[sdiy] I have found that SSM2210 transistor arrays give better performance in my latest oscillator design.

[Grant Richter]

It is a second order error function making the VCO start to go flat  
in the octaves above Middle C.

You can compensate for it. There is a constant current supply to bias  
the temperature compensation of the second transistor Vbe drop. The  
output of that op-amp can be routed through a diode and trimpot to  
the transistor base to pull it slightly higher at the top end. Dave  
Rossum published it in one of the Electronote bi-monthly news letters.

That works for both triangle and sawtooth. There is another error  
term in the sawtooth core, cause by the reset time. That can be  
compensated for by adding some resistance in series with the timing  
cap. That is called DeFranco compensation.

While technically it doesn't cancel out the Rbe error, just  
compensates for it,  the compensation extends accurate tracking to  
about 4 more octaves above middle C.

Most actual music happens in the two octaves above and below Middle  
C, so high end tracking is more of a luxury than an absolute must.

On Jun 24, 2006, at 6:53 PM, Paul Perry wrote:

> I'm not an EE, but I'm suprised there isn't some clever
> way to (exactly or approximately) cancel out the Rbe.
> Anyone? I'm sure EDN would publish it...
>
> paul perry melbourne australia