[sdiy] vco's

[Scott]

Hey, thanks for putting that into more words, because that is exactly what I was doing before I asked... I looked at various VCO's on my scope, and while looking and listening, there was not much (any that I could tell) difference.

Then, the very subtle differences....  we rarely use a raw vco signal, so the subtleties must be getting mostly covered up when wave folders and filters and fm modulation and everything else gets introduced.

-----Original Message-----
From: David G. Dixon [mailto:dixon at interchange.ubc.ca] 
Sent: Tuesday, April 07, 2009 12:31 PM
To: Scott; synth-diy at dropmix.xs4all.nl
Subject: RE: [sdiy] vco's

> Is there really much difference in sound between different VCO's?
> cgs vs motm vs ray Wilson vs Oakley, etc etc. (assuming they all have the
> same inputs, and same output wave forms, and not counting special VCO like
> quadrature, etc)
> I read a lot of stuff where people LOVE the sound of such and such VCO..
> But, isn't a 10v p to p saw wave (or whatever wave) the same regardless of
> which VCO its coming out of?

This is a question I've wondered about as well.  My VCO, which is basically
a Thomas Henry VCO-1 triangle core with added adjustable sync, switchable
edge/centre PWM (a la Oakley), JFET triangle-to-saw converter (thanks Scott
Gravenhorst) and OTA sine shaper (thanks Ray Wilson), produces virtually
perfect triangles, nearly perfect squares (made more so with the addition of
little speed-up capacitors on the comparator) and very sharp saws with just
the tiniest barely perceptible glitch spike in the middle, significantly
reduced from the original massive glitch by putting a 100p capacitor across
the diode I added which faces back from the JFET gate to the triangle
schmitt trigger.  The sine is "adequate", with a little angular bit at the
tops and bottoms (I replaced one of the three trimmers in Ray's circuit with
a resistor, the one on the OTA gain control, so this may be a factor -- I
probably should have done more research with a pot to determine the correct
resistance before committing to 560k).

It seems to me that if two different VCOs make identical waveforms on a
'scope, then they are producing the same sound.  One thing about the Thomas
Henry triangle core is that the discrete schmitt trigger (two npn's) does
not produce a perfect square wave, but the negative half has a downward
slope and a sharp negative peak just before going positive.  This trigger
output drives the OTA input to different polarities to change the direction
of the integrator, so one would expect this to create a non-linear triangle
up-slope.  However, the OTA is driven well beyond its linear range, as the
(roughly) +/-5V trigger output is only attenuated by a factor of ten before
being fed to the OTA input, and no linearizing diodes are employed.  (OTAs
give nonlinear responses to differential input voltages of more than a few
millivolts, and the input here is half a volt!) This ensures that the OTA
output current feeding the integrator is virtually a perfect square wave
(yes, I checked).

The extent to which any little glitches and non-linearities will affect the
basic sound of the VCO is anybody's guess.  I know that the little glitch on
the saw is quite audible (it sounds like an ant complaining, although the
capacitor has definitely appeased him) if steps are not taken to attenuate
it.  Also, I think that round-shouldered square waves can sound
significantly different than square-shouldered ones if no speed-up caps are
employed.  Also, a square wave comparator can feed glitches back to the
saw/triangle waves unless fed straight from the buffered opamp outputs
(i.e., prior to any output impedance resistors).  Finally, all bets are off
above about 12 kHz, where everything tends to fall apart (but at that
frequency, the damn thing is nearly inaudible anyway, especially to me).
Probably if I sprung for fancier opamps instead of using TL07x for
everything, this wouldn't happen, but I'm cheap, and I'm not interested in
making music that only dogs can hear!

Probably the biggest difference between VCOs is the tracking.
Notwithstanding the general issue of expo converter transistor matching and
tempco (yawn), I have heard that triangle core VCOs track less well than saw
core VCOs.  I can't say for sure, but my simulations indicate that my
triangle core VCO will have no difficulty tracking well.  I think that much
depends on the speed of the schmitt trigger at higher frequencies.  There
are also high frequency compensation schemes (basically feeding a very small
amount of expo current back to the CV summing amp) which firm up the
tracking at the top end, but I seriously doubt whether anyone other than a
dog could really tell, without testing equipment.

I would guess that any garden-variety VCO built today tracks better than
anything Moog could build in the '60s, just because our parts are better.
Probably the imperfections of those old circuits added to their charm.  I'm
always reminded, however, of the story told by Donald Fagen of Steely Dan.
They used synthesizers very sparingly, but decided to use one for a lead
line in a song called "King of the World" (GREAT song, BTW!).  This was in
1972.  They had such a frustrating time keeping the thing in tune that when
the session was over, they lobbed it down a flight of stairs and set it on
fire!