303 clone filter and VCO not reproducable

[Martin Czech]

> HI
> as you will know you must use the matched transistors in the difference amplifier
> and the current source. These double-transistors are still available. And I used
> not matched transistors in the diode ladder. I got resonance up to oscillatio by
> putting a variable resistor in series with the two 2k2 resistors of the current
> source : With this you get a higher voltage going into the output transistor.
> I also made the filter without the matched transistors for the current source and
> the current source and it also worked reasonably.
> 
Thanx.
Ok, lets see if I'm able to get these japanese double tr. also.
And I also think that the 2x4 ladder tr. are not matched, this would lead
to difficulties in mass production. This was a cheap instrument!

> I still have a few problems and maybe you know the answers : The vco doesn't always
> start at once and then blows out q27 (that's th transistor in the transistor made
> thyrisitor) Do you know how to cure this.
> And do you know a website where there's resources about building diode
> ladderfilters
> 
Do you use the original type? For SCR action these transistors or most
probably operated out of the SOA, ie. the absolute maximum ratings may
be exceeded.  I guess base current as well as collector current is too
high during the discharge.  These are small signal types which now carry 1
A or so during the discharge.  It is possible, that the schemo was revised
later, and that Roland changed the circuit because they had problems, too.
I don't have the schemo at hand right now, but are the passive components
ok on your board ? I mean, no typos, no wrong values, everything checked ?

You could try to find a replacement for q27 that has higher pulse current
specs. But then it is no *real* clone any more...
and beeing on this *wrong* track now...

...now, I have my flameproof suit and helmet on, so be it...

I think that the osc. as such makes only 10% of the sound. Filter
distortion and/or vca distortion makes 90%, most likely the filter
allone. And I think that the SCR method is no good idea for a vco at all
(not stable). If nothing else helps, you could try to use the normal
saw vco circuit, it is easy to make a spaced sawtooth out of it with
the original shape and amplitude on the scope.  If you have a real 303,
you can try to check this out, if it makes any difference. I think the
difference of a "false" vco to a real clone should be thus so small that
it is in the limits of normal 303 production bandwidth. Two 303 won't
sound exactly equal. Again, this was once a cheap box.

Just my 0.01 DM.

m.c.