[sdiy] decoupling on modulars power bus

[Ian Fritz]

At 02:13 PM 5/24/05, Harry Bissell Jr wrote:

>The power busses are probably fine. More
>likely the VCOs are defective of design.  Most VCOs
>assume that the power bus will be perfect, and trust
>the raw power rails for some mission-critical
>voltages.
>
>These are (in the usual sawtooth core) -
>
>1) Ramp Reset level (voltage divider from pos. rail)
>2) Expo converter reference (resistor to pos. rail)
>3) Initial frequency pot (tied to pos. and neg. rails)
>4) Any front panel controls (tied to pos. / neg.
>rails)
>
>If you inject ANY noise on these rails... you are
>hosed.
>
>These points should be connected to on-board precision
>voltage references (LM4040 for example).  Any
>variation
>of supply voltage will not directly affect the
>freqeuncy.  This is good engineering practice and is
>almost always overlooked.

Hear, hear!  This is exactly my experience.  The only thing I would add is 
to watch out for radiative coupling of high slew rate signals, such as 
reset pulses.  Use careful layout to keep these signals away from sensitive 
inputs, and use shielding or twisted pair if they have to run any 
appreciable distance.  Even think about limiting the slew rate of the main 
output signals (pulse, square, saw).  I had to fight these problems with my 
5Pulser waveshaper, which has square pulses up to 100 kHz.

   Ian