Expo conv. heater (tomg et al.)

[Martin Czech]

> 
> > 2. Barry Klein suggested putting the 3046 under a styrofoam cover to
> > isolate the chip from heat loss or gain - sort of a chip calorimeter.
> > Has anyone done this?
> 
> Not me, but Berry is a pretty bright guy. Seems like it would be a little
> odd though, kinda like a 3046 happy meal or something. 

Not odd. Makes sense.

You can't make the regulation loop as fast as you want, because of
temperature diff equation time lag, this would mean loop oscillation.
But you can damp external influences via isolation. This way the loop
is fast enough.

And:

You can reduce the total chip power consumption by isolation.
As I said earlier, heating is nice, but keep inside SOA!
Too much current will lead to migration und junction degradation,
and maybe also to second breakdown. That would be fatal.
The overall power consumption is lower (somebody complained because
he feared that the lights would go dim if the heater starts ;->).
 
> 
> > 3. How do you know where to set the temperature trim on the thermostat?
> > I mean, how do you know where to set the temperature and how do you
> > measure a silicon chip's temperature accurately anyway? Not having the
> > temp too high or low was discussed, so this is important.
> 

Test No 1

Build the temp sense circuit. I think it is a current source, the
part of the chip that does the sensing (one tranny as diode) and maybe
some sense amp.  Take a discrete diode and use it as dummy temp sensor.
Take a thermometer and prepare a glas of water with the same temperature
as the air. Measure Udiode in air.  Drop the diode in the water. Measure
again. If there is any significant difference it means bad temperature
of the water (check again or wait longer), or leakage current in the
water. In the later case use oil (kitchen) or clean water for automobile
battery. Now the leakage should be gone.

This way you have now prooven a large heat capacitance fluid without
leakage.


Test No 2

Now exchange the diode with the real chip.  Heat the medium to the wanted
temperature. Dip the IC, wait until settling down, and measure.

Now you know what voltage the sens amp will give for your wanted
temperature.  The trim voltage has to be of exactly this value for a
good PI regulator that will have no residual error.


Sounds stupid?

No!

We measure our Hall sensors in the lab exactly this way. Fluids have
better thermal contact and larger heat capacity. Heating with air is only
second choice.  Of course we use special oil, but I translated that to
home use. No poison.  

(I don't know if normal water will be ok, in the USA they are
bacteriophobic (sp?)  and put tons of Cl into the water, you can't drink
it any more. We don't have that much here, and nobody is harmed).


Of course I know I=Is*(exp{U/Ut}-1), the problem is that Is is not defined.
Varies from piece to piece and also with temp, but one does not know excactly
how with a specific chip. Better measure.

m.c.