A few thoughts on heat tests and heaters

[Martin Czech]

> I did the heat gun, and freeze spray tests and after I had tried it, I had
> a few concerns about the results.
> There is a temperature stable point in every VCOs expo convertor, when
> dVbe=0 then there is nothing to drift with temperature. This depends of
> course on the chosen frequency at which the test is performed. To make any
> use of the measurement, one would have at least have to measure two
> frequencies, one below and one above the temperature stable point. 
> 

I think that the most  obvious thermal dependency occurs at the highest
possible frequency? So it would be good to tune the osc to, say 17kHz
and then do the heating.


> Then obviously there is the problem of temperature gradients, when using
> the iron to heat up the chip, nothing makes shure that the heat is
> distributed evenly on the package and the chip. 

Right. The system should have time to settle down at the new temperature.
You don't need the ability to cycle temperstures in practice, just settling on 
other conditions. 

> The second problem I encountered was when using freeze spray, because there
> will be water condensing on the circuit which causes leakage currents.

Right. Would be a very bad thing, since little currents are involved.
Better heat only.

> But one thing that isn't often mentioned is that some transistor parameters
> degrade with increasing temperature. For example the leakage current!
> The leakage is interesting only in VCOs that have the cap to the supply
> because 
> then the sinking transistor will have a rather large voltage across it. (In
> VCOs where the current flows from a virtual ground this is not an issue.)
> 
> Another thing makes this even worse: the heater can't make the chip cooler
> than it is. So if you choose your heater temperature too low, the
> regulation won't work at temperatures above that, and if you make it too
> high the performance of the transistors may degrade.
> 
 
The heater tranny should not leave the recommended operation area for shure!
There should allways be some measures to enshure that this can't happen,
even if the regulator freaks out. If you are in recommended area, the array
does not degrade (irreversibly), but parameters change due to temperature,
like you say, leakage, or hfe etc.

If you want to let the thing oscillate at very low currents you need some kind
of leakage compensation anyway.

m.c.