[sdiy] expo pair "heat shield?"

[Spiros Makris]

I have built two different VCO prototypes some years ago, using different
methods of thermal stabilisation.
The first was two 3906 transistors that I strapped together, putting a
resistor between them as a heater. I added copious amounts of thermal
paste, since they are supposed to be at the same temperature. The second
used a monolithic transistor array from THAT in a thermal oven type of
configuration; two of the transistors formed the pair while the other two
worked as a heater and temperature sensor, creating a thermostat. The first
one kinda worked but was very sensitive; the second version is fairly
stable as long as you don't place it on strong air currents.

According to the analysis for the expo pair, the two transistors need to be
at the same temperature and the input voltage is scaled for that
temperature. Change the temperature and you have to change the scaling; the
two most common methods are to either keep the temperature constant or
observe its variations and actively scale the voltage in order to keep the
tuning stable.

My first prototype was an open-loop version of the constant temperature
method. The idea is that by dissipating enough heat on the pair, you can
heat it some degrees above ambient, and that will eventually reach an
equilibrium temperature. This final temperature is strongly dependent on
ambient temperature as well as variations. A strong enough heater will
mitigate some of that, but this eventually means that your tuning is almost
seasonal, as I painfully found out.
Another interesting effect was self-heating. I was using LM13700 for the
oscillation core, so I had 2mA as my upper limit. In order to get the whole
range of the IC I used 2mA as my maximum output current, which apparently
is enough to heat up the transistors. The result was that my tuning would
change if I played long enough on the higher registers. Now that was crazy!

My version did not use any kind of thermal shielding, I think that if you
encapsulate the whole thing in a heat-shielding compound, you will make
this process more predictable and improve stability. I'm not well versed in
heat transfer but I think that eventually after enough time has passed, the
encapsulated pair will stabilize on a different temperature that is still
dependent on ambient, but this process will happen so slowly that your jam
session would be finished before that.

The second version was a closed-loop version of this idea. It required
20-30 minutes to reach the final temperature (the heater transistor could
not go above a few mA) but could do so reliably in all seasons and would
even resist light air currents. The target temperature you pick should be
above the highest ambient temperature you might ever encounter since your
system can only add heat, ie increase its temperature. I am located in
Greece, so I assumed 40C as the highest ambient temperature, and set my
pair to 55C.

The temco solution is also closed-loop and as such, can correct for various
ambient changes. Its weakness is that there is a lot of hysteresis between
the actual thing whose temperature you care about (expo pair) and your
sensor, meaning that it will adjust for average temperature, but will not
fair well with transients. Of course, most synths are inside a case so
that's usually not a problem. Positioning the circuit away from
heat-generating parts, such as power regulators, and placing the tempco
close (stepped on?) the pair would make the performance more predictable.

The biggest weakness of my monolithic solution is that these kinds of
arrays are rare in the market. I needed 4 PNP transistors and THAT was the
only option I had, however, there's lots of monolithic transistor pairs
that could be used, if the thermal loop becomes external. It would be
interesting to encapsulate the pair, heater and sensor and see what kind of
performance you can get from it.

Good luck!






On Fri, 14 Oct 2022 at 05:30, drheqx via Synth-diy <synth-diy at synth-diy.org>
wrote:

> We use a thermal potting compound to encapsulate military modules. The
> silicone is not type one
> ( acetic acid ) and has minimal odor. The key ingredient is something like
> sand or glass spheres
> that conduct the heat away from the devices. The thing is in this case do
> you really want the heat to go away
> or do you want it to become a constant?  Old Tektronix vector scopes that
> I used to repair by the dozens used a
> brass looking "oven" to encapsulate parts in several areas. These
> things had 25", I think, CRT displays and they were mostly
> analog. They definitely had items that needed to stay thermally
> consistent, but not necessarily cool.
>
> Just a thought about the oven idea.
>
> Mike
>
> On Tue, Oct 11, 2022 at 1:28 AM Neil Harper via Synth-diy <
> synth-diy at synth-diy.org> wrote:
>
>> hey all,
>>
>> just wondering if anybody thinks it's worth encasing an exponential pair
>> for stability. while this is in my modular case, there must be some heat
>> currents moving around from the other stuff that share the same space.
>>
>> Maybe a dab of hot-glue-gun (hot snot), or a silicone or something?
>>
>> I'd hate for it to bite me in the arse in the future, maybe corroding
>> the copper traces or the tin legs or becoming impossible to remove. What
>> do the pros use?
>>
>>
>> --
>> /// Neil Harper
>> /// Every Wave is New Until it Breaks
>>
>> _______________________________________________
>> Synth-diy mailing list
>> Synth-diy at synth-diy.org
>> http://synth-diy.org/mailman/listinfo/synth-diy
>> Selling or trading? Use marketplace at synth-diy.org
>>
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy
> Selling or trading? Use marketplace at synth-diy.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/20221102/3135d96b/attachment.htm>