[sdiy] Thought experiment...

[Magnus Danielson]

From: "D A F" <polaris30 at ncx.com>
Subject: [sdiy] Thought experiment...
Date: Mon, 21 Mar 2005 20:08:23 -0500
Message-ID: <003d01c52e7b$a5b86f30$df63e7cf at squirtgy7dv5qe>

> Hi all,

Dave,

> I was taking a look at the schematic for the music from outer space VCO
> board, when I got an idea concerning a possible way to elliminate tempcos
> from VCO's.  I'm sure someone has probably thought of this (like 40 years
> ago) but in any case, here goes...   If you had a quad of matched transistors
> on a single substrate (such as in the new THAT 300 series), or at least two
> thermally coupled pairs, couldn't you send one pair a positive-going CV and
> the other pair a negative-going CV and then send the two outputs to a
> differential amplifier, such that only the difference signal would be
> amplified.  This would elliminate the change in current with temperature from
> the final output, because it would be equal in both pairs, assuming perfectly
> matched and thermally-coupled devices, and the diff amp would 'ignore' it.
> Any thoughts?

Yes, you are incorrectly assuming an offset error, but that is not the problem.
The problem is that the error is multiplicative on the logarithmic side.
The NP-junction diode of a tempco-transistor is

          V  *q
           BE
          -----
I  = I (e  k*T  - 1)
 C    S

q is unity charge
k is Boltzmans constant
T is the temperature in Kelvin
Vbe is the Base-Emitter voltage
Is is a characteristic "constant"
Ic is the collector current

We toss our CV into the Vbe side, but the temperature error changes the scale.
A correct solution scales the CV with a measure of T before sent to the Vbe.
Such solutions have been discussed here before and the masters behind it can
step forward right about here...

Cheers,
Magnus