AW: Re: [AH] Regelwerk

[Haible Juergen]

> > I've just taken delivery of a Doepfer Regelwerk, and it is a great
 >> controller. 8 CV & gate outs!
 >> The one thing that bothers me is that if you connect more than one synth
to
> > a CV-out, the voltage drops and the synths cannot be tuned.
> > Does anyone else have this problem?
> > 
> Anyone who did this would have this problem ;-)

You only have this problem when the outputs are either
cheap, badly designed, or simply not intended for
precise V/Oct CV's. (I don't know the circuit of the Regelwerk,
so this is not meant as an offense. I'm trying to give a few
helpful hints *in case* that there is a problem with voltage
droop - for this unit, or for similar output stages. Wheather there is such 
a problem with the Regelwerk or not I don't know.)
 
> Any synth's CV input is a current drain; the higher the current drain,
> the more the voltage seen at each synth's CV in will sag.
>
> To solve this you can build a simple circuit based around an Op-Amp
buffer.
> I went off to search the net for such a thing, but came up blank -- anyone
> else got a good circuit reference on line?  Radio Shack has an Op-Amp
>cookbook
> that should get you going.

As someone said before, such a buffer should probably have been built into 
the box in the first place.
Admitted, these buffers are a little tricky. It looks easy at the first
glance:
An opamp buffer has such a low impedance that you can connect more
synths that you probably have, without any noticeable voltage droop.
But then there is a problem with capacitance: If you connect a long shielded
cable, the buffer will oscillate, and be worthless. You will run into the
same problem if you build an external buffer, btw. (But then it's *your*
problem,
not the manufacturer's anymore (;->) )
The cheap solution to avoid oscillation is simply adding a resistor at the
opamp output. This works fine if you don't need precision, or if you
re-calculate
(or simply trim) the gain of your output so that it compensates the voltage
droop
for one single load. I do this on all non-precision outputs - but if you use
it
for tuned 1V/Oct connections, it's bad design.

There are several ways to avoid these problems:
(1) You can use special opamps that are designed to drive capacitive loads.
(Check the data sheets !) These will work without the resistor. Problem:
They are a little more expensive than some other opamps, and some of
them are rather slow (slew rate!), so you might run into troubles if you
want to make audio rate FM from the same output. (This might apply
to an external buffer, not so much to the regelwerk, I guess.)
(2) You can build a special circuit that uses a resistor, too, but this time
it's
*inside* the opamp's feedback loop, so that the voltage droop is
automatically
compensated. The opamp needs a second (capacitive) feedback path then,
to ensure stability. Calculating the components is tricky. But you can get
around
if you use a prooven design with a certain opamp type and the fitting R and
C
values. I think there are examples of this on the Rane site.
(3) If you use some slow opamps (caveat see (1)) it might even work without
the capacitor. I have seen LM324's with 330R resistors (inside the feedback
loop !)
in some Serge designs, if memory serves. Don't know it still works on long
shielded cables, though. 

It's not a question of "designing around physics", as someone posted. It's
part of the price you pay when you buy low budged gear. You don't see things
like this from the outside, nor does it usually appear in the specs. But it
can make a difference.

JH.