M6 noise gen further investigated.

[jorgen.bergfors at idg.se]

Hi all.
Back from Germany and the Cebit fair, I made some more experiments with the Oberheim Matrix 6 noise generator yesterday.
To make sure Marjan's schematics are correct, I opened up my own Matrix 6R and reverse-engineered the noise circuit.
As Marjan suspected, both the 4k7 resistors in his original schematics are really 4M7. Apart from this, his revised schematics are correct (transistor orientation wrong originally).
The op-amp used in the Matrix 6R is a C301AC . I suppose this is the same as an LM301. At least it is a single op-amp.
It turns out that changing op-amp type makes a large difference.
>From measuring with oscillosciope and swapping components I get the impression that the circuit works like this:
The op-amp works like an oscillator, that is frequency-modulated by the noise from the transistor.
It the transistor doesn't make enough noise, there is a tone that can be clearly heard. The output oft the first op-amp swings between the rails. This is then half wave rectified and low pass filtered.
I don't quite understand why it must be rectified. Maybe this increases the high frequency content?
Can any of you circuit theory experts confirm this function description?
Anyway, I got the best results with an NE5532 op-amp and an old BCX59Y transistor. With these it does sound like white noise.
A TL072 doesn't sound quite as good. Here the oscillator tone (68Hz) is slightly heard even with the BCX59Y. With other transistors it sounds more like engine noise. Useful maybe, but not what I was after.
I also thried 2N3704 (not good) and BC170B (better), but BCX59Y was clearly the best of the ones that I got. The M6R uses a MPS5172. It is the only transistor of this type in the machine, so I suppose it was chosen because it made much noise.
Adding a 22p capacitor between the inverting input and output of the op-amp usually reduces the pitched component and improves the sound.
A 4558 sounded almost the same as a TL072, but a little thinner.
A 1458 didn't produce much noise at all. Only a slightly instable 60 Hz tone.

The amplitude of the noise at the output was 2,5 V p-p. To get that up to 10 V p-p, I changed the 3k9 resistor to 2k2, the 1k after that to 10k and the capacitor to 1n. I then put in a 10n capacitor in series and a 1M resistor to ground to get rid of the DC component. Otherwise it will swing between 0 and +10V because of the rectification.
A ran my circuit on +-15 volts.

/Jorgen