From: Larry Miller ([email protected])
Date: Mon Jan 22 2001 - 14:02:00 PST
There is a big difference to the voltage regulator control loop between caps
that couple the voltage planes together by way of the ground planes (in
which case the caps are in series, but the ground --which is a signal node
in the voltage regulator circuit-- is in the middle) and caps tied directly
between the planes. The voltage regulators are comparatively low-frequency
circuits and are type 0 servo loops. If the caps are tied from voltage to
ground, the capacitor adds a pole to the voltage regulator loop equation
which reduces its bandwidth, and this is generally OK. Caps tied between
voltages are at least a lead term in the equations and also cross-couple the
regulator loops so that they interact-- foot in mouth disease, to use an old
scientific servo term...
Larry Miller
-----Original Message-----
From: [email protected] [mailto:[email protected]]
Sent: Monday, January 22, 2001 1:44 PM
To: Larry Miller
Cc: 'Zabinski, Patrick J.'; [email protected]
Subject: Re: [SI-LIST] : Decoupling between non-ground power rails, yes
or no??
If there are multiple voltage rails on a board that are properly decoupled
to
the same ground, there is a significant amount of capacitance between the
voltage rails, even without explicitly adding caps between them. Why does
that
not cause a problem?
Thanks,
Vinu
Larry Miller wrote:
> This instability is not a surprise. Putting caps between the voltages
> introduces a lead term (treble boost) into the voltage regulator feedback
> loop which can raise havoc by pushing the loop bandwidth out farther than
it
> was designed for.
>
> Offhand, I would say that this is overall a Bad Idea.
>
> Larry Miller
>
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]]On Behalf Of Zabinski, Patrick
> J.
> Sent: Monday, January 22, 2001 9:04 AM
> To: [email protected]
> Subject: RE: [SI-LIST] : Decoupling between non-ground power rails, yes
> or no??
>
> Michael,
>
> Not exactly answering your question
> directly, but I did want to bring up one cautionary
> note.
>
> In at least two past board designs in our group, folks
> have decoupled power planes directly to one another
> (e.g., +5 to +3.3) in addition to decoupling them to
> ground. In both of these board designs, oscillations
> were found in the board. When we removed one set of
> decoupling capacitors (i.e., breaking the GND-to-3.3-to-5-to-GND
> loop), the oscillations stopped.
>
> We did not have the time to dig into the root cause,
> but I have since made it a rule to avoid having strong
> decoupling between power supplies.
>
> Pat
>
> > Hi All,
> >
> > I recently saw something on a design that I
> > felt was questionable and wanted to see what
> > your thoughts were, especially as there appears
> > to be a number of folks that have a focus on
> > PDS and decoupling.
> >
> > While in the process of performing a PDS analysis
> > of a new board, I noticed that there was a signi-
> > ficant amount of decoupling between the power
> > rails. For instance, there were caps between 5
> > and 3.3, 3.3 and 2.5, etc. From discussions,
> > it appears that this is there to facilitate
> > return currents.
> >
> > So the obvious question is, Is this accomplishing
> > what it is believed to be and is there in fact a
> > better way to do this. Is a properly decoupled PDS
> > (below target impedance through frequency range)
> > all that is required. For the sake of argument,
> > assume that all the planes are whole.
> >
> > Thanks for your opinions
> >
> > Best Regards,
> >
> > Michael C. Greim Sonus Networks
> >
>
> **** 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
****
This archive was generated by hypermail 2b29 : Tue May 08 2001 - 14:30:42 PDT