Driving TEC elements

[blueskull]

Recently I am playing with some temperature sensitive things so I think I can use TEC coolers to heat or cool the sensitive part.
In this case, 0C~30C is fine, so I do not need accurate temperature control. However, I do need a simple and low cost solution.
My question is, can I connect a TEC rated for maximum 3.6V into a 5V system, and drive it with unfiltered PWM? The 5V can be lowered to minimum 4.5V, but that is still more than the TEC's rated max voltage.

[Pjotr]

My question is, can I connect a TEC rated for maximum 3.6V into a 5V system, and drive it with unfiltered PWM? The 5V can be lowered to minimum 4.5V, but that is still more than the TEC's rated max voltage.

NO! The AC ripple current greatly impairs its cooling performance. You can use PWM (and it is almost always done) but you need to filter carefully to get the ripple current through the TEC as low as possible. Preferable < 10%.

[Kleinstein]

The cooling power of the TEC follows a parabola, with a maximum cooling at about the maximum / rated current of the TEC. So more than the rate voltage will reduce the cooling effect. So a 3.6 V rated TEC at 5 V will have a similar cooling effect as at 1.4 V - less if cooling of the hot side is not perfect. This applies to the instant current with PWM too - so unfiltered PWM at 5 V is not a good idea. Even with just the nominal current it is not a good idea, as this will reduce the efficiency and lifetime of the TEC due to thermal stress.

So it's better to add a resistor or use linear control of the current than using more than the nominal current/voltage.

Also filtering with a kind of switched mode control (filter inductor and free wheeling diode) is an real option. There is no need to get the ripple really low. Something like a 10-20% ripple should not be a problem, so a capacitor parallel to the TEC is optional.