dV/dT for a shottkey diode?

Joerg schrieb:

But be aware, tempco depends significantly on forward current, have a look to Onsemi's datasheet for the 1N5819, Fig.7.

Regards, Dieter

Bottomline I would not use a diode to compensate.

Fred, why don't you run a closed loop system and measure on the battery side? I mean, a TL431 reference or a better one doesn't cost an arm and a leg.

a diode to compensate.

Amen, Jeorg. I've been through that, trying to get diode compensated chargers to work well on sealed lead-acid batteries used as backups on a fire alarm system, with several units on porches where the temperature ranges from about -20C in the winter to +35C in the summer. We finally bit the bullet and reworked the circuit with TI UC3904 lead-acid charge controllers, and the problems essentially vanished. They aren't particularly cheap, but they are a lot cheaper than replacement batteries.

Cheers, Tom

diode to compensate.

Make that +45C for use in Northern California in August. Or +65C or whetever when the sun hits it :-)

And in Minnesota it can get quite a bit colder than -20C. They call that temperature "winter on the mild side".

The formula is actually the voltage varies by -0.3 percent; your signal will get smaller too as your forward drop goes down, and a 1N5819 isn't terribly ideal in its adherence to the diode-equation curve anyhow. Most people use a transistor with collector/base shorted together for thermometry.

Accurate determination of the forward voltage is not part of most semiconductor test procedures; you might find a batch to be uniform, but wide variation in any given part number (2N3904 or 1N5819) is to be expected. You will want to calibrate, or use pre-trimmed integrated circuit temperature sensors, if accuracy is required.