Re: [SI-LIST] : Q: Plane-jumping return currents

Doug (doug@dsmith.org)
23 Sep 1999 14:19:58 -0700

Hi Larry,

The board had solid ground planes and terminations for both paths. One path spent 1/3 of its length on the opposite side of the board. Granted there were no devices on the board to load it down. I was tempted to include a picture of the board and test results, but that would have resulted in too large a message. After the paper is published (or sooner if not accepted) I will post the material on my website.

Doug

On Thu, 23 September 1999, Larry Smith wrote:

>
> Doug - The 20 mV drop is in the capacitor/via/power-plane loop. If
> the supply voltage is 2 volts of so, 99% of the voltage is still
> across the capacitor and it's ESR. From an SI perspective, this is
> not too bad. But if there are 10 active vias instead of one, we
> have 10X the di/dt in the inductive loop, 10X the voltage drop (200mV),
> and the single discrete capcacitor does not seem very effective at
> keeping the power planes quiet. That is why the high quality
> (low inductance) capacitance of the power planes is so important.
>
> I agree that 20 mV can be an EMI disaster. To fully evaluate this
> problem, we would have to know the size of the power planes and the
> losses. The disturbance between the power planes created by signal
> via return current diminishes greatly as it propagates out radially.
> If the via continues to create disturbances at a repetitive frequency,
> it can set up resonances on the power planes. Bare power planes act
> like a pretty high Q structure. But when populated with ASICs and lots
> of decoupling capacitors, there is a fair amount of loss. So the
> 20 mV of local noise between power planes does not create as much EMI
> as you might think. If it did, our EMI problem would be a lot worse
> than what we observe with measurement.
>
> You have reported that a single trace produced a 26 dB difference in
> the emissions for a test board. Was the trace properly terminated?
> Did it have a good reference plane and path for return current for the
> whole length of the trace? I have always thought that a well
> terminated trace (either microstrip or stripline) with a good return
> current path should not produce much EMI.
>
> regards,
> Larry

-----------------------------------------------------------
___ _ Doug Smith
\ / ) P.O. Box 1457
========= Los Gatos, CA 95031-1457
_ / \ / \ _ TEL/FAX: 408-356-4186/358-3799
/ /\ \ ] / /\ \ Cell Phone: 408-858-4528
| q-----( ) | o | Email: doug@dsmith.org
\ _ / ] \ _ / Website: http://www.dsmith.org
-----------------------------------------------------------

**** To unsubscribe from si-list: send e-mail to majordomo@silab.eng.sun.com. 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/si-list ****