I need a controller for a 12 V - 1A fan, blowing on the cold side of a Peltier element to keep it at constant temperature. I have the hot side under control.

I have seen multiple cheap PWM ready-to-install controllers online that do the opposite: make a PC fan faster as the sensor gets hot (like this controller) or (like this controller).

I thought about three speed ranges, (so the fan does not keep readjusting non-stop): fast at 11 °C (52 F), medium at 13 °C (55 F), slow at 16 °C (61 F). That way the temperature would be measured every 60 sec, and adjusting the fan speed in intervals, allowing the peltier to cool down until the next measurement 60 sec later.

How can I change the outcome of the pre-assembled controllers? Would a PNP transistor work? Or a Mosfet that decreases conductivity as the voltage increases? I attached an image for reference (but I am a basic electronics user).enter image description here

  • 2
    \$\begingroup\$ Welcome to EE.SE but I advise you to edit your question and remove the last bullet point as shopping questions will result in fairly prompt closure of your question as per site rules. While you're at it, can you clarify your application and specify what temperature the COOL air is at? Is COOL higher than 15°C? \$\endgroup\$
    – Transistor
    Feb 18, 2019 at 19:21
  • 2
    \$\begingroup\$ The problem wasn't the Amazon link - it's the "I would like to find someone who can charge me to design the circuit and manufacture like 5 PCB for me" bit. \$\endgroup\$
    – Transistor
    Feb 18, 2019 at 19:30
  • 2
    \$\begingroup\$ OK, it's getting a little clearer although we're still missing the spec for "COOL". Your phrase "a way to always blow the cold side at the same temperature" suggests that you think fan speed controls air temperature when in fact fan speed will control the rate of heat transfer. \$\endgroup\$
    – Transistor
    Feb 18, 2019 at 19:45
  • \$\begingroup\$ tried to slim down your question to focus on the core statements. \$\endgroup\$ Feb 18, 2019 at 19:48
  • 1
    \$\begingroup\$ Many of the cheap fan controllers are PWM, and use a microcontroller to generate the PWM value based on a value read from an analog input. You might be able to reprogram one of these to work in the opposite sense, or reverse engineer the one you have pointed out, which looks to be a more discrete design, with a couple of comparators on board, probably set up as an astable oscillator, or maybe running in linear mode (hence the big heatsink). I'd bet that just swapping a couple of connections around on that board would save you creating one from scratch. \$\endgroup\$
    – Phil G
    Feb 18, 2019 at 21:56

1 Answer 1


You cannot simply invert the output of a controller. Then the thermal negative feedback loop becomes a positive feedback loop and latches in one direction.

The inputs must be reversed with thermal negative feedback so that the thermostat regulates the cold temperature or minimum value rather than the hot.


You need to avoid adding heat with a fan on the cold side yet remove as much heat as possible on the hot side.

THis is a simple way to make a cold thermostat.


simulate this circuit – Schematic created using CircuitLab

A better approach may be to maximize thermal conduction on hot side with continuous forced air exhaust to remove heat and cold side with large passive heatsink and then regulate Peltier device current for regulating cold side temp. Efficiency improves by reducing the temperature drop.

  • 1
    \$\begingroup\$ I appreciate your suggestions, @Sunnyskyguy EE75, but I do not want to deal with the hot side (I have that covered). My challenge is not related to the Peltier itself, but to control the speed of the fan on a progression of cold (not heat as the ones on Ebay do). \$\endgroup\$
    – Paul Pn
    Feb 19, 2019 at 20:12
  • \$\begingroup\$ the Ebay ones might not work if the polarity of feedback is inverted to prevent over temp rather than under temp. But you have my design,If you use an open collector comparator add pull up to Base = 2k, Vref must be stable if derived from Vcc \$\endgroup\$ Feb 19, 2019 at 21:34
  • \$\begingroup\$ They could be modified by2 cuts and jumpers to swap +/- comparator inputs and some offset for low temp but I cannot tell you how \$\endgroup\$ Feb 19, 2019 at 21:39
  • \$\begingroup\$ Thank you. I will do some research. I appreciate your help. \$\endgroup\$
    – Paul Pn
    Feb 19, 2019 at 22:25
  • \$\begingroup\$ I wonder what's wrong with the person who voted -1. There is nothing obviously wrong here, and those controllers are black boxes without documentation to modify \$\endgroup\$ Feb 20, 2019 at 5:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.