[sdiy] dare I say - decoupling?

[James Patchell]

I have never found this nessesary.  In some cases...it might cause more 
problems than it is worth.

What I try to do is contrive to not have any large currents flow in the 
ground circuit.  Sometimes this is not possible (like in the case where you 
use digital IC's which have to run from +5 volts...or less).  You just need 
to be very careful ...although...that only comes with experience...

If you are going to have a large amount of digital ICs (74xxXXX type 
stuff)...you might want to have a digital and analog ground...these need to 
be tied to a single point ground somewhere...but again...I have generally 
found this is not needed....if you are careful.

At 01:28 PM 9/3/2005 -0400, Michael Ruberto wrote:
>For my new synth I want to use a clean ground + dirty ground setup. What 
>are some guidelines as to which devices use which ground? Especially in a 
>discreet VCO...
>
>Also, since I am planning to have no less than 5 VCOs what should I do to 
>prevent soft-syncing (ground coupling?).
>Maybe it would be helpful to me and others if an overall system grounding 
>strategy was outlined.
>
>I noticed some people use resistors on the power rail inputs for each 
>module. What exactly is gained from this?
>would it be a good idea to apply it to all modulation sources?
>
>Thanks, keep the info rolling, I'm learning alot already!!
>
>M. A. Ruberto
>
>
>
>>From: James Patchell <patchell at cox.net>
>>To: "Michael Ruberto" <frankentron at hotmail.com>, 
>><synth-diy at dropmix.xs4all.nl>
>>Subject: Re: [sdiy] dare I say - decoupling?
>>Date: Sat, 03 Sep 2005 09:46:49 -0700
>>
>>And just to reinforce what Paul says...
>>
>>If a rail to rail bypass seems to work better than rail to ground...you 
>>better look at your grounds!
>>
>>I did a linear power amplifier a few years ago that had a persistent 
>>oscillation.  Replacing the bypass cap from rail to ground with a rail to 
>>rail got rid of the problem...sort of.  Despite the ground traces being 
>>0.2 inches wide in 2 oz copper...they were inadequate for the job...the 
>>rail to ground just made matters actually worse...but what this was 
>>telling me was "Hey, you dummy...you need to add another layer that is a 
>>ground plane!"...which I did, finally (I didn't want to...this increased 
>>the cost)...but the ground plane solved the oscillation problem and the 
>>rail to ground bypasses worked like they were supposed to.
>>
>>Harry likes to use the analogy of "Ground == Ocean"....very important 
>>here...In general, you do not need a ground plane.  But when things don't 
>>seem to be working they way they should...well...
>>
>>By the way, that power amplifier was for a servo system and would put out 
>>about +/- 45 volts at 27 amps with a bandwidth of about 20KHz...I might 
>>have been expecting just a bit too much from .2 inch wide ground traces 
>>(they were not expected to carry 27 amps, but they were the return for 
>>the bypass caps on the rails which did carry that current....).
>>
>>-Jim
>>
>>At 10:46 AM 9/3/2005 -0500, Paul Schreiber wrote:
>>>Rail-to-ground.
>>>Rail-to-rail doesn't do *anything* worthwhile. In fact, it can actually add
>>>noise :)
>>>
>>>Paul S.
>>>
>>>
>>>----- Original Message ----- From: "Michael Ruberto" 
>>><frankentron at hotmail.com>
>>>To: <synth-diy at dropmix.xs4all.nl>
>>>Sent: Saturday, September 03, 2005 10:28 AM
>>>Subject: [sdiy] dare I say - decoupling?
>>>
>>>
>>>>Well we know who the decoupling experts are here, and they do shine so 
>>>>brightly. At the risk of sounding like the moron I am I must pose 2 
>>>>questions to the decoupling kings.
>>>>which is better, decoupling caps from each rail to ground, from rail to 
>>>>rail, or both?
>>>>what's the difference?
>>>>what is the air-speed velocity of an unladen swallow?
>>>>thanks in advance oh wise ones ;-)
>>>>M. A. Ruberto
>>
>>         -Jim
>>***************************************************************
>>http://www.oldcrows.net/~patchell
>>
>>***************************************************************
>>
>
>

         -Jim
***************************************************************
http://www.oldcrows.net/~patchell

***************************************************************