[sdiy] Nice info on decoupling caps

[David G Dixon]

Fair enough, Harry.  I generally put my decoupling caps right next to the
power connector on the module, so that any rail noise would be limited to
just that module.  I did have one LFO module which affected other modules,
but only if it was situated next to them on the panel...?  Otherwise, I've
never noticed anything being any better or worse when I've added or removed
decoupling caps.  However, your advice is obviously pretty sage.


> them's fightin' words, pardner....
> 
> (well OK not quite, but...)
> 
> A decoupling cap per supply pin on each chip will usually 
> prevent at least most crosstalk, you might even need some 
> series resistance or inductance or both (small resistor, 
> ferrite bead, etc.)
> 
> That is overkill but the only downsides are
> 1) increased cost, but probably not much...
> 2) extra board area required
> 3) extra inrush current from the power supply
> 
> the upside is
> 1) you will probably not have any interaction between stages 
> that you didn't intend...
> 
> Too much decoupling is usually not noticible from a 
> performance standpoint
> 
> Now if you have too few decoupling caps, the circuits will 
> probably interact in unusual and probably undesirable ways.
> 
> I once had some EFM modules on my bench... a diode VCF and a 
> VCO. When you turn up the VCF cutoff, the VCO pitch would change.
> 
> Woah what's up with that ?
> 
> The VCF was breaking into ultrasonic oscillation... and it 
> was being coupled through the power supply to the VCO reset 
> comparator... which reset early on the noise pulses.
> 
> Naturally, there were no 'excess' decoupling caps, and the 
> reset comparator was located closer to the power input and 
> farther from the one decoupling cap that ~was~ on that supply.
> 
> Adding a decoupling cap on the comparator voltage reference 
> pin fixed it. Using a voltage reference IC to generate 
> critical bias voltages fixed it even better...
> 
> My vote is that you should be liberal and generous with 
> decoupling caps unless you KNOW that they will not be needed.
> 
> (this might be different if you are designing a high volume 
> product where the pennies really count... if so you put the 
> caps IN and remove them until the performance suffers... then 
> add one bask in...
> 
> H^) harry
> 
> 
> 
> 
> ----- Original Message -----
> From: David G Dixon <dixon at mail.ubc.ca>
> To: 'Tim Parkhurst' <tim.parkhurst at gmail.com>, 'Synth-DIY' 
> <synth-diy at dropmix.xs4all.nl>
> Sent: Thu, 19 Jan 2012 14:54:04 -0500 (EST)
> Subject: Re: [sdiy] Nice info on decoupling caps
> 
> > A nice little intro on decoupling caps for the uninitiated. 
> > Some good links for more info at the bottom of the article too:
> > 
> > http://www.analog.com/en/all-operational-amplifiers-op-amps/op
> > erational-amplifiers-op-amps/products/RAQ_JBryant_In_and_Out_I
> > ssue76/resources/faq.html?display=popup
> 
> I found this article to be frustratingly terse.  "Many ICs 
> have internal circuitry which generates HF noise on the power 
> rails" doesn't really tell me anything!  Am I to believe that 
> an opamp generates HF, just because?
> Why?  If so, then wouldn't every transistor also require 
> decoupling?  Where does it end?
> 
> I tend to use only one pair of decoupling ceramics on each 
> module; two pairs if it is a big module, or if it has digital 
> stuff on it (which is fairly rare).  I've never had any 
> problem which could be solved with more decoupling.  Maybe 
> I've just been lucky, but I think that, for most analog synth 
> circuits, this advice of putting caps on every IC is simply 
> needless overkill.
> 
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
> 
> --
> Harry Bissell & Nora Abdullah 4eva