## [SI-LIST] Re: 90 degree bend

• From: <jeff_latourrette@xxxxxxxxxxx>
• Date: Mon, 30 Aug 2004 12:59:22 -0600

```I remembered answering this at least twice a long time ago & dug this =
out from SI archives-it also contains an older reference.=20

Hope it can help,

Jeff LaT.
=20

[si-list] || [Date Prev] [08-2002 Date Index] [Date Next] || [Thread =

[SI-LIST] Re: PCB tracks
From: "LATOURRETTE,JEFF (A-SanJose,ex1)" <jeff_latourrette@xxxxxxxxxxx>=20
To: "'johnlipsius@xxxxxxxxx'" <johnlipsius@xxxxxxxxx>,SI-LIST =
<si-list@xxxxxxxxxxxxx>=20
Date: Thu, 8 Aug 2002 11:09:41 -0600=20
John, Philippe:

Another reason for 45 degree angle or curved traces instead of 90 degree =
bends=20
is to reduce what ends up being excess capacitance due to an abrupt =
bend.  This=20
capacitance occurs as the fields change from a nice even distribution on =

straight microstrip to very concentrated on the inside and less =
concentrated on=20
the outside of the bend.  This problem is more pronounced with wider, =
lower=20
(like 30 to 50-ohm) impedance lines, becoming much less significant at =
75-ohms=20
or 100 ohms. =20

If you are stuck doing 90 degrees and working at high frequencies, you =
can=20
miter (remove metal) at the bend to reduce  excess capacitance in your=20
high-speed signal traces.  You can also think of it as slightly raising =
the=20
impedance of the line at the bend, which when done correctly can =
significantly=20
improve return loss.  This is probably even harder for PCB houses than =
45=20
angles, so be sure your application really needs it.

As a quick check you can take that corner, mitered off, calculate its=20
capacitance and then reactance at your frequency of interest.  For most=20
frequencies below 1 GHz, this becomes insignificant and experiments can =
show=20
that mitering does little good at very low frequencies.  This technique =
is used=20
on most microwave designs where they are ringing out every last dB of =
Return=20
Loss performance from a design.  A lot of digital systems & components =
don't=20
even specify VSWR/Return Loss and may be able to tolerate more mismatch.

From an MTT article on how to optimally miter a microwave bend, mitering =
can=20
improve performance at frequencies as low as 1 GHz, so your comment on =
not=20
needing to worry at 200 MHz is probably right on.  It just depends what =
kind of=20
mismatches (VSWR) your system can handle, and how many harmonics above =
your=20
signaling rate you are designing to.

I think I posted this reference here once before, but for more:=20
=20
"Experimental Study of Symmetric Microstrip Bends and Their =
Compensation" =20
Douville & James, IEEE Transactions on Microwave Theory & Techniques, =
Vol.=20
MTT-26. No. 3, March 1978, pp. 175-181 =20

This article, due to its age, only covers out to 3 GHz, but the =
principles in=20
it are applied successfully up into 10's of GHz, although if you're up =
in that=20
range, you can easily verify the effect on your field solver.

I agree with John that the amount of energy launched into evanescent =
modes is=20
probably a 3rd order effect and not worth worrying about.

I'm not sure, however, I agree that you can ignore this at 10 Gb/s.  =
Since a 10=20
Gb/s signal carries with it some very high frequencies and seems to =
start to=20
tax the limits of some PCB materials, cables and connectors, I'd think =
you'd=20
want the best possible Return Loss in your signal traces.

Now you have \$0.03,

Jeff LaT.

=20
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field

http://www.freelists.org/webpage/si-list

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

List FAQ wiki page is located at:
http://si-list.org/wiki/wiki.pl?Si-List_FAQ

List technical documents are available at:
http://www.si-list.org

List archives are viewable at:
http://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

```