[SI-LIST] Re: Stitching Capacitors, Split-planes & Return Currents ?
- From: Patrick_Carrier@xxxxxxxx
- To: wangxiaoyun@xxxxxxxxx, sjulian@xxxxxxx
- Date: Thu, 28 Mar 2002 11:57:27 -0600
Simba--
The signals will use whatever adjacent planes are present as reference
planes, regardless of what their potential is. The trick is to create a
system which allows for a continuous current path for all cases of buffer
switching.
The drivers of your signal traces use a certain voltage for power. So, if
their signal traces induce a return current on one or more reference planes,
you need that return current to return to the voltage pin of your driver.
If you are using drivers which operate on 2V, and you induce return currents
on a ground plane, a 2V plane, and a 1.5V plane, you need to have a return
path connection between 2V and ground, 1.5V and ground, and 1.5V and 2V.
This would mean stitching the two planes with several small caps, and
placing decoupling caps from 2V to ground at both ends of the system, and
decoupling caps from 1.5V to ground and/or 1.5V to 2V at the one end of the
system that is using 1.5V as a reference plane.
I would suggest using more than one capacitor to stitch the planes, because
signals sharing a return current path (a capacitor) will have greater
crosstalk.
Hope this helps. --Pat
-----Original Message-----
From: Wang Xiao-yun [mailto:wangxiaoyun@xxxxxxxxx]
Sent: Thursday, March 28, 2002 3:39 AM
To: sjulian@xxxxxxx
Cc: si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Re: Stitching Capacitors, Split-planes & Return
Currents ?
Hello, Simba:
Are you sure the signals whose return current concerns are using the
power plane as reference rather than the continuous ground plane?
I'm doing design with voltage supplies ranging from 1.5V to 2.5V,
besides the main 3.3V, but those voltages are used as core power supply
for many large chips. It seems to me not a good idea to have core
voltage supply as a path for return currents. My understanding is when
you decide to jump the current from one reference plane to the other via
a cap as in your case, you have to provide another cap somewhere to lead
the current back to the original plane before it finally goes back to
the its source.
Moreover, quite a few chips I'm dealing with consume most of their
currents in cores. Connecting capacitors between different planes
provides paths to let the large current going anywhere it can. It does
not seem safe to me.
I'm also suffering from the fact that different power supplies exist
almost everywhere on my board, but my limited ken stops me from doing
what you are considering. I would also like to hear any gurus to provide
some better ways in dealing with the problem.
Regards.
Simba Julian wrote:
>Greetings to all,
>
>I've read numerous publications that have said not to route over split
>planes or to avoid splits at all costs. Well, we have dense boards that
>require a horde of reference voltages that would force me to have stack-ups
>greater than 20 layers. I must split these planes and in my case I have to
>send signals directly over these splits.
>
>My Question is.... If I have a 2.0V plane split with a 1.5V rail should I
>place two capacitors, one on each side of the split to the nearest ground
>plane. Or should I place one capacitor across the split connecting both
>Voltage rails (on the same layer). In both cases, we are assuming that the
>signals of concern are on the layer adjacent to the split.
>
>In my eyes the second case seems more attractive.... Lower capacitor count,
>half the inductance (ESL of one cap as opposed to two) for the return path,
>real estate. Nonetheless, are there any issues that I may be overlooking
in
>regard to capacitivly (is that how you spell it?) tying two power planes?
>Would it matter if the difference between the planes were larger.... Such
as
>5V and 2V? Are there any drawbacks to this configuration whereby the first
>scenario provides a better return/design?
>
>
>Secondly,
>My assumptions is that the largest capacitor with the smallest ESL would be
>optimal. Say ...... I'll use 0.1u for stitching, Is there any guide two
>determining the circle of influence for a capacitor based on capacitance
>(This should guide the number of capacitors needed for this application....
>And is probably useful for determining bulk capacitor count in the
>decoupling realm). I'll probably need a few since I have quite a few nets
>and different busses crossing this split, but I just wouldn't want to over
>do it!!!
>
>
>Thanks... I Really appreciate all responses,
>Simba
>
>
>
>------------------------------------------------------------------
>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 archives are viewable at:
> //www.freelists.org/archives/si-list
>or at our remote archives:
> http://groups.yahoo.com/group/si-list/messages
>Old (prior to June 6, 2001) list archives are viewable at:
> http://www.qsl.net/wb6tpu
>
>
>
------------------------------------------------------------------
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 archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
------------------------------------------------------------------
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 archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
