[sdiy] Why dual grounds?

Wed Jun 9 19:12:00 CEST 2010

All good points Harry.  

[Marginally related non-sdiy story follows] 

When working at one company that built fiber-to-the-curb telephony equipment I got called to the lab once because a system test engineer was seeing low input voltage at the power supply input of an ONU. The ONU (optical network unit) is a box that sits in the outside plant, and is remotely powered from a network power supply over multiple twisted pair at +/-130VDC with a 100W safety limit. If the input voltage to the ONU power supply was too low it would not turn on.   I started tracing how he'd hooked up the ONU to the network power supply and noticed his hook-up wire was warm to the touch!  He'd used the wrong gauge wire and was dropping most of the +/-130V in I^2R loss in the hook-up wire!

Certainly an extreme case of 'modular' design gone wrong,  but the story points out that a wire, just like power distribution traces on a pcb, is not zero ohms, and if you put enough current through it you can get a significant voltage drop.

Then there's the story about the network power switching circuit that had an oscillation in the control circuit - the power FETs got so hot they'd unsolder themselves...

-Dave

 On Wed 10/06/09  8:38 AM , "Harry Bissell" harrybissell at wowway.com sent:
> ~most~ of the issues imho with power supply grounding come from
> MODULE DESIGN and POWER SUPPLY DESIGN.
> 
> A good module design (this is what I'd say 'Paul' does)
> does not corrupt its own ground or the grounds of other
> modules. In this case, two ground wires in parallel have
> a lower impedance, therfore better performance.
> 
> Bad board design includes sharing ground paths between
> low current (reference) circuits and power grounds (things the
> module designer KNOWS are going to draw high and transient
> currents). Other no-no's are running power traces to high
> current chips, then extending the same trace to a reference
> supply. (EFM VCO4d comes to my mind here).
> 
> A good module treats the power supply as a nasty source of energy
> that must be tamed and refined before it reaches any circuit that it
> could disturb. Use of adequate decoupling caps, NOT using the power
> rails for reference voltages, ~maybe~ ferrite beads or inductors, and
> on-card regulation are good techniques.
> 
> Good power supply designs have the necessary (I'll call it)
> 'oompf' to be robust in supplying power. Large enough caps, etc
> to handle all loads, including transients.
> 
> I've done designs that had all the jacks grounded to the front panel
> (most modular designs) and also all the jacks (save one) isolated from
> the panel, usually for stand-alone effects. Both techniques have their
> places.
> 
> I'd say that if you are a big fan of Vactrol designs, the separate
> powerand signal grounds are a really good idea (Buchla anyone?)... otherwise
> this is not needed
> 
> H^) harry
> 
> 
> 
> 
> ----- Original Message -----
> From: Dave Manley <dlmanley at s
> onic.net>To: Stewart Pye <stewp
> ye at optusnet.com.au>Cc: s
> ynth-diy at dropmix.xs4all.nlSent: Wed, 09 Jun 2010 11:11:29 -0400 (EDT)
> Subject: Re: [sdiy] Why dual grounds?
> 
> I can't speak for MOTM, but Paul Shreiber often posts to the list.  As 
> the Oakley pdf referenced by Graham indicates the two MOTM connector 
> grounds are joined on the board. I'm guessing Paul would say if your 
> board layout is correct, and your power distribution is good, then you 
> don't need separate clean and dirty grounds back to a star point.
> 
> -Dave
> 
> Stewart Pye wrote:
> > Since many of the modular synth standard power
> supply connectors > (MOTM, Oakley, Blacet etc) have two ground
> wires, is one of these used > for a "clean" ground and the other
> "dirty ?>
> > Regards,
> > Stewart.
> >
> > Dave Manley wrote:
> >> The way I've understood it: current spikes
> can cause the local ground >> reference to rise, since the ground
> connection isn't perfect and has >> non-zero impedance.  If you isolate the
> circuitry that generates >> current spikes (digital logic, LED drivers,
> 555 timers, etc) to the >> dirty ground, then the current flowing
> through the dirty ground can't >> affect the clean ground reference and couple
> switching noise into >> your audio path.
> >>
> >> If you connect the two on the pcb and there
> is a poor connection back >> to the power supply, then the ground at the
> pcb can get disturbed due >> to current spikes on that board.  On the
> other hand if you have a >> separate path for both grounds back to the
> power supply, then the >> clean ground stays quiet.
> >>
> >> -Dave
> >>
> >> David Ingebretsen wrote:
> >>> Thanks Paul. I've been reading about
> ground loops, star grounding, >>> and other
> >>> grounding concepts as I've been puzzling
> over this.>>>
> >>> David
> >>>
> >>> ~~ -----Original Message-----
> >>> ~~ From: synth-diy-bounces at dropmix.xs4all.nl
> [mailto:synth-diy->>> ~~ bou
> nces at dropmix.xs4all.nl] On Behalf Of Paul Perry>>> ~~ Sent: Tuesday, June 08, 2010 11:43
> PM>>> ~~ To: s
> ynth-diy at dropmix.xs4all.nl>>> ~~ Subject: Re: [sdiy] Why dual
> grounds?>>> ~~ ~~ As close to the power supply as
> possible.>>> ~~ So there is as little 'ground' in
> common as possible.>>> ~~ ~~ Another thing to read about in
> grounding , is "star grounding".>>> ~~ Plenty of references on the net for
> this ;D>>> ~~ ~~ paul perry Melbourne
> Australia>>> ~~ ~~ ----- Original Message
> ----->>> ~~ From: "David Ingebretsen"
> > Won't these two grounds eventually >>> have to be
> >>> ~~ connected together though?
> I>>> ~~ > guess that's my main question. I
> can layout the PCB with separate>>> ~~ grounds,
> >>> ~~ > I'm just baffled as to
> where/when these two grounds touch each >>> other.
> >>> ~~ >
> >>> ~~ ~~
> _______________________________________________>>> ~~ Synth-diy mailing list
> >>> ~~ S
> ynth-diy at dropmix.xs4all.nl>>> ~~ http://dropmix.xs4all.nl/mailman/listinfo/synth-diy>>>
> >>>
> _______________________________________________>>> Synth-diy mailing list
> >>> S
> ynth-diy at dropmix.xs4all.nl>>> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy>>>
> >>>
> >>>   
> >>
> >>
> _______________________________________________>> Synth-diy mailing list
> >> S
> ynth-diy at dropmix.xs4all.nl>> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy>>
> >
> >
> _______________________________________________> Synth-diy mailing list
> > S
> ynth-diy at dropmix.xs4all.nl> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy>
> >
> 
> _______________________________________________
> Synth-diy mailing list
> S
> ynth-diy at dropmix.xs4all.nlhttp://dropmix.xs4all.nl/mailman/listinfo/synth-diy
> -- 
> Harry Bissell & Nora Abdullah 4eva
> 
> 
> 
> 