Feedback on circuit design

[Paul Maddox]

Marc,

> 1) Do I need any buffering stage between my resulting signal
> (cut-off/pitch) and the monotron's vc points ?

nope, if you hook into where it says "cutoff" and goes into pin 6 of
IC4B (LM324) you'll be fine. You'll need an input resistor (10K like
the output of the LFO has) between your cutoff CV and that point.

> 2) There's this strange 'bias' solder point in the monotron's
> schematics that people seem to use and is apparently some kind of
> virtual ground. I'm not using it but I wonder if I should. Right now
> the cutoff signal influences the cut-off in both directions (with
> respect to the monotron's knob setting). Maybe if I used it it would
> be only in the positive part ?

the "bias" point creates a virtual ground.
Opamps (and there's plenty in this schematic) need both a positive and
negative rail.
If you're running of a battery of DC wall wart then you only have a
positive rail.
the quick (cheap) way to do create a dual rail from a single rail is
to make a virtual ground that is half way between the supply rail and
ground. This gives you a "virtual" ground and a "virtual" negative
rail.

so, if you had say 9V of power coming in, you power rails would be +9v and 0V.
using the virtual ground trick, you have a power rails of +4.5v
(actually 9v), Gnd and -4.5v (actually 0V)

remember, supply rails are all relative to something.

If your current voltage brings the CV negative, then you'll need to
use the virtual ground (or bias point), however you need to be VERY
careful.. you must *NOT* connect any ground on your circuit to the
'real' 0v of the monotron, you must ensure that your ground connects
to the virtual ground (or bias) only.

> http://beatnic.jp/manuals/monotron-midi/img/mono++midi-schem.PNG

IC2B is being used to create the virtual ground.
The problem with this technique, is that you need to be wary of the
current flowing to ground, might be worth looking up the LM324
datasheet to see what the maximum current it can output is.

> 3) I'm planning on using the first channel of the mcp4822 to generate
> a DCO that will be fed in the monotron's line input to beef it up with
> a bunch of DCO / noise / whatever goes in my mind. Do I need to buffer
> it after the mcp ? (I guess it's pretty much the same question as 1)

No, again, Korg have provided you one,
IC4D, mixes the "line in" and the "oscillator output" together.
You'll need a 1uF capacitor (like C14 and C17) and a resistor (I'd
start with 100K like R58).
you need to wire the output of your MCP4822 into the capacitor, then
the resistor and then to pin 13 of IC4D.

> 4) A side question. How do you choose resistor values for voltage
> dividers ? I've put a 47k because I thought it would minimise the
> current and therefore consumption but it's a blind choice and I don't
> like that too much :D

So, R2 and R1 in your schematic are diving the voltage by half.
R3, R4 and C1 provide a low pass filter (I assume the teensy pin 3
output is PWM?).
The resistor values are a balancing act, or trade off..

put to high a value in there and you may not have enough current to
drive the destination.
put to low a value in the and you'll load the output of the driver and
the output voltage will sag.

47K/47K is a good start, the teensy MCU (what ever it is) should be
able to push a few mA out of its pins, but more power used at the pins
means a shorter battery life. I wouldn't go below 10K, but it should
be fine with 47K.

Just be aware though, that the resistance you use at that stage (R2)
will affect the slope of your low pass filter.
Ideally I would say keep R3 ten times the value of R2 and R1, but
that's not always do-able, you've got pretty close there with 22K,
giving you 5 times, so it should be fine.

> From there on, I'll have a nice platform to do a pretty interesting
> Digilog synth.. even if it will miss a VCA... but I don't think I mind
> too much. And I could add it later on too.

Only thing I'd add is, don't expect miracles from your DCO, the
MCP4822 is a fine DAC, but to use both channels takes a lot of CPU
cycles, you need to write 16 bits to change a value on a single
channel, if you're planning on "bit banging" the pins on the Teensy3
this will eat a lot of CPU time. if the Teensy3 has a hardware SPI,
it's a lot less of a load, but still slower.
i.e. don't expect a 96Khz sample rate on your DAC ;)

Paul