home made PSU for TEC? [Archive] - XtremeSystems Forums

View Full Version : home made PSU for TEC?

Shockerr

03-16-2004, 12:47 AM

I think Im going to get a 226watt Peltier (TEC) for my CPU in a few months, and will watercool it with my current setup, or maybe get a car heater core.
But, Im cheap, and dont want to get a dedicated PSU for the TEC.
So, I figure Ill make one myself; Ill need a capacator, diode, and transformer, capable of at least 250watts, maybe up to ~500 if I decide to TEC cool my vid card with another TEC.

I havent done the math yet, since I have a project and test tomorrow, but, Ill figure out exactly what ratings of components Ill need. But for now, how does my idea look (schematic below). Also, where can I buy these parts? (I may be able to from my University, but, havent checked yet).

Basically, Ill put it together on a nonconducting surface (acrylic maybe), and box it in once its finished. Ill need to soder the parts together, obviously.
Im not going for perfect consistent voltage, Ive heard TEC's can take varying amount pretty well. Id be willing to guess with a basic setup I could get it to a maximum variation of 0.5v. Does anyone know if that would be OK, since im not 100% sure on that
Id have the AC go to a transformer, probably 9 to 1, giving about 13.3volts AC. Then, to a diode, then a capacator and TEC hooked up in parallel. then back to the transformer.
Since the resistance is so low (power demand = high), the capacator will need to be fairly beefy. All the components will need to be I guess...I dont want to blow anything apart.

Anyway, comments, ideas, suggestions?

Ragnarok

03-16-2004, 03:59 AM

why only a 1/2 wave rectifier? use a bridge... for full wave rectification..

u gonna need a big :banana: :banana: :banana: :banana:ing cap if u use a 1/2 wave...

http://ourworld.compuserve.com/homepages/g_knott/elect207.htm

Vlad Draculea

03-16-2004, 09:18 AM

FUGGER
Xtreme Owner

Registered: Jun 2002
Posts: 2780
barfing
I removed the crap you posted.

__________________
Intel P4 3.4EE ES
Asus P4C800 Deluxe
OCZ PC4200 EL
ATI Radeon 9800XT

2k3 9915

2k1 30440 DX9

AquaMark3 68781

05-15-2004 11:31 PM

cwatson

03-16-2004, 06:14 PM

sweet! sell me one!

Vlad Draculea

03-16-2004, 07:23 PM

Shockerr

03-16-2004, 07:43 PM

Ragnarok: That seems like a good idea with the 4 diodes, but, AC isnt perfect, and works on a sin wave, so, the DC voltage would vary a whole lot, so, wouldnt I need to smooth the power before sending it to the load (TEC)? I guess id need a capacator, but, it would have to supply much less power, so, it could be smaller.

Vlad: I would be interested to see that spec sheet, if its not a problem. What components would determine the power; load obviously, but what else? Capacator and ?
For a 226Watt TEC from danger den, id want about 14v, dictating about 16A to give 226Watts.

STEvil

03-16-2004, 08:42 PM

Ooh, i'll be watching this like a hawk.. i've got a whole bunch of 200v caps from old PSU's and other items (forget their farad ratings at the moment) and i've been itching to build something to run TECs/car amps off of ;)

Ragnarok

03-17-2004, 01:20 AM

Of course u need a smoothing capacitor, using a bridge is just a lot more efficient than a single diode...

on the other hand, just grab a Mean Well PSU, they are pretty cheap if u look in the right places

U can get 250W MeanWell psus for like 90 AUD.. ~60 USD..

enzoR

03-17-2004, 05:42 AM

A smoothing capacitor?

I would use no less than 5,000mF per 1V

Since your draining such high amperage, i would suggest about 10,000mF per V for a decent quality power.

and use devices rated for alteast twice the power your gonna put through them... for Capacitors, use around 30 Volt ones, and for the bridge i would seriously recomend you to run 2 or 20 amp bridges. they will have to be heatsinked. if you wanted 1 40 amp bridge you would have to really cool it down with some kind of a CPU cooler. Good luck

Vlad Draculea

03-17-2004, 05:58 AM

enzoR

03-17-2004, 07:48 AM

yea but remember that this is after the transformer. A spike will toast those caps easyly... thats what i heared.
yea i guess around 10000uF would be fine for him.

as far as capacitance,
i use 2x 1 Farad caps. i know quite overkill, but i also use it as just a bench powersupply to power crazy contraptions that i make.

Shockerr

03-17-2004, 02:56 PM

Thanks for the suggestions thus far, Im going to talk to my physics lab prof tonight after my test.
So, AC is at 60Hz, with full wave rectification, the peaks will be 120Hz (come every .0083 secbonds, or 8300 microseconds).
So, I need the RC time constant to be mush longer then that, so, say, the voltage drop isnt more then 5%. So, the RC time constant will need to be 20 times the time for each cycle, or, .166 seconds (166000microseconds).
Id have to test the peltier, but at the max rated voltage for the 226W one from danger den, it works out to be about 1ohm (at 12v its 1.5ohm). Im gonna figure the rest of this out with 1 ohm.
so, I need RC to equal 166000microseconds:
166000 microseconds = (1)(C)
C=166000micro farads, or .16 Farad.
Hot damn, thats gonna be a big ass Capacator!

But, the circuit is always supplying power, just not the peak. Since its a sin wav, I can find the area under the curve, subtracet from the total area will give the power needed by the capacator....
area under curve between full voltage is ~2, when full area is pi.
So, the capacitor needs only 36% of the orifinally calculated power, or, .06Farad (60000 microFarad).
Does this make sense?

I guess I could also use an IC regulator to control the voltage; just feed the loop before it with more current and a smaller capacitor, so the DC current before the IC will be more than what I want and vary more, but the DC after the IC will be clean and level.
I do have many heatsinks lying around that I can use, btw...lots from old busted PSU's, actually :p

enzoR

03-17-2004, 03:08 PM

yup 60000mF is about right but you can easily get away with half of that

Shockerr

03-17-2004, 03:30 PM

That would mean the voltage would vary by about 10%, ie, from 15V to a low of 13.5V, with 60000F it would only go as low as 14.25, which should give better cooling performance or be easier on the TEC (exactly how sensitive are they to fluctuating voltage anyway?)

Vlad Draculea

03-17-2004, 03:55 PM

STEvil

03-18-2004, 01:18 PM

A pair of 1F caps?? Jeeze.. the mig welder and plasma cutter ($1500 for a new control board for the plasma cutter..!) I fixed recently only have one .5F cap each..

EDIT

the 200v caps I have are from 470uf to 2000uf

L337 M33P

03-18-2004, 01:59 PM

DIY 13.8V 40A Switching PSU (http://www.qsl.net/xq2fod/Electron/PS40/PS40.html)

For those of you that want a fair heft of current :D

enzoR

03-18-2004, 02:50 PM

Originally posted by STEvil
A pair of 1F caps?? Jeeze.. the mig welder and plasma cutter ($1500 for a new control board for the plasma cutter..!) I fixed recently only have one .5F cap each..

EDIT

the 200v caps I have are from 470uf to 2000uf

yea, were only 40 euros, couldnt resist :D

Boyne7

03-18-2004, 04:12 PM

hmmm, if this works i may have to build myself one also

Shockerr

03-19-2004, 12:04 AM

Hmmm, thats a very complicated power supply.
Finding a 60000microFarad ~20V cap is pretty much impossible, but I found soem that were about 20000, so, I could wire a few of those up in parallel to get what I need.
My physics prof and lab guy were useless; Im heading over to the electronics building this afternoon (about 12 hours from now), so,I hope to get some good advice there.
ill also have time to come up with some possible designs this weekend (this week is packed with tests, etc), and figure them out (applying class work to the real world, finally)

enzoR

03-19-2004, 06:54 AM

you can use 6 10k mF caps, it doesnt matter. and if i were you, id use 30v caps. ive had enough explode on me, and yes, i was doing what u are about to do. 20v caps, @ 15 volt. a small power surge, and your over their rated.