit includes a counter circuit that receives the reference signal from the source device, and an AND gate coupled to the counter circuit. A line supplies the skew detection signal from the sink device to the source device, and another line supplies the reference signal from the source device to the sink device.
Referring to the drawings, and initially to
The data bits D0 to DN−1 of the digital data signal D have a frequency of f bits-per-second, and are considered to be a continuous data stream. The transmission order of the N data bits D0 to DN−1 is defined by:
D0;m, D1;m, . . . , DN−1;m , D0;m+1, D1;m+1, . . . , DN−1;m+1,
where m is the mth set of N data bits in time, and (m+1) is the (m+1)th set of N data bits in time.
The reference line 18 carries characteristic timing information to allow for phase/delay detection, but is not part of the data stream of digital data signal D. The reference signal REF may, in one advantageous embodiment, be an alternating high-low sequence of bits having a frequency of f bits-per-second. Such a reference signal REF allows a phase detection of one (1) unit interval (UI).
The source device 12 receives the feedback signal FBC in an AND gate 32, where the bits Ri are extracted and assigned to the channels i, i=0, . . . N−1, under control of a counter 34 (CTRA). The bits Ri are supplied from the AND gate 32 through N individual loop filters 360, . . . 36N−1 (LFi, i=0, . . . N−1). The filtered digital signal from the loop filters 360, . . . 36N−1 are supplied as the control signals for the variable delay units 200, . . . 20N−1.
The counters 34, 30 in the source device 12 and the sink device 14, respectively, must be synchronized in order to assign the information for the bits Ri to the correct data channel corresponding to the bits Di. A preferred way is to make counter 34 a slave to counter 30, and provide a separate synchronization line 38 connecting the counters 34, 30. Other means of synchronizing the counters 30, 34 will be described hereinbelow.
It will be appreciated from the above description that the digital circuit 10 includes N data lines that are delay compensated, and that each line is individually delay compensated. The reference line 18 carrying the reference signal REF provides a timing reference from the source device 12 to the sink device 14. The feedback signal or channel FBC from the source device 12 to the sink device 14 provides information about the individual channel delays, or phase offsets, when the feedback signal or channel FBC is used in a time-shared manner between the individual data channels. It will also be appreciated that the source device 12 implements the delay compensation and control of the delay compensation to compensate for skew in the digital data signal D, and that this is the more complex part of the circuitry used to compensate for skew. The sink device 14, by contradistinction, implements the phase detection or measurement between the data line and the reference line, which is the less complex portion of the circuitry.
Specifically, in the initialization condition, an initialization digital data signal DetPat is supplied on the data channels or lines corresponding to the bits Di and the signal on the reference line or channel 18. The initialization digital data signal DetPat has a deterministic pattern of defined length that allows detection and compensation for delay variations of more than one unit interval (UI). In this instance, of course, the unit interval of the initialization digital data signal DetPat is greater than the unit interval of the digital data signal D, that is, UI.
In the circuit 110 of
If the circuit 110 is switched back into the operating condition, an “undefined” digital data signal resets the detection interval to one (1) UI. The variable delay units 200, . . . , 20N−1 determine the compensation range, however, as hereinbefore described.
It will be appreciated that the present invention allows relaxed timing requirements for the source device 12, and less complexity at the sink device 14.
Although specific embodiments of the present invention have been described in detail herein with reference to the accompanying drawings, it is to be understood that the invention is not limited to those precise embodiments, and that various changes and modifications may be effected therein by one skilled in the art without departing from the spirit and scope of the invention as defined in the appended claims.
Priority is claimed from provisional application Ser. No. 60/288,375, filed May 3, 2001.
Number | Name | Date | Kind |
---|---|---|---|
5568526 | Ferraiolo et al. | Oct 1996 | A |
6232806 | Woeste et al. | May 2001 | B1 |
6516040 | Lecourtier et al. | Feb 2003 | B1 |
Number | Date | Country | |
---|---|---|---|
20020179938 A1 | Dec 2002 | US |
Number | Date | Country | |
---|---|---|---|
60288375 | May 2001 | US |