[sdiy] 2164 expo VCO tracks!

[cheater cheater]

David,

On Sun, Oct 18, 2009 at 5:07 PM, David G. Dixon
<dixon at interchange.ubc.ca> wrote:
> Well, Damian, at the risk of offending our favorite Finn, who has apparently
> HEARD enough about this module, I'll answer your questions:
>
>> I think it's important that you get the right JFETs!
>
> Absolutely.  Based on my experience with this VCO, it's THE critical thing.
>
>> Also, maybe think about trying the approach with two or more HF trimmers?
>
> Yes, I've thought about this.  It seems to me that, given a "decent" JFET
> and a "reasonably fast" comparator, the key to excellent HF tracking is the
> shape of the I-V response from the HF trim circuit.  Unfortunately, a diode
> (or diode-connected transistor) simply does not give the correct response.
> It undercorrects at the lower end, overcorrects in the middle, and
> undercorrects again at the high end.  There is a simple circuit one can
> build with a single opamp and a few diodes and resistors which will give a
> steadily-increasing piecewise-linear gain response, with breakpoints and
> slopes determined by the input resistors.  If the breakpoints were designed
> to coincide with the required gain curve for "perfect" HF trim, then this
> circuit would give excellent HF tracking, without trimmers.
>

I think it's important to find some of that stuff. I heard some good
opinions about radio application JFETs when I was looking into the
mixer idea, see if they can work for your applications.. ON Semi
MPSW42G and KSC2330 were the ones I was recommended. They're
supposedly very good for VCA purposes, but they might work with the
VCO too. They have a bandwidth of 50 MHz and do 300V on Vpp. Seems
like they should switch fast.



>> HF frequency is very important for FM and divide-down and the like.
>
> The thing to understand here is that all of the HF tracking issues in my
> circuit are around the integrator components, and have nothing to do with
> the 2164 expo converter, which is essentially perfect.  Better integrator
> components will give better tracking.  I simply did not have better
> components on hand, but I always use IC sockets, so I can simply unplug the
> opamps and replace them with better ones.  (The 5485 JFET will have to be
> desoldered when I get my 4391's).

Do you think that the sockets themselves might have something to do
with the HF crappiness? What's the capacitance and resistance of a
socket? I think it's worthwhile to check it out.

>
>> Regarding the switching glitch:
>> it's a feature, not a problem! If you can make its 'depth' or 'amount'
>> controllable, then that's something very useful.
>
> It's very tiny -- barely perceptible on my scope, and totally inaudible.
> I'd use a waveshaper circuit to mess up the waveforms.

Ah, but a waveshaper exists at a different point in the circuit
already... So it would work differently :) I wonder if the glitch can
be exploited and made bigger.. but that's a question for a different
thread altogether.

>> Another important thing:
>> how does it behave around other modules? Does it put switching noise
>> on the voltage bus?
>
> Absolutely not!  I've put 10uF electrolytics and 100nF ceramics on the
> rails, just after they come onto the board (as I always do).  This
> effectively isolates the module from the rest of the synth.  There are no
> other decoupling caps on my board.  I could add a couple, but I personally
> don't think it is necessary.
>

Out of curiosity - if you wanted to put in more decoupling, where
would you do that?

Do you think it's worthwhile to decouple different parts of the
oscillator from one another?

Cheers
D.