I have read a few threads concerning voltage regulation, most of them from these boards, however I must admit I still feel like I've waded a bit too deep into unfamiliar territory. I have purchased 5 127W Pelts (15V 8.5A max) that I am trying to run in parallel, connected to a voltage regulator of sorts. I want to be able to run the TEC's through the 12V rails on a PC power supply. Ideally, I'd like to be able to set and maintain a temperature, and adjust it using a knob or LCD control. The snag I've run into is a power problem. The amount of amperage needed to supply 5 TEC's is pretty high, and I am afraid of shorting equipment. For this project, ambient temperature and condensation are not an issue, so the controller wouldn't have to be complex at all. I had the notion of being able to run this through some PC fan controllers: http://www.frozencpu.com/products/9..._PWM_6x_Fan_Controller_525_Bay.html?tl=g34c17, http://www.zalman.co.kr/eng/product/view.asp?idx=274&code=017 but again I think the Amp draw would fry those in seconds. I am experienced with hands on electronic repairs (soldering, re-wiring, trouble-shooting, throwing things I broke into the trash), but I am an 100% certified n00b when it comes to proper circuitry. I have read about PWM regulating for temperature control, and I have found a possible regulator here: http://www.maxim-ic.com/quick_view2.cfm/qv_pk/5103/t/al I would greatly appreciate any ideas, products, threads that might be useful. Thanks again. -TTL Locke
Is this for oc'ing? I'm not sure why you want full temp control? You can make a regulator to handle all those pelts - but i'm not 100% on how to do it. You'd need some beefy transistors such as big mosfets, it wouldn't be that cheap to make, probably $100, maybe less? And just in case you aren't aware, to run all 5 thats 42.5amps @ 12v so you'd want a psu that can handle that constantly, not one that's rated at 42 or even 45 amp max. You'd want 50amps max at the bare minimum from a reputable company. A 60 amp or higher rating is much nicer however, something like this You'd want nothing else running off it, besides some load on the 3.3v & 5v rails, to balance the psu out a bit. If you want a psu that will run a pc and those pelts than you gonna need something bigger... how big depends on your setup - unless you just get two psus. But you probably know that^
With proper temp control you can hold it just above dewpoint - no need for insulation etc that way, just like a waterchiller.
He says ambient temperature & condensation are not an issue tho... maybe he's making a fridge? It doesn't sound like oc'ing, hence the question
Thanks for the quick responses guys, and to answer the question this is not an OC project, so no need to be super precise and worry about condensation on components. Gusta, that PSU isn't bad, I have definitely been looking at those with 3-4 stable 12V rails such as the one you mentioned. The one I have in mind is the Zalman ZM750-HP 80A total on the 12V, so that should be enough to power petls and other components I would imagine. It's running between $175.00 to $189.00 on Newegg, right now, so I'm just waiting for it to hit the right price, or find one on ebay/craigslist for $25.00-$30.00 cheaper. Mofsets, that sounds workable. I'm not sure what those do, but I can look them up. For simplicity purposes, let's assume I'm making a TEC fridge with temp control. What is hands down the best way? What is the cheapest way? I'd like to be somewhere in-between the best and the most affordable. Let's say this is budgeted at $100.00 Thanks again, your input is appreciated.
I say use a good water cooling setup with plenty of radiator to cool the mosfets. Otherwise you'd have to go large heatpipe heatsinks with powerfull loud fans and the peltiers won't work aswell. You know there is another way, you could just have the pelts running at 12v, maybe 4 will be enough, and the only thing you have to regulate is the pump! Much easier i think, and would give good temp control cos if you slow the pump right down the pelts will warm up quickly and if you turn the pump up the pelts will cool down quickly - provided you have plenty of radiator in place to keep the water cool. Airflow over the rads doesn't need to be loud either. I think there are ready made temp. regulated controllers for fans you might be able to use to control the pump if not, it won't cost that much to make one. I'd rather controll the pump than the fans because it will give a much faster response time in terms of temp control. If you went with the large heatpipe option then you can just use a fan controller but again, the peltier temps won't be as low compared to water cooling.
Interesting, I never thought of the pump idea. To clear things up a bit as I may have been misleading, I didn't plan on regulating temperature by controlling the fan speed. Rather, I thought I might be able to control the voltage being supplied to the TECs using a fan controller. I think voltage regulation might be the most effective way in the long run. I have 10 solid copper injected heatsinks that will be liquid cooled per se on the cool side, while heat is dissipated by 2 120mm fans in a seperate chamber above. I assume that with the TEC's running in parallel will increase efficiency down the line, by reducing the power the next one has to move (at least that's what I hear). At 12V these things will run at about 6A each, which is 360W of power by itself.
I didn't mention earlier that yes, a fan controller will burn instantly or overload protection will kick in if you hook up peltiers to it! Yes peltiers cool more as the power increases & vice versa and the hot side also gets hotter with power increase & vice-versa. It's just controlling temps by varying peltier voltage means having a $100 voltage regulator - which is hard to design/make - if one that powerfull exists ready made it would be like double or triple the price. But if you just controll the pump voltage instead, you only need a regulator that can handle a couple of amps or so = cheap, and easy to make or buy ready made. The money you save can be put into the water cooling setup. Even if you made a powerfull vreg for the tecs you would still want a good watercooling setup. Tecs need to be cooled on the hot side, and the cold side cools down whatever it is you wanna cool. The cooler the hot side is run, the colder the cold side will get, within the limits of the TEC. That's why varying the pump voltage/speed is a much easier option. With a water cooled tec of about 70w, i got plenty of condensation and within 20 minutes of turn on, ice had formed, covering the tec, about 5mm thick. Turning off the pump with the tec still on melted the ice quickly, and the tec will get so hot it will fail if you don't cool it so i had to turn her off soon after.
I figured the fan controller would crap out, but I assumed you could simply make one large enough to handle the necessary power needed. I don't want to watercool the TECs, because I don't think I'll have the space for the components. Since the TECs won't be directly attached to hot PC components I think this will be ok. I have solid copper heatsinks for the top and bottom of the TECs. Top: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&rd=1&item=270119557404&ssPageName=STRK:MEWN:IT&ih=017 Bottom: http://cgi.ebay.com/Lot-of-6-Nord-p...oryZ4660QQssPageNameZWDVWQQrdZ1QQcmdZViewItem I got the TECs with the bottom heatsinks, I thought it was a pretty good deal actually. THe TECs will be arranged in a semi-circular arrangement. (I had a mock drawing here, it failed) This will be enclosed in a seperate section with 2 120mm fans providing constant circular airflow through the fins of the top heatsinks. I think this should be sufficient to cool the top of the TECs. I have the entire rig drawn up in AutoCAD. Pics in a few minutes.
Here, this is relatively what it will end up looking like: http://www.imagehosting.com/show.php/1515665_TECArrangement.JPG.html
In case you didn't know, the amount of current that a pelt will draw is proportional to the supplied voltage. For example (pulling these numbers out of a hat) it might draw 8.5A at 15V, but at the 12V you're running it at, it'll draw a little less, maybe 7.5A.
You're absolutely correct. In fact, it's not a straight line either, more of a curve. So around 6A would be a more realistic value. 6A at 12V = 360W total power (with 5 TECs)
Indeed. I was disappointed to find this out when I wanted to run my 172w 24v pelt at 12v. I didn't get near half of the performance.
Could I use something like this: http://www.linear.com/pc/productDetail.jsp?navId=H0,C1,C1010,C1603,P1975 To control voltage to the TECs? Its rated to 5A output, so could I hook up 5 of them in parallel? All I want to be able to do is control the voltage being supplied by the pelts from the turn of a knob. Is that too complex/expensive for what I'm trying to do? I appreciate your help on this everyone. -Locke
In parallel, at full power, you will need number of tecs * current at full power. So 5 TECs with 8A current draw will need 5*8 amps to run them in parallel, at full power... 40 Amps, as someone mentioned previously. Due to the high power, PWM might be a better option as it allows you to use much smaller heatsinks on the components that power is going through. If you use a linear control (i.e. not PWM), either you want one big mofo transistor which can handle that current (TO-247 package and associated heatsink), or many more in parallel (5 TO-220s or something). 45A MOSFETs cost about £5, plus the heatsink. What circuit you want depends on if you want to set the temperature with your control knob (i.e. like an air conditioner thermostat) or if you want to set a 'power level' and just go with whatever temperature that leads to. The latter is simpler, you can just use a potentiometer to control the current through some kind of high power transistor (MOSFET or BJT). Going a step further you can use a potentiometer to set the current through a current regulator (such as an LM317, although that probably can't handle 8 amps without a monster heatsink). If you want it temperature controlled (e.g. you set the target to 10'C and it takes care of the rest), you need to set up a negative feedback loop. One one end, you have a temperature sensor connected to one of your TECs [cold side]. In the case of most thermistors you can work out the temperature with some function of resistance provided by the manufacturer, and vice versa. You would use the thermistor as part of a potential divider. Then, you would have another potential divider consisting of your potentiometer which you use to set the target voltage. If the thermistor scale isn't linear you might want to either linearise it somehow or do something to the potentiometer's output to make it have the same sort of 'function'. Then you compare the two, and use the difference to control the power to your TEC (positive difference, increase the power, negative difference, decrease the power). So. this way is more complex, but if you can get a more linear thermistor then it can be still quite simple. In either case your output can be used to modulate a PWM controller, if you go that way. ciao
Thanks Cinn, this is starting to make sense to me. I don't really need the unit to be set at a specific temperature, as I don't have to worry about condensation or anything. I planned on using the Zalman fan controller display to monitor temperature from 4 different areas, so I'd still be able to see what temp certain things are at. Ideally, I would be able to control the power going to the TECs through an external control (Like a volume control knob) Can you show me a circuit setup that would show a potentiomenter controlling current through a current regulator? I would target the output current at 5-6 A max for this application. How big a heatsink do you think I would need for that? I am getting an idea of what you are throwing together, but any pictures or diagrams would really help me out, I'm such an eletro-n00b. Thanks again. -TTL Locke
Another way is to have one or two of the TECs running 100% and control the others. Putting TECs in parallel sounds a bad idea, you're better having each on its own switching MOSFET using a common signal voltage from a PWM or on/off thermostatic control system.
You're probably right, that much power in one line can't be a winning solution. Could I have all the MOSFETs controlled by a single thermostatic control safely? CP, you are brilliant. I don't know what my problem is. I could easily control the power simply by having 1-5 of them on at a given time. Amazing, sometimes the simplest solution evades you. Now, how could I rig up a control knob to act individually as a rocker switch for 5 different devices? Or is there a better way? Thanks again you guys, this is headed in a possible direction.
IC1 puts out a 12V "high" or 0V "low" signal and will turn any number of parallel n-MOSFETs on & off, each with a TEC load. IRF530 Cheesecake.
You could have any 5 posistion rotary switch that controls 5 high power relays. I'd suggest big relays like 30a+, what they use on car horns or bigger. You'd want such high ratings as the high current eat at the contacts in the relay. I'm sure cp & cin know of a way to reduce this, maybe a capacitor in parrallel with the contacts? EDIT: JUST SAW CPEMMA'S POST, NO SURPRISE HER METHOD IS BETTER THAN USING RELAYS If you control temps this way, it's simple but it's not as accurate as using voltage control - you can go pwm but i still think controlling the pump flow rate is the easiest & most accurate way of controlling temps. You have got to use a much better cooling method btw, if you aren't gonna watercool then you are gonna need big powerfull heatsinks with big powerfull fans. Those heatsinks you have are no-where big enough and all you are gonna have is 5 heaters. Those pelts will not get cold at all - pelts are better at heating than cooling, you need a really powerfull air cooling setup to get them cold, and a good water cooling system to get them really cold. The "cold" side will always be cooler than the "hot" side, but the "cold" side won't get cold unless you keep the "hot" side cool. You should hook one of those peltiers up to power for about 10 seconds with that little heatsink on it. But before you touch it to see how hot it is, but beware, you might burn yourself, lick/wet your finger first
