I'm designing a TEC circuit with a Peltier element, LM35 as the sensor and Arduino uno as the controller. I simulated it with Proteus and to my knowledge it was working, I suppose. But when I tried to replicate the circuit on a bread board with the components, I had problems with PWM. Can you guys please have a look at my circuit and advise for any changes/improvements needed. Thanks. I've attached the image of my circuit. Please find it in the attachment.

  • \$\begingroup\$ Could you please either: Upload a higher resolution image, or re-draw the schematic in the forum's schematic editor? I'm having a very squinty hard time reading values and names. That said, I think it's very likely to be your transistor choice. \$\endgroup\$
    – Asmyldof
    Jun 20, 2015 at 15:21
  • \$\begingroup\$ @Asmyldof thank you for your comment. I'm on the road and I'll try to get you a better picture asap. \$\endgroup\$
    – Ram.V
    Jun 20, 2015 at 17:22
  • \$\begingroup\$ what if the motor is brushless. Then what to do? (posting on behalf of Ram) \$\endgroup\$
    – user79283
    Jun 21, 2015 at 15:10

2 Answers 2


Looks potentially OK. Transistor drive may be too low.

Place a reverse polarity diode across fan so it can freewheel during PWM off.
Fan needs to be DC brush motor and not brushless (BLDCM) for PWM control. .

You should say what "problems with PWM" mean.
What is desired Peltier current.
See other answers here re driving Peltiers. Olin has things to say about it :-).

Better diagram copy is needed. I assume diagram is available at better resolution and has been munted by downsizing. Imgur allows files of any sensibly desired size.

Assume 5A Peltier current.
Assume 5V Arduino.

TIP122 datasheet here
Beta (current gain) is 1000 min at 3A.
Use 1000 for now.
Vbemax (from datasheet) = 2.5V (!) At say 5A Peltier current Ib = 5A/1000 = 5mA.
Max Ib is about (Vcc - Vbe_darl)/Ib = (5-2.5)/1k = 2.5 mA = too low.
That's wity worst case Beta (at 3A) and worst case Vbe and you can HOPE real device may be better BUT you must always design to worst case.
Making Rb say 220 Ohms or 33 Ohms may help.

Please advise correct figures for assumed value above.

  • 1
    \$\begingroup\$ Look at figure 10, collector saturation. Assuming the Arduino can supply 5-10 mA of base drive, for a 5 A TEC the transistor will dissipate about 8 watts. Heatsink time! \$\endgroup\$ Jun 20, 2015 at 22:24
  • \$\begingroup\$ Thanks Russel, it was a brushless motor not supporting PWM. We then added a single order RC filter before transistor which solved the issue. Also changed the transistor to BC517 as its more sensitive to low current (i only had this or tip122). Regarding peltier, we have adjusted the base resistor in accord with 2500 gain and Ic=4A..using 2.2K ohm at base. @ryeager, I have that if Peltier is operated above 300Hz PWM then it doesn't have problems \$\endgroup\$
    – user79283
    Jun 22, 2015 at 17:43
  • \$\begingroup\$ (It was a group project Ram.V and I are in same group.) \$\endgroup\$
    – user79283
    Jun 22, 2015 at 20:30
  • \$\begingroup\$ @StunningKun I STRONGLY suggest that you look at Olin's Peltier AC/DC comments. There is more than one issue. I assume that the 300 Hz PWM comment relates to thermal cycling. Olin hs thiongs to say about efficiency. NB - don't simply believe what he says - Olin is right about 92.3% to 99.7% of the time depending on external factors :-) Work through it with examples and be sure you agree. It's a non intuitive area. I need to go back and have a look some time. Part of my brain says I'm not yet 100% happy with the whole thing. \$\endgroup\$
    – Russell McMahon
    Jun 23, 2015 at 9:01

Another thing to note in your design: You also do not want to directly expose the TEC to your PWM due to it's thermal cycling limitations. When you PWM, you provide an "effective" rms voltage at the TEC, however, the TEC is actually still seeing voltages that go from 0 to VCC at various duty cycles, thus exposing the TEC to the max and min currents every period. Basically, this is bad and will reduce the life of your TEC. You need a passive second order low pass filter that has proper transient and frequency response based on your design requirements in order to make that voltage ripple go to somewhere less than 10%.


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.