Fav. Designs

[Haible Juergen]

Wow, what a lot of great responses over night!

Paul wrote:

Ø	The best cap to use is a metallized polypropolyne.

Please, tell us more about this! What are the advantages? I used to 
think
that Polystyrylene (sp?) (or "Styroflex" as we call them here") would 
be the
best ones, at least in terms of temperature stability. (?)

Ø	A) Yes, in the Audio Amateur magazine. Everyone should subscribe. 
Best source of audio design out > there. Back issues available. 
AND...they sell KITS!!!!

Sounds interesting! How can I get a subsription (in Europe) ?

I think you also brought up the 3080's problems with CV feedthru. I 
got very encouraging
results with a pair of selected BC550C transistors (they are 10cent 
parts here in Germany,
and have remarkably good noise figure. Simply used them in a 
Minimoog-like discrete
VCA scheme. No audible pops even with full-level 1ms short envelopes 
!



JDM wrote:

> How many oscillators can do thru-zero PWM?  Thru-zero FM?  Not every 
VCO
Ø	design lends itself to these capabilities, does it?

Now this is one of the strong points of the CEM3340: You can build a 
thru-zero-
FM-VCO with only a handful of external components. The reprint of old 
Curtis
literature in Barry's book shows how to do it with a CEM3345, but it's 
almost
as easy with a 3340 as well.

Ø	But they distort in a really cool way.

Ok, sometimes you want clean multipying, sometimes you want "soft" 
distortion. In
VCF's for example, I absolutely want somewhat rounded transfer courve 
of a
non-linearized gain cell. I have my problems here with the CEM3320: 
Yes, it's
a very clean sounding filter, but if you overdrive it, distortion 
comes in too apruptly
IMO. I'm not sure if this comes from the use of (linearized) gilbert 
cells (can
someone confirm or correct this?), but I remember well how I built 
this SEM filter
clone for my Modular, trying to "enhance" it using 13600's with 
predistortion,
but finally cutting the resistors that feed the diodes out again, 
because it sounded
better.


Kimmo Koli wrote:

> A: Noise, more accurately phase noise !
>Noise in oscillators affect the oscillating frequency, so noise in a
>VCO is like small amount of FM with white or shaped noise. And VCOs
>are the noisiest frequency sources ever built, some may perform 
better
>and some worse. Also some amount of cross- and selfmodulation due to
>limited PSRR gives even more characteristics to a VCO sound.
>And phase noise has nothing to do (okay, some but not much) with long 
term
>drift. A VCO may be very temperature stable but can still have lot 
of
>phase noise. And even the quality of the PCB may affect the sound.

Maybe this would explain part of the VCS3's unique (?) sound. (I mean 
the
raw VCOs, with saw wave - of course there's a lot more that sets the 
Synthi
apart from other designs ...)
They use a simple 4-transistor expo converter. No opamp to in the 
usual
feedback loop which keeps the reference current constant, but just a 
single
transistor plus a second transistor connected as zener diode to keep 
the
expo pair's reference current constant. Now zener diodes are known to 
be
noisy as hell ...
Or think about all these first generation synths that used 741's to 
process
the CV summing etc.! Later publications pointed out that it does 
indeed
matter what opamp you use in the CV paths (and not just in the audio
paths) - and draw the conclusion that you would rather use better 
opamps.
But that's only part of the story ...
Maybe what we need is long term stability, but still a remarkable 
amount of
phase noise in VCOs ???  If you modulate a VCO with noise (like the 
Minimoog
or the Prophet 5 provides it), you need a rather high noise amplitude 
to get
this typical "thunder" sound. I think I will do some experiments with 
low level
noise modulation on a 3340 or other clean VCO on the weekend ...

Grant wrote about heated VCOs:

> By hand toggling the heater circuit to on/off I can pull down a
> high quality linear supply (<0.1 ohm output impedence) by ~20 mv
Ø	which in the circuit I'm using gives ~10 hz change.

The trick would be using a separate power supply path for the heater.
Not a separate psu, of course, but the heater current would get its 
own
simple regulation, and would be drawn from the unregulated supply.
It is as simple as this:

A little power transistor (BD139 or the like) has its base connected 
to the
regulated, precicion +15V (which supplies the VCO circuitry), the 
collector
is connected to the unregulated +20V (or similar), and the emitter 
feeds
the heater. (Don't forget a 100nF ceramic bypass there).
That's how all applications I know with the SSM2030 are built.


I highly enjoy this whole discussion!

JH.