From: Michael Nudelman ([email protected])
Date: Fri Nov 10 2000 - 06:38:32 PST
My two cents.
Patrick:
Have you tried putting caps not across the split, but from the edge of every
split to the solid plane?
By putting them across the split you, though eliminating discontinuity,
defeat the effect of split plane (planes are split not just for the sake of
voltage difference but also for noise reduction, like analog/digital,
hi-speed/digital, etc), introducing the noise from more noisy plane to the
quiet one.
However, putting caps along edges from edges to Solid plane, you provide
almost just as good return path without coupling the noise.
Mike.
------- -||- ---------- versus -------------|
|----------
== ==
| |
----------------------- ---------------------------
"Zabinski, Patrick J." wrote:
> Ken,
>
> Prior to starting the experiment, I thought there would be some
> kind of relationship between capacitance and edge rate, but within
> the limits of what I had available, my original thought was wrong.
>
> To provide a little more background, the test board consisted of
> four layers:
>
> --------- ---- Split plane
> --- Upper Stripline
> --- Lower Stripline
> -------------- Solid Plane
>
> The striplines ran orthogonally to the split. The traces were
> 10" long, and the upper and lower planes were tied together
> with vias only near the end of the lines (thus, the two planes
> were connected 5" from the split).
>
> I left the soldermask off the split plane to allow easy addition
> of stitching caps, etc.
>
> We kept our measurements to TDR and TDT, with 35 psec, 250 psec,
> 500 psec, 750 psec, and 1000 psec (1 nsec) edge rates.
>
> The reason for the upper and lower striplines was to see if it
> made a difference. The theory going in was that the upper striplines
> were coupled much tighter to the split plane, so they would see
> more of an affect than the lower striplines. We quickly found
> the lower striplines to behave just fine with the split. In
> the majority of measurements, we could not even detect the split.
> However, the upper striplines demonstrated a strong coupling to
> the split plane, and we could easily see the discontinuities/coupling.
>
> >From here, we then tried to insert stitching caps along the split.
> I epoxied (non-conductive) four different value caps (0805 package
> size) to wooden sticks: 0.001, 0.01, 0.1, and 1.0 uF. I verified
> the caps were not shorted and kept their appropriate capacitance
> values, after which I played around a while. After much experimentation,
> I could not detect any difference between the effects from
> inserting the different capacitance values. This was quite unexpected.
> I then took the blade of a small screwdriver (calling this an "ideal"
> cap), and used it to short the planes at different locations. What
> surprised me even more was that the screwdriver produced the
> same effects as the caps. After I came to this conclusion, I again
> double-checked my caps to make sure they were "caps" and not "shorts".
>
> Anyway, within the limits of my experiments, I was not able to detect
> a difference in performance using different cap values.
>
> Pat
>
> > Very interesting. Did you select your cap value appropriate
> > to the edge
> > rate of the signals, or was it a constant value? I have not tried the
> > "mirror plane" underneath the gap before. Did you make the
> > Co of the mirror
> > plane consistent with the edge rate or not?
> > Ken
> >
> >
> >
>
> **** To unsubscribe from si-list or si-list-digest: send e-mail to
> [email protected]. In the BODY of message put: UNSUBSCRIBE
> si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
> si-list archives are accessible at http://www.qsl.net/wb6tpu
> ****
**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****
This archive was generated by hypermail 2b29 : Tue May 08 2001 - 14:30:03 PDT