Is PWM for control of Peltier Element logical?

[MagikRevolver]

I'm making a circuit to control the temperature of a peltier element, and I'm trying to figure out the cheapest way that will actually work over the long term, and not burn out my components fast. At first I wanted to just use standard power regulation to control the power to the element and thus it's temperature, but since I need the same side to both heat and cool, I started considering essentially using an H-bridge and PWM as a control. This would kill two birds with one stone I feel.

Question is, would this sort of cycle potentially damage a peltier element, or otherwise cause early system failure? My peltier element runs at 10.47 V and 2.1 A. It peaks only about 100 mA above this on powerup. Everything I read recommends not to do this, efficiency and longevity are two cited reasons. But I am not a peltier expert, and am still curious if a low frequency cycle still wouldn't work.

 

[sandy.12345]

I'm making a circuit to control the temperature of a peltier element, and I'm trying to figure out the cheapest way that will actually work over the long term, and not burn out my components fast. At first I wanted to just use standard power regulation to control the power to the element and thus it's temperature, but since I need the same side to both heat and cool, I started considering essentially using an H-bridge and PWM as a control. This would kill two birds with one stone I feel.

Question is, would this sort of cycle potentially damage a peltier element, or otherwise cause early system failure? My peltier element runs at 10.47 V and 2.1 A. It peaks only about 100 mA above this on powerup. Everything I read recommends not to do this, efficiency and longevity are two cited reasons. But I am not a peltier expert, and am still curious if a low frequency cycle still wouldn't work.

am facing the same problem output current is in mA range and when connecting peltier at the output it drops to zero . I filtered the PWM input to a varying signal aby an RC filter can u please post your circuit diagram for cross checking .

 

[Windadct]

I can not see why it would not "work" - they Peltier documentation may not want a direct PWM signal (e.g. pure square wave signal applied) - but with some small filtering, this output would basically be a DC Power supply. You can probably find this all on one chip - less the filtering of course.

 

[Enthalpy]

You do need a filtering coil. A square current would be bad because it would waste power in the Peltier resistance instead of using the current in the Peltier thermal voltage. Peltier elements are so bad that with even more losses you might get no cold more.

What about a relay to invert the current, or even a mechanical switch? You'd have a buck regulator then, instead of an H bridge. Easier to drive as well.

 

[Mike_In_Plano]

You might look into the LT1070 sold at Ti. It's an old workhorse and used to be cheap. Then, it's straightforward run the regulated voltage to a slow, dumb H-bridge.

Even cheaper would be to toss in a pair of PFETs for the top of an H-bridge with a small cap and resistor at each gate to Vcc - 10nF, a 2.7K resistor on each, to pull it up (and off), and an NPN to gnd to pull it on.

Then use two logic leave NFETs at the Bottom.

Say you wish to heat, you turn on the right PFET and PWM the left NFET. For cooling, you turn everyone off, wait, and the turn on the Left PFET and PWM the right NFET.

Toss in an inductor suitable to keep your peak to peak ripple under 30% of your average current.

During it's off time, each PFET doubles as a freewheeling diode, but if you wish to improve efficiency, you can place a schottky in parallel with the PFETs and have a lower voltage drop.

I also like to place a series RC snubber from each leg of the bridge to ground to reduce ringing and noise. For something like this, I'd guess a 1nF cap and a 22 ohm resistor would be a start for the snubbers.

One last tidbit. I never trust micros to always have the gating right. For the price of one logic chip, you can save yourself all manner of worries. Just ensure that an NFET cannot turn on when his partner PFET is active.

- Mike