Thermoelectric (TE) Coolers

[jollyrgr]

I have a question about using PWM to drive a Thermoelectric (TE) peltier junction device. While researching these I found a number of references to using PWM to get higher effiency out of these devices. What kind of duty cycle and frequency works best? This should be quite simple to do with a 555 timer and a power transistor such as a 2N3055. While the spec sheets suggested using PWM, none gave any specifics.<p>None of the commercial coolers I have ever seen use PWM. Is it worth the trouble?

[Ron H]

Looking at a datasheet from TI,<p>http://www-s.ti.com/sc/psheets/slos365a/slos365a.pdf<p>there is an equation on page 8 that relates the loss in cooling ability to the ripple current. Higher ripple current yields lower cooling capability. Therefore, you should minimize ripple current (within reason). This gets easier to do as the switching frequency increases (smaller filter components). However, you do have to take into account the efficiency of your switcher as a function of frequency.<p>The efficiency gained by using PWM is all in the regulator - not in the Peltier device itself.<p>I bought a commercial cooler (for food), but returned it when I realized that it could only maintain about 30-35 deg F differential between inside and ambient. I don't know if the regulator was linear or a switcher. When running off a car battery (as this was), it may be cheaper for the manufacturer to use a bigger fan to remove the heat from a linear regulator than to use a switcher, since there has to be a fan anyway to remove the heat from the Peltier device.<p>If you use an LC filter, remember that the inductor has to handle its peak current without saturating the core.<p>If you decide to try a 555 for your PWM oscillator, I would use a FET instead of a 2N3055 for the switch.<p>Ron

[bodgy]

If a Peltier device is going to maintain an approximate temperature only, then it can be powerd by straight DC. If on the other hand you need it to run some form of feedback system- in other words it needs to stabilise its cold side to variations on the hot side -like a thermostat then it should be run from a pulsed source like PWM. This avoids strain on the Peltier Device.<p>colin

[Chris Smith]

If you want a colder temperature, stack them.

TE's are used to reach close to absolute Zero. <p>You can multiply them both parallel and in series to extract both quantity of heat as well as absolute heat temperature.