RE: [SI-LIST] : What's your favourite Screwy SI Concept?

About this list Date view Thread view Subject view Author view

From: Jian Zheng ([email protected])
Date: Fri Jan 14 2000 - 09:58:54 PST


Hi, Lee:

As I mentioned earlier, for a signal as high as 1 GHz, you do not need to
worry any problem with right angle bends if your traces are less than 200
mils from the ground plane. For a typical PCB, the trace to ground distance
is only 5 mils. You may not need to worry about the effect until 40 GHz. For
an edge as fast as 62.5 ps, 99.9 of the power spectrum should be below 15
GHz. For sure, the right angle bend (trace-to-ground distance of 5 mils)
will not have any problem. Thanks!

Best regards,

Jian Zheng

> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]]On Behalf Of Lee Ritchey
> Sent: Friday, January 14, 2000 8:53 AM
> To: [email protected]
> Subject: Re: [SI-LIST] : What's your favourite Screwy SI Concept?
>
>
> Are you talking about right angle bends in traces on PCBs or in
> wires suspended in space. If you are talking about traces on
> PCBs, Todd Hubing of UMR did
> a set of lab tests last year on atest design created by Doug
> Brooks. The results were published in Pritned Circuit Design
> magazine in early 1999. The
> conclusion is that right anlge bends in PCB traces do not
> represent a measurable source of EMI nor do the casue degradation
> of fast edge signals, at least
> with edges as fast as 62.5 pSEC.
>
> Why do people insist on making right angle bends a major villian
> in SI and EMI without some real proof that it is true?
>
>
>
> Jian Zheng wrote:
>
> > Hi, Doug:
> >
> > Following are the comments I would like to make on right angle corners:
> >
> > Normally, one bend should not create much radiation at relative low
> > frequency. The radiated power from a corner is normally very
> small compared
> > to the transmitted power when the trace to ground plane distance is much
> > smaller than one wavelength. Assuming your signal's frequency
> is 1 GHz and
> > you are using silicon substrate, one wavelength in free space is about 1
> > foot at 1 GHz. One wavelength in the silicon substrate is about
> 4 inches.
> > You do not need to worry about the radiation from the corners if your
> > substrate thickness is smaller than 0.2 inches (200 mils). The radiation
> > from a bend normall will be smaller than 1% of the transmitting power.
> >
> > The problem is many bends. If one corner radiates 0.1% of the
> power, then 10
> > corners will radiate 1% of the power. 100 corners will radiate
> 10% of the
> > power. Not only that, the biggest problem is the resonances
> caused by the
> > corners. When you have multiple corners, there might be some
> reflection back
> > and forth to create resonances. When a resonance happens, it is possible
> > much of the power will be radiated.
> >
> > Thanks!
> >
> > Best regards,
> >
> >
> ------------------------------------------------------------------
> ----------
> > ------------
> > Jian-X. Zheng, Ph.D
> > Zeland Software, Inc., 39676 Mission Blvd., Fremont, CA 94539, U.S.A.
> > Tel: 510-797-8109, Fax: 510-797-8241, Web: http://www.zeland.com
> >
> ------------------------------------------------------------------
> ----------
> > ------------
> >
> > > -----Original Message-----
> > > From: [email protected]
> > > [mailto:[email protected]]On Behalf Of Doug McKean
> > > Sent: Thursday, January 13, 2000 3:34 PM
> > > To: [email protected]
> > > Subject: Re: [SI-LIST] : What's your favourite Screwy SI Concept?
> > >
> > >
> > > Well, to continue, here's another one that has
> > > me scratching my head frankly.
> > >
> > > SI Concept: Avoid Right Angle Corners in Traces
> > >
> > > Underlying Concept: Right Angle Corners in traces
> > > decrease signal integrity/
> > > increase emi ...
> > >
> > > Why It's Used: There may be some truth to it.
> > >
> > > Now, as far as Maxwell goes, we should see an
> > > increase in radiation. But as far as I've read
> > > and experimented, I just did a little mockup in
> > > the chamber moments ago, the effect is minor.
> > > It may in fact signify nothing.
> > >
> > > For instance, I take a six inch piece of wire,
> > > stick it in one end of a barrel connector and
> > > stick the other end of the barrel connector
> > > to a coaxial cable which is connected to the
> > > tracking generator of a Spectrum Analyzer.
> > > This is my output.
> > >
> > > >From 0 to 1.3 GHz I see a particular maximized
> > > profile on the SA at 3 meters. I then force a
> > > sharp right angle in the wire and maximize again.
> > > The profile at 3 meters has changed but slightly.
> > > but, this is far field.
> > >
> > > I disconnect the antenna, disable the antenna
> > > factors, connect directly to the SA a near
> > > field probe and measure along the wire. At the
> > > bend of the wire when compared to a straight
> > > geometry, there is approx (very hard to tell)
> > > a +3 dB increase. And there could be a host
> > > of errors here. But I would expect some sort
> > > of change.
> > >
> > > Now this might not say anything for emissions.
> > > As far as a tdr in concerned, I doubt it would
> > > show a significant change in impedance to worry.
> > > BUT, the change in near field amplitude makes me
> > > suspect a corresponding increase in crosstalk
> > > (perhaps minor) say in a bus architecture. A
> > > tight high speed bus architecture.
> > >
> > > The above experiment strictly an observation.
> > >
> > > Comments? - Doug McKean
> > >
> > > **** To unsubscribe from si-list: send e-mail to
> > > [email protected]. 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
> > > ****
> > >
> > >
> >
> > **** To unsubscribe from si-list: send e-mail to
> [email protected]. 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
> > ****
>
>
>
>
> **** To unsubscribe from si-list: send e-mail to
> [email protected]. 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
> ****
>
>

**** To unsubscribe from si-list: send e-mail to [email protected]. 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
****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Thu Apr 20 2000 - 11:34:38 PDT