[sdiy] The TL072 , part 2

[Ian Fritz]

>While I agree 100% that TL072 is a bad choice for precision CV summer, the 
>input offset voltage spec (which is what was being talked about) doesn't 
>make any difference in this case as it is simply cancelled by the tuning 
>offset.

No, that's wrong.

A DC offset at the base of the exponential converter leads directly to a 
temperature drift (tuning drift, not scale factor drift).  This is not 
often discussed, for some reason, but it is true.  The relation is that 1 
mV of offset gives 129 ppm/K of drift.  This is carefully explained -- with 
a detailed derivation -- on my website, as linked to in my previous 
reply.  My dial-a-tempco expo converter allows for this drift source to be 
compensated.  I've verified with careful measurements that the above is 
correct.  IOW, I measure the tuning drift, calculate how many turns of the 
trimmer is required for compensation, made the trimmer adjustment and then 
verify that the tuning drift is eliminated.

It's interesting that the early synth designers never seem to have 
understood this.  Thus you sometimes see ridiculous compensations schemes 
that are totally unnecessary, as a single fixed resistor could have been 
used.  (I have a drawing labeled Emu 1201 that shows an example of such a 
scheme.)

We are being told today that you don't need calculus for sdiy.  Well, drift 
is a temperature derivative, so it is virtually impossible to understand 
without calculus.  In fact, without a calculus *trick*, it is still very 
difficult to understand, as the equations become very messy and quickly 
expand to many pages.

But people will spout off anyway, won't they?

Ian