Another DIY Peltier - Adivce needed

[Exie]

Hi Folks,

I'm thinking of embarking on a little water-chiller project. I'm dreaming up something similar to the Swifttech MCW-Chill.

Here's my first question: If I get 2 x 226w pelts - Can I run them off a standalone 450w PC power supply @ 12v ?

I'm planning on fixing some custom copper heatsinks to the "hot" sides of the Pelts, just how Hot should I expect it to get ? eg. 50c , 60c , 80c hiigher ?!?

The design in my head isnt final just yet, but if the consensus is that 226w Pelts iis "too big" I might settle for 2 blocks of 80w pelts in a similar config to the MCW-Chill.

What do you think ?

[Brians256]

Originally Posted by Exie

Hi Folks,

I'm thinking of embarking on a little water-chiller project. I'm dreaming up something similar to the Swifttech MCW-Chill.

Here's my first question: If I get 2 x 226w pelts - Can I run them off a standalone 450w PC power supply @ 12v ?

I'm planning on fixing some custom copper heatsinks to the "hot" sides of the Pelts, just how Hot should I expect it to get ? eg. 50c , 60c , 80c hiigher ?!?

The design in my head isnt final just yet, but if the consensus is that 226w Pelts iis "too big" I might settle for 2 blocks of 80w pelts in a similar config to the MCW-Chill.

What do you think ?

No.

Detail:
Look at the amps supplied on the 12V line. Your power supply is actually several different power supplies all rolled into one package, and the rating (450W) is only a sum of the wattages available on all the different parts. You could run the two 226W pelts on a 450W 12VDC power supply (maybe), but not on a 450W power supply that provides 250W on the 12V line, 150W on the 5V line, 48W on the 3.3V line and 2W on the -5V line. Substitute rail for line if you like.

Also, 226W pelts don't consume 226W. They move (at max) 226W of heat. For example, one 226W peltier consumes 364.8 watts of electricity. This is of course moving 226W of heat while maintaining zero temperature differential. Max efficiency is different than max heat moving is different than best cooling. Setups will vary.

Get two PC power supplies (one for each pelt) and go for it. As for how hot the hotside gets... that's the wrong end of the horse to be looking at. If the pelts are moving 226W of heat, and the heatsinks are 0.1C/W and the ambient temperature is 26C, then the heatsinks will be 26 + (0.1 * 226) = 48.6C. Are fanless heatsinks going to be 0.1C/W ? Probably not. There's a reason people watercool peltiers.

Sounds fun, though. Keep telling us how things go.

[Exie]

WOW!

Thanks for the insights. I hear what your saying with the lines/rails in a stock PC powersupply. I just wanted to start off simple with cheap/common parts where ever possible.

I've just ordered 2 x 80W pelts (40mm x 40mm) which might consume about 130watts each, which hopefully my PSU will deliver. (I hate measuring heat in "W", I do fall for the trap of thinking in terms of electricity.)

I have milled out a prototype water block from some MDF I had lying around, and am pretty happy with how it looks. I'm now hunting down a block of copper I can mill out, and I'll see how it goes.

Can I ask a silly question, what does "0.1C/W" refer to ? heat absorbed by copper heatsink from the Pelt ? Some sort of themal efficiency ? <shrug>

At this stage, I have 2 Thermalright XP90C's which are old CPU heatsinks (with "heat pipes"), I was going to bolt on to the Peltier blocks. My equation might look like:
25c ambient + (0.1C/W * 80W Pelts) = 33c which should be no problem at all ?

I accpet that 80W pelts may not cool my water block & coolant by much, but I'd be happy with modest gains. Coolant temp currently sits at about 35c, if I could knock 5 or 10 degrees C off it, I'd be happy.

Thanks again for the insight.

[Brians256]

Originally Posted by Exie

Can I ask a silly question, what does "0.1C/W" refer to ? heat absorbed by copper heatsink from the Pelt ? Some sort of themal efficiency ? <shrug>

It the ratio of temperature difference per watt of heat moving through the heatsink. So, if you have 0.1C/W and you have 40W, 80W or 100W, you would have a hot side that is 4C, 8C or 10C hotter than the cold side. The cold side is the one exposed to air, water or other coolant. The C/W figure is constant for any ambient temperature and heat being moving through the heatsink (within the normal range that we work with). Look at all the waterblock reviews we publish.

Originally Posted by Exie

I accpet that 80W pelts may not cool my water block & coolant by much, but I'd be happy with modest gains. Coolant temp currently sits at about 35c, if I could knock 5 or 10 degrees C off it, I'd be happy.

I'm guessing that you'll be lucky to knock 1-2C off your coolant temp. But, I'd love to be proved wrong.

Why? Secondary thermal paths, and the fact that your peltiers won't actually be moving 160W of heat since the temperature differential won't be zero. Peltiers get a LOT more efficient when you make their heatsink better.

[Exie]

Just thought I'd give everyone an update on this project. I ended up revising my design a few times along the way.

Fortunately, I found a guy that had an end mill which enabled me to channel out a bigger water block. (see pictures).

I had some trouble sourcing the 80W pelts as the store closed down after I ordered them (fortunately they never charged me). So I picked up a 65W pelt, which was all I could get at the time. It's mounted up and after "warming up" I'm getting about a 3-4c drop in coolant temp.

It's running off the stock PC powersupply and draws about 50W of power. The hot side of the peltier just has a basic heatsink on it at the moment, and it gets upto about 58c.

So I'm looking for better heatsinks to make the peltier more efficient, I now see what your saying about a second water circuit to absorb the heat. I may eventually end up going this way, but I wanted to avoid getting too complex if possible.

I'll keep you updated on how things progress.

[blue68f100]

At looking at your endmilled parts, you are not quite setup properly. I have a endmill that I use for making AL molds. Have been working with Al and SS for over 25 yrs. I have had many machinest ask me how I get clean cuts on AL. For some reason this soft metal gives them problems. The biggest mistake is running too fast, feed and bit. Most use to working with CS.

Use a 2 flute endmill (NON Carbide, HSS), more flute generate more heat. Run a slow speed (600-800rpm), use WD40 for the cutting fluid. Limit your cuts to 0.100 or less for rough cuts, and 0.020 for final, less if using small dia bits. Always move the bed toward the bit cutting edge. If you go the wrong direction it will force chips between the bit and part. And keep your chips cleaned out while it is cutting. Running too fast will load up the bit, gets hot and start sticking to the bit. Running on slow speed will give you super clean cuts. If you see smoke from the WD40 running to fast.

New project are always fun.

[Exie]

haha! Thanks for the advice! wish I'd had that before I started. This was actually my first go with such a device, and I discovered everything you mentioned, if we went too fast it clagged up, if it got too hot (eg. no cutting fluid) it clagged up.

I agree it's not the best job in the world, but I kinda got there in the end, and it wasnt too bad for the first go.

I wasnt sure how to adjust the speed on the machine, the workshop I was at just said "Yeah, theres a machine out the back, go for your life!" ... the really good thing was they didnt charge me anything to use it, and said I'm welcome to stop by any time as long as I sweep up after myself.

Once I get the design sorted and working really good, I'll make a fresh one, and hopefully it will come out much cleaner.

[Marci]

Suggest registering at OCAU and having a read of this thread: http://forums.overclockers.com.au/sh...er+chill+water

LOTS of theory on the subject and practical demonstrations along the way...