Heated expo-converters

[Grant Richter]

>    I've been mucking around with a MAT-04 and the AN-299 article from 
> National, trying to make a heated exponential converter out of it and 
> haven't had much luck.  Well, that is, I have and I haven't.
>    My first problem is that it doesn't track worth a hill of beans.  Well

> that is, it doesn't conform to 1V/Oct.  I guess I'll have to go back 
> through the circuitry to find out where I goofed.
>    My main problem is that I don't think the heating regulation circuitry

> is doing anything.  In fact, I can't tell that the chip is dissipating 
> any heat at all.  But boy does that current limiting resistor though!  In

> fact, I am too scared to put a 33 Ohm resistor in there because a couple 
> of 470's in parallel sure do get hot!
>    I'm tempted to just slap two or three 470s over top of the chip and 
> create a nice little thermal shield for it!
>    Okay well let's see, the heater is in two parts, the heater element 
> and the regulator.  The regulator should turn off the heater when it gets

> hot enough and turn it back on when it cools enough.  Mine ain't doing 
> it, so maybe there's another problem.  But then I "still" don't think 
> that the chip is warming up.
>    Anyone got some advice?
> 

Check the polarity of the op-amp connections. The op-amp positive goes to
the sense transistor connected as a diode. The negative goes to a 690 MV.
voltage reference. That indicates VBE goes DOWN with temperature. The
op-amp should produce a rounded square wave at around 200 Hz. depending on
the value of the capacitor. The chip won't get very hot. The 690 MV. value
seems to be just above room temperature. The Serge CA3046's don't get very
warm either (same circuit). If you put your finger on the chip for a moment
the op-amp will stop oscillating.

The Wiard VCO's use a CA3096 transistor array thermally stabilized by the
AN-299 circuit. The two PNP's left in the package are used in Bernie
Hutchins classic dual PNP converter. This is nice because it provides log
and linear current sources, unlike the Terry Mikulic NPN-PNP circuit.