[sdiy] PCB layout question

[ASSI]

On Sonntag 02 November 2008, Seb Francis wrote:
> I thought bypass caps are more needed to stop switching transients
> spreading through the power supply.  And actually limiting the
> bandwidth helps this too.

A bypass cap proper delivers current when the transient is so fast 
that the inductance of the power supply network simply refuses to let 
it come through.  That is the sole purpose of these things.  Not 
having fast transients in the first place greatly diminishes the need 
for bypassing as well as a low-inductance power distribution.  If you 
go light on bypassing, it is still a good idea to check where the 
power supply network has its self-resonance (inject a squarewave and 
see if and where where it rings) and kill it with a few extra caps if 
necessary.  Unfortunately it doesn't need much gain to have a 
marginally stable part of the circuit go bonkers when incident RF 
from the outside hits the right frequency.

> For example, try a simple inverting summer with a fast opamp (e.g.
> OPA2132).  Bang a square wave through it and you will see it ring
> (on scope, you won't hear it!).  No amount of bypassing caps on the
> supply will change this.  But put a small cap in parallel with the
> feedback resistor and then ringing will disappear.   (The cap needs
> to be small enough to still keep the bandwidth well above audio
> range).

No amount of bypassing will help in that situation as it's the 
insufficient phase margin on the opamp that produces the ringing.  
While unity-gain stable, it is only marginally so as the phase margin 
on the OPA2132 at unity gain is just a measly 45° and to get rid of 
the ringing you need about 20° more.  Putting a cap in parallel with 
the feedback resistor is curing that by rolling off the gain much 
earlier.


Achim.
-- 
+<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk]>+

Factory and User Sound Singles for Waldorf Blofeld:
http://Synth.Stromeko.net/Downloads.html#WaldorfSounds