Re: [SI-LIST] : return current distribution in diff pairs

D. C. Sessions (dc.sessions@vlsi.com)
Mon, 27 Sep 1999 17:26:42 -0700

tomda wrote:
>
> I'm not sure I made myself clear. I will agree that pure differential
> signals with very high coupling will have minimal non-canceled ground
> currents. But the real world is not like that. There is skew in output
> buffers, see the [Diff Pin] section of the model. There will be skew
> introduced in routing a differential pair if not done properly from
> differences in line lengths, discontinuities, line coupling to other traces
> and variations in board properties. The rise/fall characteristics are
> going to be different, look at ECL which has an active pull up and a
> resistive pull down.
>
> The differential signal that you are hoping to propagate down a pair of
> wires will have both differential and common mode characteristics.
> Unfortunately, you are going to have to propagate both.

Perhaps more accurately, you *will* propogate both. It's not necessary,
nor desirable, but it is going to happen in the absense of specific
measures against it. An example of such measures would be a shared
ferrite bead (at least at frequencies where ferrite is useful.)

A while ago we had a thread on the effects of crosshatched grounds.
Common-mode filtering is one of the useful properties that such ground
patterns have since they preferentially increase the even-mode impedance
relative to the odd-mode impedance (they force ground-path currents to
take a long path but facilitate odd-mode coupling.) Of course, by
increasing the loop size for the common-mode signals they could increase
EMI but if the signal is heading into an unforgiving environment (such
as twisted pair or even just another ground system) it may well be worth
attenuating the common-mode signal first.

> -----Original Message-----
> From: Chris Cheng [SMTP:hycheng@3pardata.com]
> Sent: Monday, September 27, 1999 4:14 PM
> To: si-list@silab.eng.sun.com
> Subject: RE: [SI-LIST] : return current distribution in diff pairs
>
> that's why you need a differential receiver (not a complimentary
> receiver pair) to reject the common mode component. your
> differential signal portion will suffer ,independent
> of whether you are crossing planes or not, if you have edge
> asymmetry. don't mix the two effects up.

-- 
D. C. Sessions
dc.sessions@vlsi.com

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