WG: Re: AW: Compensated Exp generators a

[Haible_Juergen#Tel2743]

> > A common way to compensate that effect is a simple resistor
> > (or trimpot) in series to the capacitor. If You have a discharge
> > time T and a capacitance C, then R=T/C should compensate to 100%.
> > Make sure that the comparator input is connected to the current-
> > source/trimpot node and not directly to the capacitor.
>
> Hmm... this works of course, but I think it does cost you linearity
> of the sawtooth shape itself, as you're now charging the capacitor
> via a resistor.

No. The capacitor is charged by the constant current from the
expo converter, just as before. The resistor is just in series of the
cap and doesn't change the current value at all. The only thing
that the resistor does, is adding the same offset voltage that You
are generating by other means (only at the other input of the
comparator).
Thinking of it a second time, I find that it also has the same side
effect like You described (slight change in amplitude with frequency).
The only remaining advantage is that this single-resistor addition
is probably the most simple method to get the desired result.
Ah, em, if You meant linearity of the *discharging*, I should have
pointed out that the discharge transistor is connected directly
to the cap. (only the *comparator* input sees both, cap plus resistor!)

JH.