[sdiy] SSM2164 Phaser - another way?

[Neil Johnson]

Tom,

> I don't understand the implications of what you're saying here, Neil. Can you explain a bit further?
> Are you suggesting that there's situations in which we should be using 47K on the input? on the I-to-V?

As Mattias pointed out, you select components according to how you
have designed your circuit.  The numbers in the datasheet have been
worked to show the 2164 in its best configuration (remember:
datasheets are marketing documents, their first job is to sell the
chip, so make it look great!).  In real applications you need to
account for your circuit's signal levels, polarities, supply rails,
and so on.

For example, if you're concerned about audio quality (e.g., in an
audio VCA) then you'd want to keep the 2164 around the 0dBu region for
most of the time.  And as I said, in a professional audio product such
as a mixing desk where internal levels are likely to be +4dBu, you
would need to either attenuate by -4dB before the 2164, OR use 47k
input resistors (and adjust the output stage as well).  Or in the case
of a consumer product which might be at -10dBV you would need to do
some other scaling as appropriate.  In the case of Eurorack modules
which seem to operate at a nominal +13dBu (5V peak, 3.5V rms) the
input resistor should be higher still, around 137k (130k NPV).

But then the question arises as to what the 2164 input likes to see in
terms of external impedance.  It could be that 130k is too high --
need to build it and test it extensively before saying otherwise.
Also in terms of audio signal quality, you really don't want 130k in
series to the audio path - lots of thermal noise - so it might be
better to attenuate the input, run it through the 2164, and then bump
it back up.

The same is true of other VCAs - for example the THAT 2180 has lowest
THD for a Vin of 0dBV (1V rms)  and 0dB gain and with Rin = 20k.  So
for +4dBu you might be tempted to experiment with 24k resistors (not
too big an increase in thermal noise), while for Eurorack you'd need
somewhere around 68k (NPV), which might be better done with a
low-value resistor attenuator, then a 20k input resistor, and then
amplify it up again on the output.  See figure 10 of the THAT2180
datasheet, and remember that data is with Rin  = 20k.

Neil
--
http://www.njohnson.co.uk