• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Pelt power supply question

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Ct. Strangelove

Member
Joined
Aug 25, 2002
Location
bleek abyss
here is a simple on for all of you.

I want to buy a 510watt power supply to power my 226w peltier.
But at 12v 226w goes down to about 183w(not sure).
I want to use a full 15v. Can take the 12v and the 3.3v lines of a ATX power supply and combine them for my 15volts?

thx
 
I dont know if that would work but one way to test would be to get an old and CHEAP power supply and connect it up if it goes bang It dont work :)
 
There is another guy with a question in this forum, who has a 15V psu, see if getting that one would be as cheap as getting the 510W psu. Then you would have 15V without messing up the wiring/loads.
 
Their 226 watt pelt MOVES 226 watts of heat.
It CONSUMES 364.8 watts of electricity to do this.
Do the math from their website's numbers....you'll need more psu than you are thinking!

I'm not picking on you bud, but everyone seems to take for granted that the wattage rating is both it's power moved and it's power consumed, that is simply not so. I wish websites would make this more clear.
 
Cheers Diggrr I recently purchased a 156w peltier and it said on the web site max 16A @ 14.4v. so i calcuated this and it came out as 230w, and I thought how can 230w be 156w now you have made it clear and your right websites should make it more clear about how much power pelts consume!
 
i only gave the website link for how to run the powersupplys in series to get hte higher wattage i wasnt saying do exactly what they said. it was a here is how to do it.
 
Couldn't you just bump up the 12volt line a little to get you a little closer. Like how they do 3volt and 5volt lines in a PSU. The reason I ask is cause I'll be getting a Antec 510 watt PSU to power my 226 watt pelt.
 
Diggrr said:
Their 226 watt pelt MOVES 226 watts of heat.
It CONSUMES 364.8 watts of electricity to do this.
Do the math from their website's numbers....you'll need more psu than you are thinking!

I'm not picking on you bud, but everyone seems to take for granted that the wattage rating is both it's power moved and it's power consumed, that is simply not so. I wish websites would make this more clear.

Only took a couple of hours to fry my pyramid psu at 15v. Psu rated 15v @ 32 amps, adjustable. Know why they have a little line at 14v now.:(
 
You can find the following paper on the web:

Universal Thermoelectric Design Curves
Richard J. Buist, Marlow Industries Inc.

I'm too tired to look up the link right now.

In short it says that at 12 volts on your pelt you will get about 95% of it's total heat pumping capacity. Powering the pelt at it's Vmax will put 137 more Watts into your water cooling system than powering it at 12V, and it would only increase the pelt's heat transfer by 11.3 Watts.

The increase in the temperature of your water will more than offset the increased heat transfer of the pelt. Your CPU will actually run hotter if you power the Pelt at 15V than at 12V.

It would make more sense to run a second pelt as a water chiller.

And to answer your original question:

"Can take the 12v and the 3.3v lines of a ATX power supply and combine them for my 15volts?"

No, you can only get the difference between the two voltages.
+12V - +3.3V = 8.7V

In addition connecting a 226W pelt between these two voltages would almost definitely fry your whole system.
 
Hey [EG]~NaTz~, that's actually a pretty usefull article. I've got one I made from that article, and there's one guy here who (I believe) has 4 or 6 of them linked together. Works like a charm (at least till I got my big 40 amp Lambda <sp?> laboratory psu).

I'd stick to 12 volts though like Since87 said. There's not much to be gained by running them at full voltage except a bunch of heat that needs to be removed.
If you lower the voltage it's running at, you'll also lower the current draw it requires, though I'd always buy bigger than needed by 25+% just to be safe, and to allow the psu to run cooler.

Again...just trying to help, hate to see a computer die.
 
Holst said:
Read this on hot to make your own pelt supply that runs from the mains.

using computer PSU is a very poor way of making a pelt supply, I just wouldt trust it to be reliable, PSU are not designed for those sort of loads, even expensive 550watt Antecs you will kill a expensive PSU.

http://www.coolhardware.co.uk/modules.php?name=Forums&file=viewtopic&t=152

Yes, there is a lot to be said for using an unregulated supply as discussed in that link. Switching supplies are in common use these days because they can be smaller and maintain high efficiency over a wide range of loads.

To power a pelt at a constant voltage, an unregulated supply can be a cheaper way to go. It's going to be really heavy and bulky though. Not likely you'd want to fit it into your case.

As far as using the Antec: I don't know what current the rest of the PC draws from the +12V rail, but the 226W Pelt will pull 80% of the Antec's maximum current rating from the +12V rail. That leaves only 6 Amps left over to power Fans, drives, etc. before you are exceeding the max spec of the Antec. If you are not careful it is quite possible that the Antec will be destroyed by the combined load of the Pelt and PC.

I do believe the all around best method of powering a pelt is with a dedicated switching supply. However, there is a wide range of practical options.
 
The bits for an unregulated supply will fit in a fairly small box, and it is much cheaper then a regulated job.
Difficulty is getting enough capacitance to remove any AC ripple (thats very bad for a TEC)
Plus the Transformer and bridge rectifyer will get pretty hot, and there not as efficent as a swiching supply, but if your running a 220watt pelt or two i dont think electricity bill is a major concern.

Theres a link on todays front page with a guy who put a unregulated sullpy in a old PSU box for TEC use.
 
Holst said:
The bits for an unregulated supply will fit in a fairly small box, and it is much cheaper then a regulated job.
Difficulty is getting enough capacitance to remove any AC ripple (thats very bad for a TEC)
Plus the Transformer and bridge rectifyer will get pretty hot, and there not as efficent as a swiching supply, but if your running a 220watt pelt or two i dont think electricity bill is a major concern.

I think another difficult these days might be finding the transformer. I looked through a recent Digi-Key catalog and couldn't find anything beefy enough for this job.

You would definitely want a toroidal transformer, because the stray magnetic field would be horrendous otherwise.

I'd use schottkey diodes for the bridge rectifier which would get the heat and wasted power down some.

The caps would definitely be an issue. You want not only high capacitance but low ESR (Equivalent Series Resistance), because the load is only .633 Ohms (with a 266 Watt Pelt) you want to get ESR substantially below that, otherwise a huge amount of power will be dissipated in the caps.

The more I think about this, the more strongly I'd have to recommend people buy a switcher. Something like this should not be built by someone who doesn't understand ALL of the specs related to capacitors. It's too easy to build something that will work, but die an early death. The death of the big caps involved here can be pretty dramatic in itself, let alone the effect on the CPU when the pelt stops cooling.
 
Diggrr said:
Their 226 watt pelt MOVES 226 watts of heat.
It CONSUMES 364.8 watts of electricity to do this.
Do the math from their website's numbers....you'll need more psu than you are thinking!


So are you saying that a 510 watt PSU won't run a 226watt pelt.
 
Klownin79 said:



So are you saying that a 510 watt PSU won't run a 226watt pelt.

Idd say that running a 226watt TEC on a single computer PSU (even a high quality one) is likely to cause early death of said PSU.

It will work but I wouldnt expect it to last very long.

You would be much better off buying a bench type dedicated PSU.

And I think www.farnell.com is the best place for Transformers that I know of in the US.
 
Klownin79 said:

So are you saying that a 510 watt PSU won't run a 226watt pelt.

I don't wanna sound like I'm sending mixed signals here, so I'm gonna say a couple of things.
*I like ProCooling's article that EgNatz linked to, it's actually a pretty tough little psu for for the job. I've built one myself, and it works fine, though I've not used it long-term, and would have my doubts.

*A 500+ watt computer power supply would probably run a pelt fine for a while, but remember what that psu is meant for. There's no computer (retail desktop) that draws this much power, therefor you can take for granted that a computer's psu is NOT made to deliver it's rated load full time...in the real world, it probably never would see full draw, ever. Don't kid yourself into thinking the maker of the psu don't know this and isn't willing to use less robust components.

*A TEC requires that load current 24/7 every second it's on. Consider it like a burn in program for a psu that never stops. It does not ever let the psu rest like a computer does.

*You spent around $1,500 on the computer and all it's accessories, why would you not get a good quality power supply that is made to do the job. Your processor's life depends on it. It's like having toiletpaper seat belts in your car....saving a few bucks is no consolation to the person splatted on your dash.
 
Last edited:
Diggrr said:

*I like ProCooling's article that EgNatz linked to, it's actually a pretty tough little psu for for the job. I've built one myself, and it works fine, though I've not used it long-term, and would have my doubts.

*A 500+ watt computer power supply would probably run a pelt fine for a while, but remember what that psu is meant for. There's no computer (retail desktop) that draws this much power, therefor you can take for granted that a computer's psu is NOT made to deliver it's rated load full time...in the real world, it probably never would see full draw, ever. Don't kid yourself into thinking the maker of the psu don't know this and isn't willing to use less robust components.

I took a look at the ProCooling article and noticed a major problem is not addressed when the author discusses hooking PS's up in parallel. PC PSU's are not designed to be hooked up in parallel and load share.

Suppose you have two supplies of the same model, but one of them puts out 12.1V when loaded at it's max current and the other puts out 11.9V when loaded at it's max current. If you hook the two supplies in parallel and tie them to a load that draws 110% of the max current for a single supply, only one of the PS's is going to provide the vast majority of the current. The PS that sources 12.1V at max current may sag down to 12.0V at 110% load while the PS that sources 11.9V at max load sources no current at all, because it is detecting that its output voltage is at or above the voltage it is set to output.

What you have in this case is one supply running at 110% of it's maximum rated output and the other running in an unstable state due to a total lack of load current. This is just one of the nasty problems that can occur with supplies in parallel. There are much worse scenarios than this that are likely to occur if you load the parallel combination at more than 110% max rated current.

If you do want to hook up two supplies in parallel, first make sure that the 12V output is adjustable. (And, that you know how to adjust it.) Then with a load, equal to half the current output you are trying to achieve, adjust the power supply outputs to 12.00V as accurately as you can. This will help ensure that when the supplies are hooked in parallel to the full load they will each supply a substantial portion of the current. Because the values of electrical components drift over time, you should probably repeat this adjustment every 3 months to be on the safe side.

If you've done all this, you still haven't guaranteed safe and reliable operation of the paralleled supplies. You've only given them a fighting chance. If you don't understand all of this, don't try to hook supplies up in parallel. If you do understand this, and still want to hook supplies in parallel, and want to know how to do it in a way that truly is safe, let me know.

I agree with what Diggrr said about PC PSU's. They are designed with powering a normal PC in mind. The PSU manufacturers almost definitely cut corners where they can get away with it. (The profit margin on PSU's is razor thin.) Also, I looked up the spec's for the Antec 550W supply on the web. From an electrical engineer's standpoint, these spec's are seriously incomplete. There is no mention of temperature derating for example. They are not publishing the spec's an engineer would expect, when sourcing a general purpose supply for use in a new product. I personally consider a PC supply suitable for only one thing - powering a fairly normal PC.
 
Back