RE: [SI-LIST] : Matched Length Traces

About this list Date view Thread view Subject view Author view

From: Loyer, Jeff W (jeff.w.loyer@intel.com)
Date: Fri Nov 10 2000 - 08:07:04 PST


Andrew,
I recently had an opportunity to study "meanders" (serpentines) when we had
a small board come back with flight time variations which couldn't be
attributed to differences in trace length (traces were matched within 5mils;
flight times varied as much as 50pS). Here's what I came up with...

____________________________________________________________________________

Here's a quick note on my initial conclusions about "meanders" and their
effect on flight times. It's not intended to be a comprehensive analysis,
but is a quick synopsis of what I've learned in the short time I've
investigated the phenomena.

Special thanks to those who responded to my e-mail question, especially the
pointers towards IEEE papers about the subject. The specific papers I found
most useful were:

00496044, Flat Spiral Delay Line Design with Minimum Crosstalk Penalty:
talks about how to create a flat spiral to replace the traditional
serpentine to reduce the problem. Gives a good qualitative (readable)
description of the problem. It deals with LONG traces, however (serpentine
lengths in the inch range), so the effect is not quite the same as what
we're seeing. I don't know about the effect of the spiral at the short times
we're interested in.

00853205, Characterization of Microstrip Meanders in PCB Interconnects:
describes the problem with shorter serpentines (0.4" range) in an
understandable format. Gives some recommendation. I'm not sure about one of
them - "Keep the number of meanders to a minimum. Fewer long meanders are
better than many short meanders." And again, I don't know if a flat spiral
(AKA "bifilar spiral") will help us.

00868994, Study of Meander Line Delay in Circuit Boards and 00475270,
Laddering Wave in Serpentine Delay Line: quantitative analysis of the
phenomena. Have all those funny greek symbols that make my head swim. I
didn't find any use in their conclusions. Pure candy for those who like
differential equations, poison for the rest of us.

00410843, Timing Skew of the Equal-Length Serpentines Routing: quoted as a
reference by everybody else. If you find anything of value in it, let me
know - I'm missing something.

Here's what I've gleaned so far...

Increasing trace length with serpentines (AKA "meanders") does not give an
increase in flight-time directly proportional to the increase in trace
length. Coupling across the serpentine legs causes part of the wave to
bypass the serpentine (I would refer to it as a "barreling through the
switchbacks" phenomena), reducing the flight-time. What's most surprising is
the absence of any mention of the effect, when layout guidelines talk about
matching trace lengths to 5mils! The speed-up effects are reproducible in
simulations and seem to be only weakly tied to rise-time.

The effect can be lessened by separating serpentine legs, or routing in
stripline. A "flat spiral" (AKA "bifilar spiral") may be an option to lessen
the effect, but this needs to be studied at the lengths/times we're
concerned with (pS). I don't know whether many short serpentines are better
or worse than a few long serpentines. My recommendations would be:

Take pains to avoid serpentines - they're not free.

If serpentining is necessary, follow ****** layout guidelines which try to
address the issue directly by recommending an S/H (trace separation to
dielectric thickness) ratio of about 5 to 1.

Route in stripline for high-speed signals. As I've worked with TDR, I'm
starting to adopt this mantra. I've consistently seen less flight-time and
impedance variations in stripline. Note that I've only compared true
stripline to microstrip; I haven't looked at dual-stripline.

If any of you learn more of the phenomena, please let me know...

Jeff Loyer
(253) 371-8093

-----Original Message-----
From: Andrew W. Riley III [mailto:drew@taloninst.com]
Sent: Thursday, November 09, 2000 3:55 PM
To: SI LIST
Subject: [SI-LIST] : Matched Length Traces

Hello SI-LIST members,

One issue that I am unclear on is all the turns in the traces shown in the
attached file 'Layer3.jpg' and in the picture Mr. Riazi attached for
reference to his post on "Plane Splits Inspection".

We have our designs sent to another company for Layout. In a section of the
layout instructions I specified net pairs to be of equal length with an
extremely generous tolerance (I think) of 0.200" but ONLY if excessive
bends were not added to the trace. The board is relatively small;
approximately 2"x6", and densely packed.

I also referred to UltraCAD's Design Note www.ultracad.com/t001.pdf under
the Radiating Points section, which states;
"Right angle turns and "T's": A trace that extends in a straight line is
relatively clean. One that extends straight and then turns 180 degrees back
on itself looks just like an antenna (like those on a tall building!)"

Layer3.jpg is a sample of what was sent back to me for review and is what I
thought we were trying to avoid. Also, I was somewhat surprised to see
similar routing in "Figure 1" attached to Mr. Riazi's initial post "Plane
Splits Inspection". This style of routing will not present any problems?

Am I wrong in thinking the routing shown in Mr. Riazi's and my picture are
examples of the 'antennas' mentioned in UltraCAD's Design Note? And to that
end, would someone be so kind as to explain why?
Any references supplied would be most helpful, too.

I will do my best to clarify anything that I neglected if at all possible.

**** 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 : Tue May 08 2001 - 14:30:03 PDT