W's simple high power PWM TEC controller

Return to home page

A good source for links for TEC controllers is this thread of photonlexicon.com.

For two reasons I needed to devolop a high power TEC controller: first, to create a temperature controlled hot/cold plate for measuring the temperature stability of various lasers and circuits, and second, I need a powerful TEC controller for cooling 40W and 60W 808nm laser diodes for my DPSS project. With high power I mean the order of 100W, or roughly 12V 9A or so, referring to the 12709 type of peltiers that are currently cheap on ebay.

Clearly a linear regulated supply is out of the question for this power, while a pulse-width-modulated controller is relatively easy to build. Certainly there are fully integrated controllers of lower power available (eg., based on MAX1968 and LTC1923 as used in my Coherent Compass 315M driver), but these very tiny SSOP SMD devices are a pain to handle and are also "too good" for this crude application. For the present purposes, small size and extreme stability is not important, and keeping the temperature constant to 0.1 degrees, say, is more than enough. So why not throw together a simple and cheap circuit based on standard components, in particular the familiar SG2525/3525, which has served very well so far in my multi-KW ion laser SMPS.

Simplest and sufficient for my purposes is a circuit that cools (or heats) only. Most ingredients for a buck converter incl. comparator opamp are already contained in the IC, and one needs to add only one mosfet, diode, inductor and a few caps and resistors:

The circuit is straightforward but requires careful layout when higher powers are involved (to prevent instabilities), inparticular a good ground plane is important. It runs at approx 100khz. The feedback loop is a simple integrator, and R7 and C1 may need to be adjusted in order to match the thermal properties of the load.  Temperature is set by R15 while maximal voltage over the Peltier element is set by R8. Almost any P-channel mosfet with the appropriate current rating should do; similar applies to the diode D1. The voltage rating of both should be at least three times the supply voltage, and the current rating two times of the desired output. The inductor must be chosen properly, ie, must handle the power without saturation. The inductivity should be as high as possible, say 100uH, in order that the controller does not run in discontinous mode at low currents. For 100W or so output, the toroidal core of the main inductor of a standard computer power supply probably does just fine (with 10 turns, say). For further remarks, see my the notes on my ion laser SMPS.

For low power applications with max current in the order of 1A (like for cooling DVD diode lasers), all components can be small SMD devices, as the power dissipation is minimal.

Notes added:

• I used a P-channel mosfet for a reason, namely a simple implementation of maximum voltage/current control. If one prefers to use a more common N-channel mosfet, one need to change the IC to SG3527 (or insert an inverting mosfet gate driver), and since the Peltier supply is not referenced any more to ground but to Vcc, one needs to cook up a more complicated machinery for achieving max voltage/current control (ie, with an extra op amp). See my ion laser SMPS for how the topology of mosfet, diode and inductor is changed if one uses an N-channel mosfet.
• Note: this circuit works well only if there is a minimum heat load, so that there is always a current running through the TEC. When having no load, the PWM controller temporarily shuts off until the temperature rises again, which leads to an oscillation by a degree or so. To avoid this one would need to augment this circuit to realize a bidirectional TEC drive.
• Bidirectional drive of the Peltier element (for both cooling and heating) would require a full H-bridge, consisting of 4 mosfets and 2 inductors (plus a current sensor involving an opamp) which would make this simple circuit considerably more complex. See for example here for mosfet H-bridges. I intend to develop such a circuit in the future, for controlling the temperature of Yag chrystals with high precision; the popular MAX1968 driver doesn't have enough power for cooling large Yags.
• Higher temperature stability may be achieved by using a separate high precision opamp for the comparator, instead of the built-in one.
• A simple 18V laptop power supply should be well suitable for feeding this circuit up to approx 100W.
• Here a measurement of the temperature over a few minutes, at 50W heat load:


A bit problematic is to get away all the heat, including the substantial power from the TEC. Alltogether this gives more than 100W to dissipate. This necessitates a cooling fan, otherwise the heat sink gets all the time warmer until the TEC cannot cope any more with cooling, thus leading to temperature runaway. Even a very large heatsink did not suffice in first tests.

Return to home page

Vers. 0.9 -05/09