RE: [SI-LIST] : Regarding plane splits

About this list Date view Thread view Subject view Author view

From: Jeremy Plunkett ([email protected])
Date: Tue Dec 05 2000 - 19:54:36 PST


David,
If the capacitance of one pair of closely-spaced planes
isn't enough for you, one option is to add more 2-3mil
dielectric layers as needed, alternating power and ground
planes in the stackup. Every additional power/ground plane
pair will double the capacitance and cut the supply
impedance in half over the frequency range where the PCB is
the controlling factor (above a few hundred MHz).

If you need more capacitance than you can afford to put in
your stackup and your plane noise is due to load transients,
capacitors located on-die or on the IC package will be
closer to the noisy power consuming silicon and can be
effective up to higher frequencies. That only works if you
are designing the ASIC and package in question, of course.

Jeremy Plunkett
Signal Integrity Engineer
Serverworks Corp.
[email protected]

-----Original Message-----
From: [email protected]
[mailto:[email protected]]On Behalf Of David
Kaiser
Sent: Tuesday, December 05, 2000 3:58 PM
To: 'Ritchey Lee'; [email protected]
Cc: Chan, Michael; Zabinski, Patrick J.;
[email protected]
Subject: RE: [SI-LIST] : Regarding plane splits

If the frequency harmonics are too high for the 2 or 3 mil
coupled
ground/power planes to provide good enough AC coupling even
with filler on
other layers, where do you go from there? Does anyone know.
Maybe Ritchey Lee can answer this. Thanks.

David Kaiser
McDATA Corp.
310 Interlocken Pkwy.
Broomfield, CO 80021
(303) 460-4431
[email protected]

-----Original Message-----
From: Ritchey Lee [mailto:[email protected]]
Sent: Tuesday, December 05, 2000 4:28 PM
To: [email protected]
Cc: Chan, Michael; Zabinski, Patrick J.;
[email protected]
Subject: Re: [SI-LIST] : Regarding plane splits

Properly done decoupling produces a very low impedance
between Vcc and
ground at all
of the frequencies involved in switching. Since most of the
edges are a
nanosecond
or less, this means a good plane capacitor. Very few, if
any, application
notes even
acknowledge that this is needed. Therefore, most engineers
don't get this
right and
have lots of high frequency noise on Vcc.

Lee

Itzhak Hirshtal wrote:

> Hello Mr. Lee,
>
> What do you mean by saying "the decoupling is not done
well"? Can you
detail what
> is the good way to do it?
>
> Thanks
>
> Ritchey Lee wrote:
>
> > That is a DC view of an AC problem. When the decoupling
is done well
enough
> > to provide the switching currents required to create the
fast edges, the
two
> > sides of the split are at the same AC potential, namely
that of the
underlying
> > plane. That's because you shorted them to this plane
with the
capacitors.
> >
> > Clearly, if the decoupling is not done well, this is not
true. Also, if
the
> > decoupling is not done well, there will be excessive
noise on both VCC
> > planes. This will show up in many ways, one of them
being potentially
high
> > EMI.
> >
> > Lee
> >
> > Chan, Michael wrote:
> >
> > > What happen if the split is for two different voltage
planes?
> > >
> > > MChan
> > >
> > > -----Original Message-----
> > > From: Ritchey Lee [mailto:[email protected]]
> > > Sent: Wednesday, November 22, 2000 11:39 AM
> > > To: Zabinski, Patrick J.
> > > Cc: [email protected]
> > > Subject: Re: [SI-LIST] : Regarding plane splits
> > >
> > > All of these discussions fail to take into account the
fact that the
> > > ground plane and the two power planes msut be well
decoupled in order
to
> > > create a low impedance source for the switching
currents that are
involved
> > > in those same swithcing edges. If this had been done
well, the power
> > > planes will, of necessity, be shorted to the ground
plane and that
will be
> > > the path around the split.
> > >
> > > In my experience, people who see the effects of a
split have failed to
do
> > > a good job of power plane decoupling. Learn how to
do this well and
> > > splits won't bother you.
> > >
> > > Lee
> > >
> > > Zabinski, Patrick J. wrote:
> > >
> > > > Aloke,
> > > >
> > > > As I mentioned in a recent posting, there are
conditions in which
> > > > routing stripline in a configuration you describe
sees no effects
> > > > from the split. More specifically, if your trace
layer is "closer"
> > > > to the solid ground plane than the split power
planes, then the
> > > > solid plane has "more" of an influence on the trace.
As such, you
> > > > will "less" of a discontinuity from the split. I've
tested this
> > > > in the lab under several conditions, and I believe
this to be true.
> > > >
> > > > However, the reason I used "'s in the above
statements is that I
> > > > haven't taken my experiments far along enough to be
able to provide
> > > > any guidance as to how "close" is "close enough" to
reduce the
> > > > discontinuity effects to the point where your system
can tolerate
> > > > them.
> > > >
> > > > In what I call a 50/50 case where the stripline
layer is centered
> > > > vertically in the stackup such that the distance to
the solid ground
> > > > plane is the same as the distance to the split
plane, you will
notice
> > > > the discontinuities.
> > > >
> > > > Pat
> > > >
> > > > >
> > > > > Hello all,
> > > > > I had a doubt regarding plane splits:
> > > > >
> > > > > In the stackup, if there is a power plane on one
side of the
signal
> > > > > layer and a ground plane on the other side of the
signal
> > > > > layer(symmetric
> > > > > stripline config), and if the ground plane is a
solid ground plane
> > > > > having no discontinuities, then can I have splits
in the
> > > > > power plane and
> > > > > run traces over the splits? Is the ground plane
alone not
> > > > > sufficient to
> > > > > provide paths for return currents?
> > > > >
> > > > > With regards,
> > > > > Aloke
> > > > >
> > > > >
> > > > >
> > > > >
> > > >
> > > > **** 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
> > > ****
> >
> > **** 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
> > ****
>
> --
> Itzhak Hirshtal
> Elta Electronics
> Ashdod
> POB 330
> Israel
> Tel: 972-8-8572841
> Fax: 972-8-8572978
> email: [email protected]
>
> --------------------------------------------------------
----------------
>
> Itzhak Hirshtal <[email protected]>
> Eng
> Elta
>
> Itzhak Hirshtal
> Eng <[email protected]>
> Elta
> Netscape Conference Address
> Netscape Conference DLS Server
> That's myself
> Additional Information:
> Last Name Hirshtal
> First Name Itzhak
> Version 2.1

**** 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
****

**** 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
****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Tue May 08 2001 - 14:30:21 PDT