Problem driving a Peltier (TEC) module with low voltage vIN (0.8v)

Hello,
this is my first post in the community and I’m a big fan of Arduino:)
I’m generally confident in programming and moderate confident in electronic but I’m stuck right now and I’m looking for advice.

For a project I have to: drive one (several if it works) Peltier (TEC) element. Running the same face hot or cold, according to user selection. Also with the ability to set the desired temperature (hot or cold).

Therefore my objective is: to drive a TEC element, maintaining the ability to inverse the polarity. Using cost-effective components because it’s a really low budget project.

Components: for a technical reason, I need to use this very small peltier. Rated 0.8v 3.2A.

Looking for a cost-effective solution for driving a TEC with Arduino, I found several posts here and on google, suggesting a motor h-bridge.
The problem is that usually the motor drivers are rated for higher VmotorIn voltage, >3 volts.
The only viable and cost-effective solution I found is this small driver from pololu.

I built the circuit, and it actually worked. (diagram attached)
But there are several problems:

  1. I have to turn up the voltage from 0.8v to 1.4v at the driver V motor In. Otherwise, I have no response from the TEC at all.
  2. Nonetheless, I have a significant voltage drop on the TEC. I supply to the driver 1.4v but I measure 0.4v on the TEC terminals.

The TEC works, it reaches the temperature, I can switch the polarity and so on.
But I do not understand the voltage drop, and the driver runs very hot.
I would also like to find a different solution because right now I’m driving the TEC only a 1.7A (driver internal limit)
While I could safely push up to 75% of the power (let’s say 2.5A).

So my question are:

  1. how can I drive a TEC element rated at 0.8v?
  2. How can I improve my circuit with low-cost components?
  3. Can I use PID and PWM with this TEC/circuit? In other h-bridge solution, I found PID and PWM a 200khz is usually implemented.

Thank you all very much,
Paula

I've never used a Peltier device but form what I understand they are non-linear (like LEDs & diodes) and they should be driven the same way... with current source so the voltage "falls into place". But, they may be more "tolerant" than LEDs because from what I've seen lots of people do drive them from a constant-voltage source.

It's normal to get some voltage drop across a driver, and the more current the more the voltage drop. But I don't know what to expect from that driver. ..A fraction of a volt usually isn't a big deal with a 12V motor, but the voltage-drop and current are related to the power loss (heat generated) in the driver.

Somehow low budget and miniaturize never seem to work out together.

Paul

You should be able to use two of those drivers in parallel, but post on the Pololu forum if in doubt. Be sure to give the details of the TEC.

The engineers there are very knowledgeable and responsive.

The motor driver DRV8838 is only rated 1.8A so may be inadequate.

A reversing relay as used on car electric windows may be a better bet.

A current sense resistor ( 0.5 ohm?) would allow feedback to control the PE current.

Allan

If you want cost effective use a DPDT relay to reverse the polarity. The alternative is
a full blown MOSFET H-bridge using gate driver chips and all n-channel MOSFETs (which allows
arbitrarily low supply voltage for the load, but requires 12V for the gate driver chips).

Relay is simpler and cheaper.

Getting 0.8V supply without wasting power in a series power resistor is tricky - maybe there's
a buck converter somewhere rated for the voltage and current (and not part of a computer
motherboard).

Use a switch-mode current regulator meant for driving high power LEDs. The voltage will take care of itself.

A stepper motor driver like the A4988 =is= a switch-mode current regulator, BUT it relies on the inductance of the motor winding and will not properly drive any non-inductive load properly.

The way buck switch-mode regulators work, the supply voltage MUST be greater than the output voltage.

That H-bridge driver is meant for a DC brushed motor, they self-limit current at the short-circuit current. It is NOT suited to driving a nonlinear load like a Peltier module. A Peltier device is just a series of PN junctions, in this case one single big PN junction. AKA a diode.

I see no reason to believe it, but could you offer some justification for this statement?

It is NOT suited to driving a nonlinear load like a Peltier module

Many people use brushed D.C. motor drivers to control Peltier modules, without incident.

I'll bet they aren't driving 0.8V Peltier modules.

Hello, and thank you all for your useful replies :slight_smile:
I took some time to experiment and I was waiting for some components to test.
I'll try to reply to all and post the results of my experiment.

Paul_KD7HB:
Somehow low budget and miniaturize never seem to work out together.

You are totally right Paul :slight_smile: TBH, I do not need to miniaturize all the components. I can go larger as I want with the driver and the stepdown. The Peltiers need to be arranged in a matrix, for a sort of thermal display, therefore the small element.

DVDdoug:
and they should be driven the same way... with current source so the voltage "falls into place".

It is an interesting suggestion. I'm no expert of Peltier. I focused on the motor driver because "the internet said so". Do you have any specific component that I can buy and test?
Led driver can reverse polarity? (i need to reverse polarity on the peltier)

jremington:
You should be able to use two of those drivers in parallel but post on the Pololu forum if in doubt. Be sure to give the details of the TEC.
The engineers there are very knowledgeable and responsive.

Thanks jremington, I followed your advice both on the driver both on the pololu forum. Here are the updates:

Updates:
due to the fact that the driver was surely inadequate, my first attempt has been to use two drivers in parallel.
It was an unlucky attempt. One of the two always runs hot as hell, and the circuit performances did not show any improvements.
Pololu discouraged to use two drv8838 in parallel because the chips can have different timings and therefore behave oddly. As I personally saw.
They redirect me to drv8835, although less than the 8838, the chip has two channel made to run in parallel up to 2.4A.
Fairly better that the 1.8A single channel in the 8838.

I bought a couple of 8835 and tried them out without any change in performance.
The circuit works (ish) the driver runs hot. Same voltage drop. And same poor performances from the Peltier. Somewhat poorer that the single drv8838, because the Peltier takes longer to reach the desired temperature.

Therefore... I am basically back to square one :slight_smile:
I haven't been able to find and buy a reverse polarity relay addressable by an Arduino. I only found push button relay.

I'm really intrigued by @polymorph ' idea to use a led driver, letting the voltage taking care of itself.

Two questions:

led driver can reverse polarity? I need to switch the cold and hot side of the peltier.
Could you please suggest me a suitable led driver I can order?
Something I can buy and test :slight_smile:

Thank you all for the support!

I haven't been able to find and buy a reverse polarity relay addressable by an Arduino.

You can use two SPDT relays to reverse polarity, available in a single module.

You DID NOT mention this before:

"You are totally right Paul :slight_smile: TBH, I do not need to miniaturize all the components. I can go larger as I want with the driver and the stepdown. The Peltiers need to be arranged in a matrix, for a sort of thermal display, therefore the small element.".

You only wrote of one Peltier device.

All Peltier devices are diodes in series, even within one device. There is absolutely no reason to try to power them in parallel. Put them all in series and add the volts up to get the total volts for your matrix.

You are making this project way too difficult.

Paul

Paul_KD7HB:
All Peltier devices are diodes in series, even within one device. There is absolutely no reason to try to power them in parallel. Put them all in series and add the volts up to get the total volts for your matrix.
lt.

Paul

Not really - they're a load of wide contact area low-impedance thermocouples in series.

If the OP wants to make a thermal 'map' of different temperatures in adjacent cells he may need to drive the modules separately.

Goodness knows why anyone would want such a gadget.....

Allan