[sdiy] Decoupling caps in Moog modulars: "Living VCOs" PCB in 2009

[Rick Jelliffe]

There seems to be four views, not necessarily mutually exclusive:

* a "living" oscillator has good bass power, linear tracking better than 
5 8ves and linear detuning (JH)

* some human ears are amazingly sensitive to some small details

* some VCOs have good frequency stability over time when measured in the 
absense of other activity (eg Ken)

* power supply effects and other pathologies can effect VCOs in a loaded 
operational system (eg Barry).

Recently I looked through some discrete circuit synth schematics, to try 
and find any pattern to the decoupling capacitors, including the 
capacitors used for biasing. Modern designs with op-amps seem very 
uniform: 10uf on the board and .1uF at the chip if there are enough. 

Here are some for Moog modulars (I think these are fairly late versions 
of the designs, since some have 3080 and 3046):

  * 909/910 Power supply has a 3000uF cap after the rectifier, and 100uF 
caps on the +12, -6 (regulated ) and +10 (unregulated) lines.
 http://www.emusic-diy.org/MoogManuals/MoogModular?action=AttachFile&do=get&target=mod_23.gif

  * 901A Oscillator controller (expo converter) has no power supply 
stabilization or decoupling
http://www.emusic-diy.org/MoogManuals/MoogModular?action=AttachFile&do=get&target=mod_11.gif

  * 901B VCO has no decoupling on the +12 input, a 100uF cap on the 
derived 11.5V used for the reference voltage for tracking, and 400uF on 
the -6.
http://www.emusic-diy.org/MoogManuals/MoogModular?action=AttachFile&do=get&target=mod_14.gif

  * 902 VCA has no decoupling on the power supply inputs but 400uF on 
the negative supply and 80uF on the positive supply on some derived supplies
http://www.emusic-diy.org/MoogManuals/MoogModular?action=AttachFile&do=get&target=mod_16.gif
 
 * 904A VCF has no decoupling on the input supply lines but a 400uf cap 
on a derived negative supply for the output biasing 
http://www.emusic-diy.org/MoogManuals/MoogModular?action=AttachFile&do=get&target=mod_18.gif

 * 911 EG has no power supply stabilization or decoupling 
http://www.emusic-diy.org/MoogManuals/MoogModular?action=AttachFile&do=get&target=mod_24.gif
 
 * 3A control panel mixer: 0.01UF on + and -15.
http://www.emusic-diy.org/MoogManuals/MoogModular?action=AttachFile&do=get&target=mod_07.gif

Here is a good example:

 * 903A random noise generator has two noise section, each with derived 
power: so 4 400uF caps!
http://www.emusic-diy.org/MoogManuals/MoogModular?action=AttachFile&do=get&target=mod_17.gif

 * 923 filter/noise source has .01uF caps on +12 and -6, and two lots of 
220V on the +/-11.7 and +/-5.7 derived supplies

As you would expect from transistor designs, where there are a lot of 
voltage dividers for biasing purposes, the decoupling is based on 
providing power and stability where it is needed:, but it is interesting 
that different value caps are used for different positions. For example

* 904B VCF has 8ouF on the +/- 12V, 200uF on the +2.5V.
http://www.emusic-diy.org/MoogManuals/MoogModular?action=AttachFile&do=get&target=mod_19a.gif

So if we look at these designs and compare them to modern designs, we 
see that supplies for control signals (the negative supplies) tend to 
have caps in the 200-400uF range, while voltages for audio signals (i.e. 
the positive supplies) tend to have caps in 80-100uF range. This is a 
decade or two more than the caps in modern designs, where you tend to 
see a lot of 10uF and 0.1uf (with 0.01uF for logic).

I wouldn't be as bold as saying what effect this cap values regime has 
on a large running system with lots of interconnections, or indeed that 
it necessarily has any effect, but it is a difference compared to modern 
designs at the least.

And it is interesting that on the one hand we have JH saying that modern 
designs don't handle bass and others saying the modern designs are 
somehow not as interesting-sounding, and on the other some of the older 
designs with larger-than-modern values for the decoupling caps (which 
might provide more juice for lower notes but do less smoothing for 
higher audio frequencies.) And does 901A having no decoupling make it 
more susceptible to the potential interferences that Barry Klein 
mentioned?  Don't flame me, I am not saying it is necessarily so!

On the same trail, the ARP 2500 modules mostly have only 1uF caps on 
their +/-15 power supply: Again, not the 10V or .01V values so common 
now. And the EMS VCS-3 boards seem to have no special decoupling at all.
   http://www.guitarfool.com/ARP2500/images/1023_sch.jpg

I once read a hi-fi article which recommended decoupling caps for each 
decade: so if the post-recitifer caps were 3300uF, then 330, 33, 3.3 and 
.33 etc. I'd love to know whether people had any opinion on this.
 
Cheers
Rick Jelliffe