[sdiy] Any noise avantage to parallel 2164 VCAs?

Mattias Rickardsson mr at analogue.org
Mon Jun 18 10:10:58 CEST 2018

> electronicpresskit wrote:
> > I think it’s more accurate to think of the 2164 gain cells as a voltage
> controlled resistor rather than an opamp stage.

Schwarz Raphael wrote:
> > Thank you ! Now I understand why there is no resistor between the 2164
> cell and the following op-amp. The 2164 cell is acting as the input
> resistor, the gain itself is in fact the work of the op-amp.

To paraphrase Neil: No. :-)

The input resistor still exists as the Rin before the VCA. See the VCA as a
magic multiplication factor of the current that runs through the resistor
and to the op-amp. (It's not 100 % correct perhaps, but it's a simple and
working model of understanding the construction.)

Seeing the VCA as the input resistor gives lots of faulty implications. The
outer world doesn't experience any other resistance than Rin, regardless of
the VCA gain setting. And Rin sits at a virtual ground at the VCA input,
just as it would on the op-amp input if the VCA wouldn't be in the circuit.

See the VCA as something that boosts what the input resistor *does*, not
what the input resistor *is*.

On Sun, 17 Jun 2018 at 23:58, Tom Wiltshire <tom at electricdruid.net> wrote:

> > On 17 Jun 2018, at 22:23, Mattias Rickardsson <mr at analogue.org> wrote:
> >
> > 30k for the Rin maxes out the input current capability for SSM2164/V2164
> within +/-15 V supplies; with lower supplies you can handle lower input
> resistances, and the SSI2164 even lower to reach lower noise.
> I once experimented with a 2164-based SVF to see if I could improve the
> high frequency performance. One of the things I tried was reducing the
> input resistors to 15K to run the filter hotter, and it seemed to work out.
> So I wouldn’t describe 30k as “maxing out the input capability”. At least,
> it didn’t seem that way when I tried it.

The VCA starts clipping at roughly 465 uA input current, as hinted in the
SSM2164 datasheet albeit in clumsy decibel terms. (22 dB headroom above a
nominal 37 uA signal peak, IIRC.) This is what you get under +/-15 V
supplies with the 30k input resistors in the datasheet.

If you just have half that signal amplitude, you can use half the input
resistances to get the same current. (The exact values of the input
stability RC network might become important, though.)

On Mon, 18 Jun 2018 at 00:03, Neil Johnson <neil.johnson71 at gmail.com> wrote:

> You can certainly run with lower input resistors, but you need to
> adjust the compensation circuit to match, otherwise you'll find that
> at high input amplitudes the gain cell can become unstable and
> overshoot.

Yes, it seems a bit tricky. Luckily one of the two current(!) manufacturers
of 2164 is interested in doing research on this matter. I now use the new
improved SSI2164 with 15k input resistors (doubling the current, lowering
the noise) and a compensation RC circuit of 226R/1200p as per
recommendations from SSI. Seems to work very well!

I'm awaiting more advice on the RC values for handling the full current
capability of the SSI2164, which is 4 times that of the old 2164s.

On Mon, 18 Jun 2018 at 00:01, Neil Johnson <neil.johnson71 at gmail.com> wrote:

> Mattias Rickardsson wrote:
> > No no, separate input resistors and RC snubber networks for the two VCAs
> It is _not_ a snubber.

Ah, OK. I know nothing about snubbers, but Dan at SSI calls it a snubber,
so that's why i misuse that term. O:-)
What should we call it? Input stability network? Input compensation network?

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/20180618/c4e151b4/attachment.htm>

More information about the Synth-diy mailing list