[SI-LIST] Re: 100 ohm differential GHz trace width and gap
- From: pikeda@xxxxxxxxxxx
- To: si-list@xxxxxxxxxxxxx
- Date: Mon, 21 Jan 2002 16:08:13 -0800
Philip, Doug, James,
I am afraid I have to disagree with the last two responses. While I agree with
the fact that common mode rejection may be increased a little by routing traces
closely together, I believe there are more pertanant factors to consider. Common
mode rejection does only a little good unless the pairs are twisted, which of
course is impossible. After all a coplanar trace will always see one side of the
pair much more than the other will cancel it out. Simply having good spacing
between pairs, which it sounds like you have room to do, will eliminate any SI
concerns. Also be aware that any seperation of traces either at the component
pins, connector pins, vias, or to get around obstacles will result in higher
impedance at those points. For a GHz signal, it toesn't have to do that for too
long to be considered a transmission line, and then you will get reflections.
Also as James mentioned there is the skin effect problem. Plus if the traces are
very tightly coupled than the accuracy of the spacing becomes very important.
Remember coupling is squared inversly with the distance so just a couple of mils
off may throw your impedance out of tolerance. If you've ever tried to route a
trace on a diagonal with accurate spacing you would see it's not worth the
trouble.
On the other hand if the traces are loosly coupled than there are a number of
benefits. Routing becomes much more flexable, the traces may be given the same
geometries as 50 ohm single ended traces on the same layer, accuracy is more
easily attainable for both the layout person and fabricator, and the impedance
will be much more uniform along the length of the line.
The only time I like to closely couple traces is for backplane applications in
which case the environment is conducive to that if you are using differential
connectors.
Paul Ikeda
pwelling@xxxxxxxxxxxxxx on 01/21/2002 02:13:10 PM
Please respond to pwelling@xxxxxxxxxxxxxx
To: si-list@xxxxxxxxxxxxx
cc: (bcc: Paul Ikeda/Marvell)
Subject: [SI-LIST] Re: 100 ohm differential GHz trace width and gap
Doug/James,
I do the same.
I still prefer 50 Ohm Line-to-Ground & 100 Ohm Line-to-Line traces. Another
caution is if you fill the planes around the traces (for field containment)
you may have to calculate for co-planar trace impedance.
I also prefer to manage the traces near the pair (same layer) and traces in
adjacent layers. Keeping in mind what the return reference planes/path are,
I try to run these against ground planes and in a quiet area of the PWB
(sometimes this is not possible). Routing against power planes or routed
voltages adds common mode noise that you don't need especially in a receive
Differential Pair route.
Philip Ross Wellington
Mgr. Signal Integrity & EMI
L-3 Communications CSW
-----Original Message-----
From: James_R_Jones@xxxxxxxx [mailto:James_R_Jones@xxxxxxxx]
Sent: Monday, January 21, 2002 2:59 PM
To: doug.hopperstad@xxxxxxxxxx; si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Re: 100 ohm differential GHz trace width and gap
Doug,
I prefer less coupling to the ground plane, ie. narrower traces. My
motivations for this would be to have more coupling between the + and -
portions of the differential transmission line. There are common mode noise
rejection advantages, and improved AC return current paths. You should see
less SI degradation when crossing moats with tightly coupled diff T-lines.
One gotcha that I might warn about, however, is losses. If your traces
become to narrow, you may see some attenuation due to skin effect. The
usable area of the trace at high frequency is at the "skin" of the traces
and reduces with narrow traces. It should be possible to view this effect
when measuring narrow traces on a TDR. You will see a slightly rising slope
on a trace with consistent impedance.
I recommend using a good 2D field solver to get answers to specifics: Ansoft
2D or Apsim RLGC. Ansoft takes a bit longer to run, but has cool field
plotting options that helped me to visualize field directions and
magnitudes.
Hope this helps,
James R. Jones
-----Original Message-----
From: Doug Hopperstad [mailto:doug.hopperstad@xxxxxxxxxx]
Sent: Monday, January 21, 2002 2:49 PM
To: si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Re: 100 ohm differential GHz trace width and gap
I am interested in some feedback regarding routing 100 ohm differential
GHz signals on thin vs wide trace widths. I am currently looking using 8
mil traces with a 12 mil gap between the +/- pair. The routing is fairly
open and does not require very tightly coupled traces. I understand that
the tighter the traces are as a pair the common noise is improved. The
board material is FR408. A proposed core would use a dual-stripline with
a 7 mil dielectric between the planes (gnds) and routing layers and a 7
mil (or higher) dielectric between the routing layers. The routing is
orthogonal and tandem crosstalk is controlled. A stripline approach is
also an option if needed.=20
Doug Hopperstad
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
