ADN8831 cooling mode

The amount of heat (in watts) transferred through a Peltier junction per unit current (in amps) is a great deal lower in cooling mode than in heating mode. In my application, a 40 degree Kelvin temperature change in the temperature to be controlled meant a seven-fold decrease in heat transferred per unit current.

The Analog Devices data sheet for the AD8830 didn't give the formula that allows you to work this out, and the Linear Technology data sheet for their similar part is no better.

You might not be putting enough current through your junction to get the cooling you need, and/or you might not have a good enough heat sink on the exhaust side of your junction to deal with the much increased output heat flux you get when cooling.

I wrote up our project and published it

"A microcontroller-based driver to stabilize the temperature of an optical stage to within 1mK in the range 4-38C, using a Peltier heat pump and a thermistor sensor" by A W Sloman, Paul Buggs, James Molloy and Douglas Stewart, published in Meas.Sci.Technol. volume 7, pages

E-mail me (at bill dot sloman at ieee dot org) if you want a reprint. My paper does include a formula for calculating heat transfer as a function of drive current - if you know (or can estimate) the thermal resistances around your junction.

This comes up here every few years.

-- Bill Sloman, Nijmegen

Since you wrote that it doesn't cool at all, could one of the output bridge FETs be busted? Or maybe one of them lost a pin connection? Watch the voltage levels on the actual TEC pins. If one does full swing but the other floats about in cooling than that's the cause.

I have only used it at 5V and it works like a champ. You might want to feed it a little higher VDD and see what happens. But if a FET blew that's not going to change anything so I'd check those first.

Anyhow, Just in case I'll email you the name and email address of the engineer at AD who knows about this chip.

One word of caution: Some of the formulas in the app note for the ADN8831 are, ahem, not quite right. They told me that it'll be re-written (but not when).

Some additional information.

We noticed that when the linear amplifier tries to turn on to cool the TEC, the soft start / standby pin goes low. When this pin goes low, it turns off the driver section off including the linear amplifier. After the soft start delay, the linear amplifier again tries to turn on to cool the TEC and again SS/SB goes low.

Every time the linear amplifier turn on, it gets turn off by SS/SB going low.

Any ideas as to why SS/SB would be going low on the linear amplifier turns on? We have a .1uf cap on ss/sb, the same as the evaluation board.

Regards, Alex

That's an input so this is a bit puzzling. Page 14 (under stanby mode) says a logic high can be applied. You could try that and see if it cools then.

Failing that I guess it's time to swap out the chip. I know this isn't going to be fun at all. Hopefully you guys have a large via below so you can heat up the large center pad with a big iron.