TEC ideas for a forum produced product

[CNYC]

I've just spent the better part of an hour reading a thread involving Graystar and the discussion about Flowrates and effectiveness of cooling depending on flow. I must say there is a great degree of engineering intelligence on this board. I definately wont try and pretend that I understood the technical majority of the thread.

I do however have a degree of trust in the technical prowess of the people in this forum as a result.

While I am not a technical wiz, when it comes to thermal management, I do consider myself a business wiz. I came to this board asking a few questions about TEC cooling since I was trying to build something to cool my CPU without using a radiator.

I elected to not use a Radiator since a Radiator can only cool temperature down to ambient level and I was looking to cool the water somewhere 10-20 degrees F, lower than ambient. Also size and the need to have extra cable running around the computer was taken into consideration.

Here is the business angle of it all,

This all started as I looked at products that were commercially available to cool drinks at $30 and lower (manufactured cost) it was amazing that these mini refrigerators were available at such a low cost and small form factor. In some instances these products were of very high manufactured quality.

I took a look at gallery pictures of overclocked watercooled computers and of pre-made kits and hand picked kits and I saw the same thing, cables hanging everywhere and generally things looking like they took an awfully large amount of effort to put together. I believe the look of watercooled computers has turned off alot of the mainstream users whom would adopt watercooling if it was made easier and more cost effective.

Well to make a long story short my goal was to design a device that beat the kits and all in one units using radiators and at a sub $50 manufactured cost.

So I chose to use TEC, since this technology (using an 80W tec) can be purchased less than $5 each, a heatsink/fan can be purchased less than $7, a decent pump can even be engineered for the specfic purpose of CPU watercooling and likely mass manufactured less than $10 each.

I've seen directly mounted TEC onto the CPU ala swiftech and it doesnt seem to be very efficent (at this point it has a bad marketing rap) and when it works it emits too much heat back into the computer case and it prone to failure.

I'd like to create a TEC water cooled entirely external unit with a small form factor. Since finding this forum I've found that this group seems to be the right group of people that I should work with.

I'd like to have the product be designed by the forum and I will bankroll the project every step of the way into what will be a product designed by everyone here. Of course from a business I will have every person involved compensated one way or another and will have to figure that out if and when a product looks like it can be produced.

Andrew

[bigben2k]

Andrew,


From a technical perspective, if you're looking to eliminate the rad, and use pelts only (because it's possible to watercool a pelt, but it involves using a rad), then you need to understand that the design is not energy efficient.

From a business perspective, your electric bill at the end of the month will surprise you.

If you want to pursue this, I'll help you, but I need to know that you understand the limits and inconveniences/disadvantages here.

[RoboTech]

Hi Andrew,

I commend your enthusiasm but there are many, many technical and economic barriers standing in your way. I'm no expert but I have built a couple external, self-contained, water-cooled TEC coolers before...

As Ben said, TECs are not very efficient - they use about three times as much energy as they can pump. With todays overclocked CPU heat dissipation approaching 100 watts, you need a LOT of cooling power (large capacity Pelts), dedicated power supplies to run it and a large, well designed water-cooling system to take the heat away.

It is more efficient to apply a TEC directly (via cold plate) onto the CPU and water-cool the hotside (already done by many) but this also gets expensive and has a number of disadvantages.

Slightly different approach, but you might be interested in taking a look at Thermaltake's SubZeroG4 TEC system. It is NOT designed for sub ambient temps but instead uses an air-cooled TEC as a "quiet cooling soultion".

Good luck with your project,

[CNYC]

BigBen/RoboTech,

Thanks for the advice, I really believe that watercooling is in its infancy stages for mainstream appeal. I think mounting a TEC directly onto the CPU (Coldplate) will cause more condensation than using a TEC to chill the water just a few degrees below ambient.

I do know that it is energy inefficent to create such a device, but in the grand scheme of it all the device really doesnt add more than a few cents a day. I'll make sure to measure the actual draw and then do a cost comparision of my electric bill with and without before trying to commercialize such a product.

I appreciate the links and support. I am definately no engineer but rather a person who sees an opportunity to help a technology segment grow beyond its grass roots into a product group that would be used by major OEM's or at the very least by mainstream enthusiasts.

If anyone has any design ideas to share for a self contained plug and cool device that could be made higher performing and cheaper than a Exos Koolance let me know

[redleader]

I am definately no engineer

Clearly, so I'll start off with a few engineering questions to get you thinking. Given this:

I elected to not use a Radiator since a Radiator can only cool temperature down to ambient level and I was looking to cool the water somewhere 10-20 degrees F, lower than ambient. Also size and the need to have extra cable running around the computer was taken into consideration.

Where does the heat go? You'll have several hundred watts of it to deal with and its not clear from your post where you intend to send it.

Second question.

I think mounting a TEC directly onto the CPU (Coldplate) will cause more condensation than using a TEC to chill the water just a few degrees below ambient.

What is the purpose of this? You've elected to go with a waterchiller because you want to reduce the performance of the TEC to avoid condensation. Would it not be cheaper to just use a smaller TEC on the CPU to get the same effect? If you're just cooling the water a few C below ambient, why use a TEC at all? You'd have a cheaper equally well performing kit using plain watercooling provided you invested the TEC cost back into the block.

[wymjym]

CNYC,
My chiller is about as green as it gets.

The controllers keep the power usage low, the heat low, the condensation at bay, the cpu cooler than a 'plain' water cooled system.
BUT....most people don't think it is extreme enough to warrant the cost, effort, yadyadyada.
I like it, will continue to use it and an upcoming improved version.
good luck in your search.
wj
here are some pix

[qwerty57]

yeah put it this way a pair off meanwell 320w psu's runing my system in my room from different outlets with the rest of my stuff clipped the circuit braker of 20a .. pelt's use mad AMPRAGE and amprage cost $$$$$... I thing condenstaion proofing a tec/h20 block will give you 25%+ better performance and cost a lot less..

[JimS]

Not sure what others pay for electricity, but in Northeast USA it is about 7 cents/KWH. I have run a system similar to wymjym's for the past year. Photos of my chiller blocks can be seen in the for sale forum. My electric bill never increased by more than $6 in one month, and it averaged about $2-3 more than when I had no pelts running.

There seems to be a lot of opinion about the high cost of running pelts, I simply disagree. Run a phase change setup and you will probably use as much or more electricity than a moderate peltier based cooling setup.

[bigben2k]

A rough calc:

At 7 cents per kilowatt/hour, assuming that a PSU draws 320 Watts, comes out to 2.24 cents per hour.

6$ would cover ~267 hours, or 11 days, in a month.


If I recalculate with the proposed 3 * 80 W = 240 Watts (assuming a 100% efficient PSU), I get:
hourly cost: 1.68 cents
6$ coverage: 357 hours, or 14 days.


Maybe it's not so bad... I'll check my light bill, and comment.

[sevisehda]

Instead of using pelts why not mod an airconditioner or dehumidifier. Phase change cooling is a far more effocent means of moving heat than a peltier. If I remember correctly my room AC is rated at 8000BTUs or 2344W and only uses about 800W of electricity. So its roughly 3:1. Whereas a Peltier is 1:1.

[JimS]

bigben, while I agree with your calculations, you are assuming 24 hours a day PC useage. Most people don't even approach 12 hours a day so you could probably cut your costs at least in half, probably by 2/3. Your calculations show just about what my system cost me in months of high use, about $6. During average or lite use, it doesn't cost me much more than running a 100 watt light bulb continuously for a month.

Obviously when using a pelt cooled system, you would tend to shut it off when it is not in use. For those who have to have their system on 24/7, pelt or even phase change cooling is not cost effective.

If my memory is right, a refrigerator compressor uses about 1,000W per hour of use, or 1 kwh. I know that the phase change cooling systems use similar models, or maybe a bit smaller. If this unit were on 8 hours a day, it would cost $.56/day, or $15/month to run the PC for an average of 8 hours per day. About triple what it would cost to run 300W worth of pelts.

The phase change will of course provide a much higher degree of cooling, but it will also cost more.

[sevisehda]

A refridgerator compressor may use 100W per hour but a small room AC unit doesn't. The room AC unit I have is of medium strength and only uses 800W Max. There are smaller units available and they would never be run at max capacity.

My computer(s) are always run 24/7/365. The only time there off is when there not working or when I'm out of town for more than a day. It was my understanding that most people around here did the same thing. So anytime I would try to figure power usage I'd go with constant use.

If you used an old Room AC unit or small refridgerator to cool the water you would get 900W of cooling power for the same energy use of 300W of pelt. Also keep in mind that most AC units throttle themselves so its doubtful you'd even reach there maximum rated draw.

Its my understanding your trieing to make something that can go a few degrees below ambient. If you need it selfcontained then go with pelts. If you want low energy use go with a bong, if you want reliability go with phasechange to chill the coolant.

[bigben2k]

...and keep in mind that if you overclock it to the max, you might have to give the coolant some time to cool down, before the PC is turned on, otherwise it'll simply crash. That's the reason for running an extreme cooling solution 24/7.

For the rest of us non-extremists, we have a cooling rig that'll preserve the life of a PC beyond the 10 years it was designed, minus a reasonable overclock.