atlaswalkedaway

modified 10 years ago

Hopefully an arduino controlled. Hbridge driven peltier temperature controller

364

03:20:40

The goal is to control a single or set of peltier modules with an arduino WITHOUT using direct pwm switching, as this stresses the hell out of peltiers and seems to hurt their efficiency. The mosfets at the top will be replaced with logic levels and will act solely as switches. The transistors on the low side will ultimately control the amperage through the peltier. The signal for these is the smoothed pwm signal of the arduino as represented by the digital signal located bottom right. The manual two pole switches that reverse polarity will obviously be handled by simply dedicating two arduino pins to the digital signal for the fets and two for the simulated analog. To adjust polarity flip both switches.
To adjust current, scroll the pulse width.
The adjustment rate is very slow because were viewing at the common arduino pwm frequency. to allieviate this so you can more readily observe the ranges just drop the capacitor by a few tens of micros, or just slap a dc voltage source at the bottom of the right switch and scroll between 0-5v.
If anyone sees something that won't work or needs assessed I implore you correct me. I have a decent amout of experience with component level engineering but I have my foot in the door of so many other fields that I've never had the chance to truly delve into the finer nuances.

published 10 years ago

WTFCircuit

10 years ago

If you connect a peltier module backwards you'll destroy it (experience)

atlaswalkedaway

10 years ago

Did you have an adequate heatsink on what was the hot side under reverse polarity? I was recommended using the peltier as both heating and cooling in its respective polarities by an engineer who works for a peltier manufacturer.

WTFCircuit

10 years ago

Yes, I used a big heatsink that did the job when it was wired correctly but when I connected the module backwards with the heatsink in the hot side with reverse polarity it broken in a few seconds...

atlaswalkedaway

10 years ago

well, I think this may chucked up to a bad module then. You're comment prompted me to confirm so I flipped physically, then flipped polarity and the old hot side, now cool side has been holding at around 7F for 20 minutes or so. I'm going to keep it on for an hour to see for sure but I'm fairly certain they don't care about polarity, only that heat is removed from the hot side.

atlaswalkedaway

10 years ago

Looks like it actually goes into the negative degrees F if I don't measure it with my finger first. Lol

atlaswalkedaway

10 years ago

For the sake of being thorough for others the modules I got are at this link. Http://WWW.amazon.com/gp/product/B00E4GY0FC?redirect=true&ref_=ya_st_dp_summary

faceblast

10 years ago

yeah don't reverse polarity on them, they pop. put a couple of 1n5401 on one side to prevent mishaps. I put mine on 200mA constant current supply to identify and mark the cold sides before mounting. i wonder if an adjustable constant current supply would be more suitable for controlling temperature on these peltiers?

atlaswalkedaway

10 years ago

Well its been running since the first comment and I just scraped enough frost off of it to pelt my gal with a snow pebble. In the face of opposing data coming in I'm going to have to side with what I'm looking at at this very moment. There's also a plethora of people saying this is possible online, and only a few that claim it isn't.

atlaswalkedaway

10 years ago

I do appreciate the feedback @WTFCircuit and @faceblast. It's not taken without heed. In fact I respect you too as much as any of the few other major contributors on here, and you're advice on others circuits has been quite valuable to me. So if we assume that my module working fine for hours now is even a weak proof that (for some modules, under certain conditions) reverse polarity is possible, would this then be an adequate circuit topology for the goal I have outlined? For the sake of cohesive thought we can just assume that I want to control a heating element in this rather specific and convoluted manner.

atlaswalkedaway

10 years ago

*your advice

WTFCircuit

10 years ago

Maybe some peltier module can be reversed and someone not. If it's that I was very unluky.

faceblast

10 years ago

I tried reversing one of mine with very low current and it got warm on both sides

flowDAQ

10 years ago

This stuff is interesting and confusing, it looks like the minority carrier dominates the heat flow effect, so I don't think these things could work very well when current flow is reversed. That is not to say that some heat flow won't happen, anyway the best details I found: http://thermoelectrics.matsci.northwestern.edu/thermoelectrics/index.html

flowDAQ

10 years ago

Why is it bad to directly drive the TE with PWM? My guess is a thermal shock of the P and N semiconductors stacked between the TE plates. If that is true then you need a soft start or some sort of current ramp when the power is applied. It may even be possible to ramp up the PWM time to let the heat gradient established gradually. For example in a reflow oven I recall that large ceramic capacitors need to be ramped at something like 1 degC/Sec or they can fracture inside, which can cause leakage current that may screw up the timing of a one shot (one of my present problems). I have no clue what a TE can tolerate, but if it is like a ceramic cap then the soft start would need to be fairly long.

atlaswalkedaway

10 years ago

@flowDAQ that's effectively the reason I was given by my peltier expert buddy. They'll "work" on pwm, but the constant thermal shock reduces lifespan and the heat allowed to backflow during the off portion of the signal further hampers the already low efficiency. also, the site you reference shows an example with only two simiconductors. Once you extrapolate the examples given there to the actual internal topology of the module, which is many alternating p and n doped pillars in series, its fairly easy to see that under reverse polarity it should act in a nearly exact opposite manner. I haven't looked further into the following, but I saw a reference claiming that both wires are even connected to a p type so that the internal operation is as identical as possible under either polarity. Again, don't quote me on that one cause one reference somewhere on the internet might as well be hearsay. @everyoneelse I'm a very private person, but if need be ill sign up for a YouTube account and upload a video of myself clearly running a peltier in both polarities with essentially equal performance, accounting of course for the preheating of the heatsink I use during the first run.

flowDAQ

10 years ago

I'm reconsidering what I said, Minority carrier flow probably works in both directions in an n-type or p-type semiconductor as they are used in a TE, I am getting things mixed up with BJT stuff. But if the heat flow can be flipped, the thermal shock would be a major problem. The TE probably acts nearly as a resistor, so if it was put between a typical H-Bridge the current and heat flow would slosh back and forth at whatever PWM frequency it was driven at (and that sounds bad). If an inductor was added in series and the PWM was fast enough then the current and heat flow would not flip-flop, but you could still use something like an Arduino PWM output to operate the H-Bridge. An Arduino PWM is about 500Hz if I recall, but I think it can be adjusted up some which would reduce the size of inductor needed.

atlaswalkedaway

10 years ago

Take a look at the circuit, I've got the pwm signal flowing through a risistor, charging a capacitor, to essentially convert the signal to actual analog voltage that slowly ramps.

flowDAQ

10 years ago

I've looked, but it's not making sense to me, that does not mean its wrong. I need to understand what a TE is doing before I can offer any useful advice so I'll tinker with some thermal circuits and see if I can gain some insight. I need to find an old book I have on thermal conductivity, it had some info I used years ago to model thermal capacitance and thermal resistance with SPICE elements. Thermal equivalent circuits are easy, just capacitors, resistors and current sources, but setting useful values is not so easy. In the meantime search for @faceblast "Discrete H Bridge Controller" circuit. It may need some values adjusted but seems to be a good implementation. I can imagine connecting the IO's to an Arduino output pin to PWM the H-bridge. Actually if you look at the AVR chip you will see OCnA and OCnB pins (where n =0,1,2) are the PWM output. It turns out the A/B can be setup as complementary drivers with dead time, which is nice to work with H-bridges. If you put a big coil (3mH) in series with the load resistor in that H-bridge the simulation can provide clues about how current would circulate in a TE. But I fear a TE is not a simple resistor, it is more complex to model so I need to find that book.

flowDAQ

10 years ago

http://www.researchgate.net/publication/3849593_SPICE_model_of_thermoelectric_elements_including_thermal_effects

flowDAQ

10 years ago

I found that link, it shows how to model a TC with SPICE, but I don't think EC is going to work.

alkuzaayme

10 years ago

add comment in editor