Re: [SI-LIST] : Differential Clock Signal Pair

About this list Date view Thread view Subject view Author view

From: S. Weir (weirsp@atdial.net)
Date: Mon May 22 2000 - 13:51:18 PDT


Scott,

Those are some great points. I have spent time worrying about etching
differences, but not about localized differences in moisture content. I
did not mean to imply support for the notion that two single-ended signals
of equal length make a good differential pair. I think the basis of
misunderstanding here is that I was referring only to the strength of the
coupling, not the proximity of the two traces in the pair. I agree that
the closer we can get the two traces, the easier it will be to manage
outside disturbances, as well as static physical variations.

The issue I had with Jim's post is that I interpreted it, ( incorrectly? )
as asserting that it is acceptable to have both traces with substantially
different physical lengths, so long as the electrical lengths are matched.

> > >be true. If two differential signals have matched length but differ
> > >only
> > >in their spacings to the ground plane which is their signal return plane
> > >then
> > >the two signals will have different propagation velocities and hence
> > >different flight times even though their physical lengths are matched.

I interpreted that as stating that tight coupling could overcome dissimilar
gnd heights. It sounded like he was talking to some broadside coupled
arrangement, ( and I know how much you love broadside coupling ). Such a
situation would only aggravate all of the problems that you point-out.

I think that Lee's assertion that

> > > Differential signalling doesn't by definition, or any other rule,
depend on
> > > tight coupling for proper operation. Length matching is the only
> > parameter that
> > > really counts. Sure, you need to make sure that noise doesn't couple
> > into the
> > > pair differentially.

is true at least to the extent that it is not the mutual coupling that is
really making the diff pair work for us. The more I see of Lee's postings,
the more I believe he assumes that we have adequately accounted for:

  1. Offsets and interference so that we always get reliable crossings, and
  2. That the jitter that results from displacement of the crossing
threshold, ( flight time mismatch ) and induced noise is within acceptable
limits.

He was quite explicit pointing out to Weston the CM rejection limitations
of a coplanar pair against any nearby trace on the same plane.

The way that I interpret Lee's postings and his paper Feb 1999, is that
management of the above drives us to a particular amount of physical
matching of the pair which depends on the application.

I think that we all agree that faster and faster Tr applications drive
tighter physical matching, and this usually drives the coupling up. But no
matter how tight we try to make it, the mutual coupling itself remains
quite weak and limited as to what external disturbances it can mitigate. I
think your examples go to that point, since it is the proximity of the
traces to the same geographic features that sets the flight time, not the
coupling. I agree on your bet, although it is pretty tough in a 2mm world.

Regards,

Steve.

At 02:18 AM 5/22/2000 -0700, you wrote:
>Steve,
>
>The assumption here is that the dielectric is homogeneous for both
>signals of a pair routed the same length in different geographic
>areas. This assumption is false for FR4 in particular.
>(The dielectric constant of FR4 as seen by a trace is highly dependent
>upon the orientation of the trace over fiberglass bundles and the direction
>of the "grain". Since fiberglass is porous, the water content of
>different areas
>of the board varies. This variation can cause large differences in Er and
>the accompanying velocity of propagation.) It is more
>likely that the dielectric can be considered homogeneous for tightly
>coupled side-by-side differential pairs than it can for weakly coupled
>differential pairs. It is generally most homogeneous for broadside
>differential pairs, however, manufacturing tolerance issues preclude
>their use ... except for those who have the time and money to
>control the manufacturing process.
>
>Otherwise ...
>weak coupling will subject the pair to susceptability from other
>coupling sources. This coupling will cause differential skew.
>This differential skew can be decomposed into a differential
>component and a common mode component. The net effect is
>that differential skew due to either mis-matched trace velocities
>or additional coupling sources will cause voltage jitter in
>the received signal at the differential crossing. This jitter has the effect
>of reducing the CMRR of the system. Not to mention that it
>looks "yucky" on an eye diagram.
>
>Recently, we have worked with drivers that exhibit a launched edge
>rate of 110 ps into the center of a 50 ohm line. It only takes 60 ps of
>differential
>skew to never see a differenial crossing, no matter what the source
>of that skew is ... whether driver
> ... or unmatched lines
> ... or outside non-symmetric coupling source
> ... or receiver.
>
>Any skew causes reduction in receiver voltage margins and causes
>eye closure.
>
>I think I'll place my best design bets on tight coupling of side-to-side
>differential pairs with large gaps (3X or more) from one set of
>differential pairs
>to another.
>
>I happen to enjoy good clean differential crossings.
>
>
>best regards,
>
>scott
>
>
>--
>Scott McMorrow
>Principal Engineer
>SiQual, Signal Quality Engineering
>18735 SW Boones Ferry Road
>Tualatin, OR 97062-3090
>(503) 885-1231
>http://www.siqual.com
>
>
>
>"S. Weir" wrote:
>
> > Jim,
> >
> > How do you propose to get the differential coupling much above 15% so as to
> > make the differential coupling dominant over the single-ended coupling?
> >
> > I agree with your point that it is delay matching that we want, but I think
> > that with certain reasonably valid assumptions, Lee's assertion that
> > matched length realizes matched delay holds. Given that it would be very
> > naive to either mix a stripline and microstrip, or use unequal single-ended
> > impedances on the pair, aside from etching tolerances, the velocities will
> > be equal.
> >
> > Regards,
> >
> > Steve.
> > At 02:00 PM 5/19/2000 -0700, you wrote:
> > >You need to generalize length matching to flight time matching for this
> > >to
> > >be true. If two differential signals have matched length but differ
> > >only
> > >in their spacings to the ground plane which is their signal return plane
> > >then
> > >the two signals will have different propagation velocities and hence
> > >different flight times even though their physical lengths are matched.
> > >
> > >The easiest way to match the propagation velocities is to have the two
> > >signals tightly coupled.
> > >
> > >The tight coupling will dominate over the coupling to other conductors
> > >in the PCB. So this dominate tight coupling will determine the
> > >propagation
> > >velocity instead of the coupling to other conductors. The dominate
> > >tight
> > >coupling is inherently balanced between the two signals, hence the
> > >propagation velocities are inherently balanced. The same cannot be said
> > >for
> > >two independently routed signals.
> > >
> > >Also the variations in the PCB will usually affect both signals equally
> > >and
> > >hence equally affect their propagation velocities.
> > >
> > >Jim
> > >
> > >Ritchey Lee wrote:
> > > >
> > > > Differential signalling doesn't by definition, or any other rule,
> depend on
> > > > tight coupling for proper operation. Length matching is the only
> > > parameter that
> > > > really counts. Sure, you need to make sure that noise doesn't couple
> > > into the
> > > > pair differentially. Sid by side routing does not guarantee that on a
> > > PCB. On
> > > > the contrary, side by side routing in a PCB will aslways result in
> > > differential
> > > > noise coupling.
> > > >
> > > > Common mode noise coupling only occurs when the field strength of the
> > > inducing
> > > > signal is the same in both wires. This cannot occur in a PCB when the
> > > inducer
> > > > is on the same layer as the pair.
> > > >
> > > > Lee
> > > >
> > > > Weston Beal wrote:
> > > >
> > > > > Brian,
> > > > >
> > > > > By definition, differential signals should be tightly coupled. This
> > > implies
> > > > > close spacing. This does not necessarily mean that the seperation
> > > should be
> > > > > minimized. You need to do some analysis to decide the stackup,
> > > width, and
> > > > > space that creates the differential impedance that is correct for
> your
> > > > > design.
> > > > >
> > > > > Regards,
> > > > > Weston
> > > > >
> > > > > -----Original Message-----
> > > > > From: owner-si-list@silab.eng.sun.com
> > > > > [mailto:owner-si-list@silab.eng.sun.com]On Behalf Of Brian Seol
> > > > > Sent: Monday, May 15, 2000 9:38 AM
> > > > > To: si-list@silab.eng.sun.com
> > > > > Subject: [SI-LIST] : Differential Clock Signal Pair
> > > > >
> > > > > Hi everyone,
> > > > >
> > > > > I have a simple question about trace layout design for a differential
> > > clock
> > > > > signal pair of high-speed CMOS memory packages. I have two design
> > > > > guidelines for that as follows:
> > > > >
> > > > > 1. SPACING between a differential clock signal trace pair must be
> > > > > MINIMIZED as well as matched in length in order to reduce noise.
> > > > >
> > > > > 2. Differential clock signal trace pair must be matched in length in
> > > order
> > > > > to achieve matched electrical characteristics, but SPACING
> between
> > > > > them must be MAXIMIZED in order to reduce crosstalk noise.
> > > > >
> > > > > Which do you prefer?
> > > > >
> > > > > Thanks and regards,
> > > > >
> > > > > Brian
> > >
> > >**** To unsubscribe from si-list or si-list-digest: send e-mail to
> > >majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
> > >si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
> > >si-list archives are accessible at http://www.qsl.net/wb6tpu
> > >****
> >
> > **** To unsubscribe from si-list or si-list-digest: send e-mail to
> > majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
> > si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
> > si-list archives are accessible at http://www.qsl.net/wb6tpu
> > ****
>
>
>
>
>**** To unsubscribe from si-list or si-list-digest: send e-mail to
>majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
>si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
>si-list archives are accessible at http://www.qsl.net/wb6tpu
>****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Wed Nov 22 2000 - 10:50:26 PST