Re: Differential Pair Skew

Don Abernathey (dla@pyramid.com)
Mon, 29 Jul 1996 12:15:50 -0700

Hello!

I would like to thank the folks who took the time to respond to my
questions on differential pair skew. I still don't understand the
effect skew has on differential signaling, but I'm getting
closer. I've put together a little Spice simulation to help me
visualize the circuit principles and various termination techniques.

I've included Rich and Mike's responses because they caused some more
questions. Please feel free to comment, as all input is helpfull.

--------------------Begin Responses--------------------

----------Rich wrote----------
Hi Don,

Skew kills the eye, Different modal velocities will also kill your eye
too, This is especially true for high data rates. You can prove this
to yourself with some simple graphing and waveform math. The effect
looks like edge dispersion.

... Rich Mellitz, NCR

Question/Comment - The bottom waveform would be the output of a
differential amplifier (ignoring amplitude). The output is skewed from
the input, but original bit cell time is preserved. If I did a BER eye
diagram using a differential amplifier and a scope the skew in this
case would have no effect on the eye. The only effects on the eye
would come from differences in rise/fall times, differences in Voh and
Vol between drivers, and jitter (ignoring loss).

Differential Pair with skew
+ ------\ /-----------------------------\ /---------
\ / \ /
X \ /
/ \ \ /
/ \ \/
/ \ / \
- ______/ \________________________/ \_________________
^ ^
| |
| |
Output of Differential Receiver
/---------------------------------\
/ \
/ \
____________/ \______________

----------Mike wrote----------
Sometimes it seems skew is given too much importance, but it does
have a number of subtle effects. While the differential signal
timing is theoretically insensitive to small skew, a common mode
signal is created. This is certainly a consideration in receiver
design, and needs to be limited by specification.

A second effect is on the dif'l risetime. Skew will "round off"
the corners of the differential signal, increasing the risetime.

A third effect is another aspect of the resulting common mode
signal. Perfect differential signals have the virtue of creating
zero net ground return current, so connectors with poor (or no)
ground connection and PC boards with split power/ground planes
are not a problem. Skew can turn these into real design issues.

Regards,
Mike Jenkins

Question/Comment - The output risetime of a differential receiver is
determined by the gain, output buffer, and output loading
characteristics. Of the three, only gain relates to the differential
input. Therefore the slew rate of the differential inputs effects the
rise time of the differential receiver output. Loss of energy in the
harmonics will result in decreased slew rate. Skew will not effect
differential receiver output edge rate.

--------------------End Responses--------------------

*************************
Thank you |
Don Abernathey |
(503)690-6234 |
dla@pyramid.com |
*************************