"Fleury, Bill" wrote:
>
> Judging from the number of responses (0) I got to my question about EMI
> problems from compactPCI backplanes I am assumiong that either no one uses
> cPCI backplanes or I'm the only one that has experienced problems when
> testing in this environment. I didn't think that ever happened in this
> group. I'll repost this just in case some of you didn't get the last one.
>
> This is a question regarding systems which use CompactPCI bus architecture.
> On a cPCI backplane the PCI clock is routed to every slot on the backplane
> with up to 7 peripheral slots possible on a normal backplane. My question is
> this: With the cPCI specification calling out specific trace lengths for
> clock traces on both the backplane itself and on peripheral cards, have
> other people experienced any unique EMI problems in trying to meet the
> design constraints associated with this synchronous bus; especially where
> there are unpopulated slots on the backplane?
>
> Please feel free to respond privately if the question is unclear. I wasn't
> sure if it was worded properly.
I didn't catch your former post.
As the PCI bus is a reflective based bus, in my opinion,
the PCI bus is one of the worst bus constructions ever
imagined. I've had the pleasant experience of trying
to cope with 132 MHz all over the damn place. For a
33 MHz based bus, yes, I'm talking 4th harmonics.
Harmonics can abound above that was well. Remember
that odd harmonics are only produced by a perfect
square wave. Any asymmetry to a square wave starts
including even harmonics. Add in some fast edge
rates adding to a bandwidth out to approx (0.35/Tr)
depending upon who you talk to and you've got a
potential rats nest.
Grounding the face plates of interface cards to the
chassis is especially tricky. If your box is using
regular style interface card sheet metal construction
that you would find on the back of any regular PC,
then you will have to make special care to inspect
the vertical rails that are supposed to be in contact
with the face plates. I know that's a mouthful, but
I hope you understand what I'm trying to say.
I've had to deal with a weird slot effect under some
circumstances with such a construction. A voltage
differential across the slot causes the slot effect
to radiate onto the cable making the cable *appear*
to be culprit. In my case, even though the cable
was the *radiator*, it's not the source of the
problem.
By simply trying to add a ferrite AND discovering
that the ferrite does little to solve the problem,
therein lies the hint that this slot effect is, well,
in effect and doing fine to make your life difficult.
Anywhere from about 6 to 10 dB of attenuation can be
had with a gasket. I can only reiterate and fully
agree with Mr. Cortland's comment regarding the
grounding provided by the chassis slot.
Bottom line, make sure your signal integrity
guys know what they're doing from an emi
viewpoint before you even start and make sure
to gain a close association with some unsung
heroes in the mechanical department. You just
may be needing them.
There. That's a total of about $10,000 worth of
advice, experience, theory, pain, and solution.
I expect my check in the mail by tomorrow ...
(just joking ...)
Regards and have fun, Doug McKean
-------------------------------------------
This message is from the IEEE EMC Society Product Safety
Technical Committee emc-pstc discussion list.
To cancel your subscription, send mail to:
[email protected]
with the single line:
unsubscribe emc-pstc
For help, send mail to the list administrators:
Jim Bacher: [email protected]
Michael Garretson: [email protected]
For policy questions, send mail to:
Richard Nute: [email protected]
