[SI-LIST] Re: Chassis GND and Digital GND shorting on daughterboard
- From: "o. laney" <olaney@xxxxxxxx>
- To: shuklas@xxxxxxxxxx
- Date: Tue, 29 Jun 2010 00:45:40 -0700
The reason there are so many seemingly conflicting answers is, as always,
"it depends". If you are hoping for rules of thumb or accepted "best
practices" you will be disappointed. The only adequate way through the
EMI/ESD thicket is a good grasp of the fundamentals. For instance, there
are LF tricks that don't work at HF and vice versa and you need to
understand why. Ground partitioning and current steering are often not
well suited to cookbook solutions. Really, it is impossible to ask
general questions and obtain compact, one size fits all working answers.
You'll have to do the homework on learning the physical reasoning.
Another ad naseum discussion cannot substitute for that because it's a
fool's errand. Try the books on noise control by Henry Ott, and as
previously suggested troll the archives until you understand the
reasoning (both good and bad) behind various recommendations. Good luck,
let us know how you're doing.
Orin Laney
On Tue, 29 Jun 2010 12:39:44 +0530 "Shailendra Shukla"
<shuklas@xxxxxxxxxx> writes:
> facing problem with hyperlynx as attached
> >>> Ignas Mikulevicius <mikulevi@xxxxxxxxx> 6/29/2010 7:20 AM >>>
> Hello everyone,
> Yes, the topic's back. I know this has been discussed ad nauseum in
> various
> previous posts, but I felt none of them conclusively established a
> solution.
> First of all, from previous discussions it seems to me like there
> are some
> trade-offs between ESD and EMI protection in terms of chassis GND -
> digital
> GND connections. The way I understand it, for ESD, we ideally want
> to
> connect chassis ground and digital ground with a large resistor
> (say, 1
> MOhm), to allow any ESD charge to be bled off to chassis without
> causing any
> excessive currents and correspondingly dangerous voltages across
> components.
>
> On the other hand, for EMI, I understand it may be advantageous to
> short
> digital ground and chassis ground at multiple locations in the
> vicinity of
> the I/O connector to divert any problematic currents to chassis
> ground
> before they can radiate out on the cable. So is there any way to
> reconcile
> the two and get the best of both worlds?
> And how should these recommendations be applied when designing a
> standard
> PCI or PCI Express add-in card?
> The specific scenario I am interested in is this:
> Say I have a PCI Express daughterboard (add-in card) that plugs into
> a
> motherboard in a standard PC (let's call this System A). The PCB
> bracket is
> screwed on to the I/O connector of the PCB and, of course, also
> contacts the
> computer chassis. The board sends differential signals across a
> long,
> shielded cable to another system (say, System B).
> First of all, in the case of differential signals inside a shielded
> cable,
> how much potential for EMI can there be? I would assume the
> differential
> nature of the signals would greatly minimize radiation.
> The cable shield is shorted to chassis GND. Each differential pair
> in the
> cable has its own ground wire, and, finally, there is a wire that is
> called
> "Frame Ground", tied to chassis ground at system B. On the
> daughterboard,
> there is a "chassis GND" island underneath the connector in the
> layer below.
> The connector posts as well as the "Frame GND" signal are shorted
> here. So
> basically we have the chassis GND of system A and system B grounded
> and this
> also grounds the cable shield. Now clearly I can't leave this
> floating as it
> would seem that would create an EMI nightmare.
> What would be the ideal way to connect this island and digital GND?
> A
> capacitor, dead shorting, a resistor?
> It seems whatever method I choose I can find at least one article to
> endorse
> it.
> Any thoughts or insight would be greatly appreciated. I am not
> expecting
> someone to remedy my dillemma for me, but would really like to
> understand
> the theory and thinking behind this once and for all. If someone
> could
> either elucidate me or point me to some definitive, conclusive
> articles or
> books on this subject I would be very grateful.
> Thank you very much,
> Ignas M.
------------------------------------------------------------------
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 technical documents are available at:
http://www.si-list.net
List archives are viewable at:
//www.freelists.org/archives/si-list
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
