New version and (Re: Haible_Juergen's comments on VCO)

Rick Jansen sscprick at
Mon Jun 12 13:20:28 CEST 1995

Juergen Haible wrote:

> (2) The electronic switch that discharges the cap is controlled
> by 0 ...+5V, but it switches signals in the +5 ... +10V range.
> Are You sure the switch You use is capable of that?

I thought so, but the data sheet is a bit confusing. The HC(T)4066 
can switch voltages up to 12V, but it may be that this means voltages
between -6V and +6V. In the latest version the cap voltage is 
less than 5 V, with control signals of 5V.

There is also a 4316 quad switch, which contains a level translator
so you can switch analog voltages outside the range of the digital
control signals. But I have no data on Ron for this ic, so I stick
to the HC4066.
> (3) Personally, I would avoid coupling capacitors where ever
> possible. You have so many components already, why not
> adding a little level-shifting instead of just AC-coupling?
> Every VCO can be a very useful LFO ...

I've thought about this too. Using a capacitor seemed so easy.
I could of course shift the sawtooth down. Is this what is done
usually?  The new circuit uses level-shifting.

> (4) I would replace the current source by a simple resistor.
> I'd invest in a good on-board voltage regulator instead.

I now use an LM329 (6.9V) reference and fix all voltages
with it, i.e. the reference current, and both comparator levels.
(I'll go for the two-comparator solution, one comparator detecting
the upper, tho other the lower level.)

A bit controversial may be that the power supply for the digital
(sorry!) chips is derived with an opamp from the reference voltage.
> (5) You could implement linear FM by simply adding one
> resistor ... but You surely know that!

Uh, not really in this schematic? You'd have to introduce control
after the exp generator, where would you do that? 

> (6) The heater, the heater!!! No I know what someone recently
> meant with "flametorch"! You have such a good heat regulated
> circuit in Your VCF - based on this omnipresent application note -
> Why not use it for the VCO? 

Well, the sole reason is that the 3046 transistors do not have
great specs for a *very* good exp generator. The bulk resistance is
rather big, and the current gain is only about 100. So, I decided
to go for a really good matched pair like the LM394 or MAT-02.
I think flametorch is a bit overstated :) 's just that that 
piece of aluminium acts as a stabilizing factor in the whole process,
like a flywheel. So quick variations in the environment have no effect.
I'd have gone for a temperature of, say, 45 degrees. 

Now I know there is a MAT-04 I have gone back to the transistor array
version with substrate heating. The current circuit is a LM3046 one,
but is easily modified for MAT-04.

> (7) Unfortunately, You also have used the same tuning scheme
> as in FORMANT. Offset tuning and scale tuning are very, very dependant
> on each other, and this almost drove me crazy.

True, I have experienced this too. Although the real value
for the offset in the Formant VCO's isn't that important, as 
you'll tune them to any pitch you like. I haven't bothered getting
the same frequency for exactly the same position of the coarse 
frontpanel potmeter for the different VCO's. 
> The triangle converter looks interesting. Where do You have the circuit 
> from?
> How does it work, exactly? (must be a resistor to gnd at the input, btw)

It's a circuit that is in both S. Franco's Design with Opamps and
W.G. Jung's IC-Opamp Cookbook. It's a circuit evolved from others,
I added a short explanation at the bottom of the wave shape converter.
This is a unity-gain saw-to-triangle circuit.

There are two new gifs on the web page depicting the current state,
again, comments of course very welcome!
    gif of LM3046 based sawtooth generator (31258 bytes) 
    gif of sawtooth conversion (29132 bytes) 

Rick Jansen
rick at
S&H's a module and s&h's looking good

More information about the Synth-diy mailing list