---
title: "Peltier coolers"
question: "Should you drive a Peltier cooler with PWM or steady DC?"
url: "https://nearfieldnotes.com/videos/peltier-pwm/"
site: "Near Field Notes"
video: "https://nearfieldnotes.com/media/peltier-pwm/full.mp4"
duration: "3:05"
date: "2026-11-14"
note: "https://nearfieldnotes.com/notes/peltier-pwm/"
---

# Peltier coolers (video)

Should you drive a Peltier cooler with PWM or steady DC?

3:05, with sound. [Read Note 39](https://nearfieldnotes.com/notes/peltier-pwm/) for the full answer, with figures, conditions and sources.

## Transcript

These two Peltier coolers hold the same box at 8 °C.

The one on PWM burns 30 W. The one on steady DC burns 7.

Here's why, the model that makes a Peltier easy to reason about, and the one part that fixes the PWM.

A Peltier module is 254 small legs of bismuth telluride, between two ceramic plates.

They're wired in series, so the same current runs through every leg.

In each leg, the charge carriers take heat along with them.

Half the legs are n-type and half are p-type, and the current runs up one and down the next. So in every leg the carriers move the same way, from the cold plate to the hot one.

In this module, each amp carries about 15 W of heat.

But you pay two taxes on it.

The legs have resistance, about 2 Ω in all. Half of that heat lands on the cold side.

And heat leaks back through the legs, from the hot plate to the cold one.

At 6 A, with 40 °C across it: 87 W carried. 37 lost to resistance. 22 leaked back. 28 left.

Those 28 W come out the hot side, plus the 88 W you put in. 116 W for the heatsink.

Carrying grows with the current. Resistive heat grows with its square.

So past about 6 A, more current makes the cold side warmer.

If you know DC motors, you already know this part.

Its voltage is current times resistance, plus a back-EMF. Here the back-EMF comes from the temperature difference, about 55 mV per degree.

So temperature difference is speed, and heat flow is torque.

Stalled, with no difference, it pumps the most heat: 61 W. Free-running, with no load, it reaches its biggest difference: 70 °C.

Everything else sits on the line between them.

And like a motor, it's most efficient lightly loaded. At 20 °C across, it moves 1.6 W per watt at 1.6 A. At full current, 0.5.

Now the two coolers. I simulated this module on a fan heatsink, cooling a foam box, from 12 V.

At 50% duty, it gets full current half the time: 5 A, then nothing.

Heat carried follows the average, 2.5 A. Resistive heat follows the square, so it heats like 3.5 A.

And while the switch is off, heat leaks back.

The box settles at 8 °C, on 30 W.

Steady DC holds the same 8 °C with 1.6 A at 4.7 V. 7 W.

A faster PWM doesn't help. I ran it from 5 Hz to 20 kHz, and the loss didn't move.

A thermostat that clicks on and off is the same thing, slower. Holding 15 °C that way took 17 W. Steady DC took 2.

A motor gets away with PWM because its inductance smooths the current. A Peltier has almost none.

So add it. An inductor and a capacitor after the switch, and it's a buck converter.

Keep the ripple under 10%, and it costs about 1%.

Make it a full bridge, and the same module heats and cools.

That's what holds the laser in a fiber-optic transceiver on its wavelength.

So, a Peltier: no moving parts, small, precise, and it runs both ways.

But its heatsink carries several times the heat it pumps, and it wants steady current.

Same box, 8 °C. PWM, 30 W. Steady DC, 7.

---

Source: https://nearfieldnotes.com/videos/peltier-pwm/
