[SI-LIST] Tightly coupled VS loosely coupled diff pairs

  • From: Chris Mesibov <Chris.Mesibov@xxxxxxxxxxxxxxx>
  • To: si-list@xxxxxxxxxxxxx
  • Date: Thu, 06 Sep 2001 11:23:29 -0400

Hello SI gurus,
I have been in a debate recently regarding how close differential signal
pairs should be routed in edge coupled stripline.  The debate is
centered around routing loosely coupled lines on a backplane (long
lengths) and potentially on a circuit card (short lengths). With the
benefits obtained from having tightly coupled lines when considering
cross talk , EMI and routing density, one might never consider spacing
the traces as far apart as lets say 5W.  The augment for loosely coupled
lines contend that because of this separation, 40dB of cross talk
isolation is can be obtained (for a given geometry, Dk...) and thus EMI
concerns are satisfied as well.  In addition, loosely coupled
differential transmission lines have lower loss per unit length, in the
order of .4 db/m (again for a given geometry).  This last argument
bothers me the most, as I have yet to see a direct comparison of test
results illustrating the performance of the two routing methods.
Moreover, the opposition to tightly coupled lines suggest that increased
current density occurs at the edges of the differential pairs, and that
this is the fundamental contributor to increased loss.  I know that
return current density is higher directly under the high speed trace.
It would certainly follow that in tightly coupled lines a larger
proportion of return current flows in the differential pairs as opposed
to the ground plane.  But it also is clear that the sum of all return
currents must be be equal to the source current, thus reducing return
current density in the diff pair return.  So how can increased current
density in the diff pair return be proportional to increased loss?  Does
skin effect further increase the loss?  If so, what is the nature of the
interaction (if any) between increased current density at the edge of
the diff pair traces and skin effect?
In exploring the loss argument, is there any data, studies or formulas
that illustrate these relationships?


 Chris Mesibov
 Sr. Hardware Engineer Fujitsu Network Communications, Inc.
 Phone:(845)731-2037   2 Blue Hill Plaza, 6th Floor
 Fax:  (845)731-2011   Pearl River, NY 10965

To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field

List archives are viewable at:     
or at our remote archives:
Old (prior to June 6, 2001) list archives are viewable at:

Other related posts: