Open source DIY TEC and Fan controller

[Ultrasonic2]

Of course the answers are different, as the two pic's have different variables and therefore the answers are different.(as you would expect)

However had we punched in the same variables there would still be a discrepancy.

People need to understand the difference between a calculated answer and a chart.

My calculator, when a number of variables are entered completes some mathematical equations and out puts a result. It uses the same mathematical equation no matter what TEC is selected. It in effect it assumes ALL TEC's are the same design and manufacturing process. However this is not reality as all TEC's are actually slightly different. .

A chart is produced when the ACTUAL TEC is tested and the results are plotted.

So Charts and calculations will never be 100% the same. This also means i can not fix my app because it is correct.

Custom thermoelectric TEC's are also More expensive than average TEC's so one would assume they should perform better than average, which according to the chart they do

EDIT
amperage will decrease as dt increase at a set voltage..
You are correct and my app does do this.

[Ultrasonic2]

Well getting back to the Actual topic

The controller is coming along nicely. Really all im doing now is adding in more features.

You'll notice there is NO soldering or bread board required so far. it can't get much simpler than this

IMG_1283[1].JPG
IMG_1285[1].JPG

Oh i've also got it running at 31.5khz now .sweet

[Liam_G]

looking good mate, can't wait to see it all set up and running in your PC.

[Ultrasonic2]

plenty of work before i do that

[Liam_G]

haha, yeah i know but proof of concept is a big step... how's the booster/mosfet side coming along?

[Ultrasonic2]

haha, yeah i know but proof of concept is a big step... how's the booster/mosfet side coming along?

well it largely depends on what frequency i can get the Arduino to run at and i haven't given up on 62.5khz yet.

The booster will be the last thing i do. I still need a way to turn this thing off when the machines off too.

and i still have to do the VB.net stuff as well

[Liam_G]

is that 31.5kHz or 62.5kHz the operating frequency of the PWM signal? is that not too fast for the TEC's? or is that just the operating frequency of the Nano itself meaning you can handle all the temp/humidity sensors without stuffing with your PWM signal?

I'm not familiar with the Arduino or how these things work so bare with me :P

[Ultrasonic2]

that is the frequency of the mosfet PWM . it can never in theory be to fast
The nanos code (CPU) is actually 16mhz. You can play wolfenstine on it !

[pby5cat]

I have no idea what the correct power down secquence for an Arduino is, but you cold just use a 12V relay that to command the on/off. Wire the coil to a molex conecter on the 12volt pins. that will make the relay open or close when the computer PSU turns on, and it will keep PSU power separate from the whole TEC system power.

As for turning off/on the TEC system, have the relay contacts control the neutral AC line running to the AC/DC power supply. Also you could run a contact signal to the Arduino on/off pins if it has some. Just make shure that if you cut the power to the Arduino, you also cut the power to the sensors, MOSFETS, screen, ect so you dont pull odd currents or reverse polarities.

[Ultrasonic2]

i dont believe there's a means to turn it off other than loss of power

So my current plan it to simply cut the usb cable in 1/2 and find out which wire is the +5 then bag a relay in their . now the relays power will becoming from the PC's 12v molex. Now since i'll already have a circuit board to do the above with the molex input i might as well bang a mosfet on there for controlling the FANS.

In the past i never turned the 24v psu's off because it wasn't necessary as the mosfets were blocking any current draw when the PC is off and i'll probably do the same this time too.

i've just added "keep the water temp X degrees C over dew point" even though i said i wouldn't but i figured it was simple for me to add and it's probably very hard for some else to understand whats going on.

I've also added in keep the hot side water X degrees C over ambient. which i think could be a cool mode to try.

Obviously there is the normal keep my hot or cold side at X temp.

Everything will be configurable from my VB app

[Liam_G]

hey ultra, question about the PWM frequency again, is it because your running such high frequency, 31.5 or 62.5kHz that your able to use that pass filter in your circuit? is it still possible to use a pass filter when the PWM frequency is about 16kHz??

I read something about the higher frequencies being better for a pass filter but not so much at a lower frequency because of the size of capacitors required and possibly heat or something, is that why your trying to get max frequency, or is 16kHz within range for the pass filter to be reasonable to use?

I'm asking because i'm investigating the feasibility of using the Aquaero 5 LT as the controller part. The Aquaero has 2 x PWM 12v outputs that operate at 16kHz, which might be too fast for a TECs response time without a pass filter and also might be too slow to make a pass filter feasible. Any thoughts?

I really like the package with the Aquaero 5 LT, good software, lots of options, analog fan control channels, can power my MCP35X pumps off it with PWM etc, seems like its got everything i need except the driver board/mosfet add on. But it will only be useful if the TEC and mosfet driver board can respond to that 16kHz frequency.

[Ultrasonic2]

yes the higher the frequency the smaller the cap has to be

16khz isn't to slow it just means you need bigger caps

what do you mean ?

which might be too fast for a TECs response time without a pass filter

Aquaero 5 LT have seen their stuff before Germans make such good stuff . This should work


it will be fine with my mosfet and low pass just double that capacitance i use (that assumes im running at 31k)

EDIT
it wont do anything about dew point will it ?

[Liam_G]

yes the higher the frequency the smaller the cap has to be

16khz isn't to slow it just means you need bigger caps

what do you mean ?

which might be too fast for a TECs response time without a pass filter

Aquaero 5 LT have seen their stuff before Germans make such good stuff . This should work


it will be fine with my mosfet and low pass just double that capacitance i use (that assumes im running at 31k)

EDIT
it wont do anything about dew point will it ?

ok cool, thats fine then, just wondering if it would work with 16kHz......

well about the TEC's response time, i didn't really know what to call it, but what i mean is that there are a lot of different opinions on the net about what PWM frequency is ideal for TEC's, i have done a few hours of research today and couldn't find any definitive answers as to what frequency is best, have heard as low as 120hz all the way up to 200kHz mentioned... However i have come across a few threads where it was said TEC's didn't respond well to certain frequencies, 10-20kHz was mentioned as possibly causing strange behavior in the TEC, also below 1kHz was mentioned as being too slow, and higher frequencies, say 20kHz and up were said to be just interpreted as either on or off depending on the duty cycle. Like i said i couldn't find anything definite so this could just be ppl occasionally having problems and i can't really find any specs on what the TEC's should run at so can't really tell if it's ok to run them at 16k or not.

So i'm wondering if the TEC will run at 16kHz without a filter with any issues, or at that frequency do they require a filter so it behaves more like plain old DC? have you run a TEC at this frequency with or without a filter? Also does the mosfet have to be chosen to work with a certain frequency range or do they just work with whatever they are sent?

No the AQ5 doesn't have the dew point capability, although i think you could rig something up that would work, it doesn't have a humidity sensor as far as i know but you could just do an offset above ambient or something, not ideal but it would work most of the time..... but i will just set a temp for it to maintain and i will be insulating to prevent any problems from condensation so should be sweet, I may not run it much below ambient all the time, but i want to just set a temp and leave it, for similar reason as you stated earlier in this thread, i want consistent temps and overclock, not one that automatically adjusts up or down with dew point.

And yeah the AQ5 does look sweet, can't really fault ze germans :P

[Ultrasonic2]

well i like to base my opinion on facts so this is what the TEC manufacturers say

http://www.tetech.com/FAQ-Technical-Information.html#19

d) Temperature control methods also have an impact on thermoelectric module reliability. Linear or pulse-width-modulated (frequency at least 300 Hz) control should always be chosen over ON/OFF control to ensure better reliability. The ON/OFF type of controller basically causes thermal cycling and so should be avoided.

you wont need a filter at 16khz the point of the filter was to increase reliability and LOW duty cycle efficiency

[bds71]

i've just added "keep the water temp X degrees C over dew point" even though i said i wouldn't but i figured it was simple for me to add and it's probably very hard for some else to understand whats going on.

I've also added in keep the hot side water X degrees C over ambient. which i think could be a cool mode to try.

OMG!! this is EXACTLY the type of circuit i was looking to build. where can i get one?!?! since my house here in Germany is not A/C'ed the temp (and humidity) fluctuate drasticly. this type of circuit is exactly what i was looking to try to build/acquire.

[Ultrasonic2]

OMG!! this is EXACTLY the type of circuit i was looking to build. where can i get one?!?! since my house here in Germany is not A/C'ed the temp (and humidity) fluctuate drasticly. this type of circuit is exactly what i was looking to try to build/acquire.

Yeah same here

Well the parts are listed earlier in this thread so just go buy them.

I've decided to buy an oscilloscope so i know for sure whats going on. It's should be pretty cool but like everything it all costs money, and that is the limiting factor so development on this will have to wait till i have more cash again.

im really looking forward to seeing what the T-balancer is actually doing

[Ultrasonic2]

Right well i have been working on the finer details of my PWM controller with Low pass filter.



i have gone to a D2pak Packager for my Mosfet and resisters. this is so i can surface mount them and have one heat sink that covers the mosfet AND resisters for the low pass filter
Im also gong to have 3 0.27ohm resisters for the low pass filter to lower the losses and it enables me to pass only a 1/3 of the current through each.

The mosfet is pretty cool instead of having one source wire it has 5 so that means it can actually pass far more current in reality than the ones with only one ! it also means the circuit board trace can be 10 times wide! which is important because it's only 0.035mm thick !

Now the caps have changed im now using 10 Panasonic 25v 680uf Caps this was done because caps dont have a resistance of 0 so having 10 drops their resistance by 10 and increases their max current the caps can discharge by 10. Because the caps have a resistance it's not possible to cycle one at 30 amps as it will get to hot and explode. i've gone also for high quality caps

[Ultrasonic2]

i have just realized that the low pass filter will Pretty much NEVER work with a 12v system . it's pretty dodge with 24 volts so i've gone to 36 volts now by wiring my psu's in series.

At 36v the low pass filter is dissipating 14.8watts. 4 TEC's in series
At 48v the low pass filter is dissipating 26.3watts. 4 TEC's in series

At 12v the low pass filter is dissipating 184.8watts. 4 TEC's in parallel !!
Even if you made 4 completely separate low pass filters it would still be a combined loss of 84watts

[Liam_G]

hmm, how about reverting to a lower PWM switching frequency of 1-2kHz and doing away with the low pass filter if its that much trouble?? I know nothing about these things but from discussions with a friend running the mosfets at such a high frequency of 31.5-62.5kHz will cause problems and it sounds like the low pass filter is going to have a hard time smoothing out that frequency. Like I said though, i know nothing about these things.

[Ultrasonic2]

i doubt i'd ever got to 1-2k unless i have to. i doubt i'll have a problem at 31k the mosfet is capable of 40mhz

The Low pass filter should work i just need to have the TEc's in series. Which i do now anyway and the advantages of having a low pass filter are Many


I've never sat around and worked out the the impact of having the TEC's in series would have on the losses of the other components. id always known that doubling the voltage would result in 1/2 the current but at the same time im doubling the resistance of the TEC's to keep their input voltage the same and therefore any other resistive losses like the mosfet and and wire are instantianiouly less. The result in having the TEC's in series is their is a lower voltage drop across these other resistances meaning there is less current and a lower Vdrop = FAR less losses to heat. SO I'll NEVER go back to a lower curcit voltage cos losses are exponentially worse for EVERYTHING at lower input voltages.



Since im getting an Ossilliscope i'll know exactly how well it works or NOT.

[Liam_G]

ok cool, thanks for the explanation. hope it works out.

[pby5cat]

remember if you are putting out 50A on ur drains 5 legs, there is 50A entering the source pin, which i assum is the tab so you will be fine. Are you using N or P channnel?

25volts on the cap ratings looks ok, be carful you do not exceed that 25v with your 48v PWM signal.

looks good i cant wait to try this out.

[Ultrasonic2]

remember if you are putting out 50A on ur drains 5 legs, there is 50A entering the source pin, which i assum is the tab so you will be fine. Are you using N or P channnel?

25volts on the cap ratings looks ok, be carful you do not exceed that 25v with your 48v PWM signal.

looks good i cant wait to try this out.

Thanks man
It’s N channel so the 5 legs are Source and the drain this the tab bit at the top of the pic.
At 36v it’s only pulling like 12 amps I think
The caps of course can’t handle that so I’ve just bought 10x 50v 2200uf caps for a total of 22,000uf, if I can fit them all on and a current of about 20amps.
So there should be stuff all ripple actually there should be less ripple than the input ripple from the PSU.

I actually doubt anyone else will every make a low pass filter cos everyone is obsessed with 12v to a fault

[mindchill]

Do yourself a favor and post on an electronics forum... they will point you in the right direction

[Ultrasonic2]

right so i got my Scope today and have found some very interesting things.
With the T-balancer i was right you HAVE to remove the cap from the board so it out puts a PWM signal otherwise it's analogue. Also the frequency of it is MUCH lower than what the interface tells you it's about 320hz

As for my project the arduino does switch at 31.4khz. i might be able to get more but im now worried about increased heating that may happen at 64khz.

i also noticed that i may not actually need the resisters at all to get my Digital to anologe converter to work. also i have changed my Mosfet again to one that will allow more current to flow at the low 5v gate voltage.

I have also realized that even though my TEC's maybe drawing 10amps (SAY) that the mosfet will be passing more than that as it tries to charge the caps. This also means the draw on the Psu's will be greater so i'll probably put some of my caps on the PSU side to keep the input voltage to the mosfets up