Heated expo-converters

[Tony Clark]

> Ummmmm..... Tony's right. I built two versions of this circuit. The first
> one used one quarter of a TL074 for the op-amp instead of a 741. I still
> used the 0.01 uF cap. That one oscillates (the 200 Hz rounded square wave I
> mentioned earlier). The oscillations also feed through to the VCO and I
> couldn't isolate it. The second one I built moved the op-amp to the outside
> and used the 741 as originally speced. That one comes to a nice DC level
> without oscillation. I tried upping the cap on the TL074 to 0.1uF and
> slowed the oscillation down but did not eliminate it. I'm thinking the main
> difference is the slew rate on the op-amps (0.5V/uSec vs. 13V/uSec).

   Boy I love getting myself in the thick of things!  Actually, I was 
using a TL082, half for the CV summing, half for the regulation!  I can't 
explain why yours was oscillating, but maybe you'll know better when I 
explain things out a little more:

   First of all, I totally threw out the diode and everything associated 
with Q6 in the AN-299 schemo.  So the output of the TL082 went to the 
2.2k resistor straight into the heater transistor.  I should probably put 
the diode in at least, but anyway...
   The second deviation I made was by not having an LM329 handy.  So I 
used an LM336 5.0V.  Now the problem with using this was that I couldn't 
quite square on the divider to make the damn thing regulate.  It either 
wanted to be full on or full off.  So out of desperation I threw in a pot 
there and bloody well trimmed it until it worked!

> To summarize what I think I found out:
> 
> 1. Don't substitute the 741 in the circuit.

   I don't really see how the op amp would matter.  In fact, I'd think 
that using an opamp with large offsets would make matters worse.  Ah 
well, using a trimmer should take care of that anyway!

> 2. Use a single chip package, not part of a multiple IC package.

   As long as it doesn't oscillate, no point in wasting PC real estate.

> 3. If you can, run independent power supply wiring directly to this circuit
> from the incoming supply rails.
> The ETI article specifies 670mv at 20 degC and dropping 2.2mv per degC
> above that. The Moog Source service manual has this to say (paraphrased):

   This is the strange thing, I believe my circuit was running around 
500mV!
 
> a) (After the chip has completely cooled off) Measure sense transistor VBE
> voltage with respect to ground (Vc).
> b) Estimate room temperature in degrees C.
> c) Subtract room temperature from 55 degC to determine temperature rise
> needed to reach 55 degC.
> d) Multiply result by 2mV (Transistor VBE decreases at 2mV/degC)
> e) Subtract this product from cold reading (Vc) to determine hot reading
> (Vh).
> f) Adjust TEMP trimmer to obtain VBE reading determined above. Substrate
> temp now adjusted to 55 degC.

   Ah, this might help some in determining the correct resistor values to 
plug into the circuit!

> I know the Source and Prodigy used substrate stabilized LM3046's for expo
> converters as did the Serge NTO's. The Moog circuit is built upside down,
> use 1/4 of a LM358 and a 47pF cap. Honestly, I swear I can hear a little
> "trilling" in those circuits though, I always thought it was the servo-loop
> cycling.

   This really makes me wonder about the validity of just plopping any 
old divider network into the negative input of the regulating op amp!  It 
really irks me to no end that I might have to use yet another trim pot 
($2.50 more to circuit cost) in order to get the circuit to function 
properly.  There must be a better method!

   Tony

-----------------------------------
I can't drive (my Moog) 55!
-----------------------------------
Tony Clark -- clark at andrews.edu 
http://aupe.phys.andrews.edu/~clark
-----------------------------------