[PM] Poly Modular example circuit

[jh]

Hi,

after much talk about standards and keyboard / Midi interfaces,
I'd like to introduce a little example circuit to show what I mean
with the (proposed) electrical standard of a current mode,
output level attenuator, and input summing node design.
I propose this for discussion, not to fix a standard already.
I'd like you to point out any error or general drawback of this
concept. Once we've started, it's too late. So don't be shy! (;->)

The example circuit is *UNTESTED*, and it's not sure if it will
actually become a module, but I chose it to point out a few
details about the proposed standard. The function of the example
module is that of a combined ADSR / VCA. I have only drawn one
of 4 channels, and the ADSR is just shown as a block. It's the
interface that's important here. (Try to imagine one of these outputs
plugged in one of these inputs.)

There are two signal inputs (summing), one (exponential) Modulation
input, one VCA output and one ADSR output. The VCA GAIN pot
scales the VCA (signal) output, the ADSR OUTPUT LEVEL pot
sets the strength of the ADSR (CV) output.

Though I am speaking of "CV", we actually have "CC" (control
currents) thruout the system. That means high impedance output,
but it will not cause problems with cable capacitance, because 
we go into a low impedance input ! (Early HiFi systems worked that
way.) The example circuit shows the advantages of this choice:

(1) As we need a VCA function for each output anyway (for setting
the output levels of all voices with one pot), and as we usually
use OTAs for that, it's not necessary to buffer the OTA output.
I have included a protection circuit for the OTA output, however.
(If we can get protection with less components, please tell me ! )

(2) The inputs will act as summing nodes even if they are no virtual
GND nodes. You can simply connect several input jacks in parallel.
The first idea was to drive the OTA inputs (or an opamp virtual GND
node ...) directly, but for protection of the inputs, there is an 1k
series resistor and a pair of diodes. Because of the current mode
outputs, this resistor will not hinder the true summing operation.
A similar circuit is shown at the modulation input (more details see
below.)

(3) Low (no ?) offsets. If an output is turned off (or scaled down),
these is no offset current either. So if you adjust for the input
offset voltage of the OTA (or opamp, or whatever), you're done.
You might even get rid of these AC/DC coupling switches in most
cases.

Time to talk about standard levels. I suggested 100uA output current
for "no" overdrive. And I'll add that -100uA will corespond to a positive
signal count. The reason is obvious: If you have an unipolar output
(0 ... 5V on a "normal" system), you can use a simple transistor pair
to work as output VCA, and you'd rather use npn pairs than pnp
pairs, if you have the choice. Well, we're setting standards, so let's
choose the right thing that will keep costs down in the whole system.
I can think of a single or a couple 3086 arrays to be used for a quad
output attenuator. (Remember I said it won't be much more expensive
than a monophonic modular ?)

Ok, lets asume an output like the VCA OUTPUT driving an input like the
SIGNAL INPUT shown in the drawing. With a peak audio current
of +/-100uA The OTA input will see 20mV across the 200R resistor
at its input. This is considered the right value for linear operation.
But you can drive the output harder, up to 1m peak, that means you can
heavily overdrive an input like this (input acts as a soft limiter at
approx. 100mV). In that casea bit more than 1V peak will appear across
the 1k input protection resistor, and a bit more than 5V across the 2k4
output protection resistors. This means that there is no transistor going
into saturation in the whole audio chain ! (Did I mention that I've avoided
opamps here ? Remember with a 5V Voltage-based modular system,
you have not much more than 6dB headroom before output clipping
with possible recovery effects start to appear. Not to speak of 10V
systems ...)

Let's look at the other inputs and outputs now.
The ADSR output is a current sink that can sink up to 250uA. This
can be scaled down with a pot (shared by all voices) by changing
the current distribution in a transistor pair. A trimpot for each voice
adjusts for transutor tolerances to get equal gain in all voices.
Now the emitter of the output transistor is around -500 ... -600mV,
so the output can handle drop voltages of -300mV without saturating.
At 100uA, this would be 3kOhm, and even with 250uA it should
work at a 1k impedance. Protection circuit ? The collector is tied
to the output, but all the other leads of the transistor are connected
to resistors, so it's hard to burn the output.
The MODULATION input is used to unbalance a pnp pair that will
set set the Iabc current of the 3080. The pnp pair actually multiplies
the ADSR output voltage with the exp() of the Modulation input
and with the exp() of the manual Gain (VCA GAIN pot).
Looking more closely, I guess the 20 Ohms are a typo. Make that
200 Ohms.

One little detail (optional) about the VCA input: The GND of the input jacks
is (slightly) isolated from the local GND. This prooved good for SNR
in my Frequency Shifter. GND concept needs a closer inspection
in general, though.

Hope this will inspire you (and not frighten you away). Comments
welcome, and keep in mind that this is just an untested example.

 

JH.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Example_Circuit_1d.gif
Type: application/octet-stream
Size: 11838 bytes
Desc: not available
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/19990512/4c31fa1a/attachment.obj>