TEC build log start

[toolmaker_03]

http://imgur.com/cd15M1a
ok so I have finished all my testing and decided on a general build for this unit.
this build will be sealed inside of a chest so I will not have physical access to the computer case, I would like some input on the general idea that I am presenting, I use antifreeze so freezing the coolant is not a issue, and seeing as I am increasing the cooling capacity of my water cooling system, I have also chosen new clocks for my hardware, the CPU clock will run at 5Ghz, and the GPU's will be running at 930Mhz, I have already tested theses clocks and found them to be stable for my hardware.

 

[4Ryan6]

:lol: Cross posted on the tubing. :lol:

 

[toolmaker_03]

this thing is going to be another one of toolmakers ugly builds.
but it will be functional, and that is my main concern at this point, I can spend time and money on it later to clean it up, and try to make it look presentable.
at some point I will try to get all this into a case labs TH10 case, I may need a fully custom cube built from mountain mods, there asking for $5000 to make a cube case that can be vacuumed, because it is a first for them to do something like this, and they have no idea what it will actually take to complete the cube case.

 

[4Ryan6]

Yeah you haven't even reached your actual testing phase yet and it is going to be ugly starting out, you can refine it later after you discover what works as planned and what doesn't.

You have a lot of separate components that all have to work together in the end result, and hopefully it will go as planned but don't get discouraged if something doesn't, just learn and adapt and make it work.

Many times I have considered using a Mountain Mods and fitting all my setup into one case, but I always seem to find something more important to spend that money on.

 

[toolmaker_03]

yea, by over building the system it is my hope that I can get it running close to expectations, but with the way I am building it, if 3 X TEC's are not enough, I can always add 3 more TEC's, 6 X CPU blocks, 3 X air coolers, 3 X relays, 3 X switches.(I should have enough power supplies),
and I may need a few more radiators, but as I am stacking everything already, I now have the room to do this if needed.

 

[toolmaker_03]

as for my hopes for this system, I do not think that my expectations are to high, I have a 3930K CPU and I have seen them running at 5Ghz for up to 3 years now, if it is keep cool, most of them are phase change systems, but still I will try to keep the CPU under 50C at load at 5Ghz. my GPU's are the 580GTX and I would like to have them running at 930Mhz, I have seen this to running for two years, before they upgraded, my hope is to have them stay below 50C as well under load at 930Mhz.
at these clocks my system, it will bench mark in the same ball park as a 5930K system with 2 X 980GTX's at stock settings.
its not great, but its good enough for me.

 

[4Ryan6]

yea, by over building the system it is my hope that I can get it running close to expectations, but with the way I am building it, if 3 X TEC's are not enough, I can always add 3 more TEC's, 6 X CPU blocks, 3 X air coolers, 3 X relays, 3 X switches.(I should have enough power supplies),
and I may need a few more radiators, but as I am stacking everything already, I now have the room to do this if needed.

I think you're going to find out your 3 TECs are going to be quite sufficient to do the job, you are at least at some point going to have to justify the cost of even running this setup, you are going to be using quite a bit of electricity and even though this is like a super sized science experiment, are the ends going to justify the means, or are you going to end up with a shelf of parts and a nice story to tell your grand kids?

If you don't put limits on this, you will never get it completed.

 

[toolmaker_03]

I will also be doing a memory change to the system from the 1600Mhz memory to the 2133Mhz memory as this system at the new clocks will benefit from the greater bandwidth of the memory.
http://www.newegg.com/Product/Product.aspx?Item=N82E16820233659
http://www.newegg.com/Product/Product.aspx?Item=N82E16820231510

 

[toolmaker_03]

I think you're going to find out your 3 TECs are going to be quite sufficient to do the job, you are at least at some point going to have to justify the cost of even running this setup, you are going to be using quite a bit of electricity and even though this is like a super sized science experiment, are the ends going to justify the means, or are you going to end up with a shelf of parts and a nice story to tell your grand kids?

If you don't put limits on this, you will never get it completed.

[/quotemsg]

don't get me wrong I don't want to do any of that.
what I was trying to express, is that I am willing to do whatever I need to, so that the system it is up and running, while keeping the system cool.

 

[4Ryan6]

OK I got You!

 

[toolmaker_03]

what I noticed with the test build was this I hope I can explain this right.

ok the TEC's are capable of a 70C change in temp and with nothing on the cold side they get a 65C change to the water putting the water at -35C and the hot side water stayed at 30C
with a load on the cold side things changed drastically.
the cold side water held at 0C
and the hot side water held at 40C
this is my theory on why
1 I lost about 10C of coldness in the process of cooling the water.
2 I lost about another 10C of coldness because of the higher hot side water temps.
3 I lost about another 10C of coldness because of the heat from the CPU being dumped into the cold side of the water loop.
for a total of about 35C of coldness being lost.
and this is the reason that the water on the cold side would hold at 0C with a load on the system.
my hope is that I can keep this ratio consistent with the larger system.

 

[toolmaker_03]

so how did I get the CPU blocks sanded down and flat?

the final process that seems to work well is this.
I start with a belt sander, which has 130 grit belt on it, and hold the block as flat as I can to it, I can only sand this way for about 80 seconds at a time before the block gets hot. so it takes me about 8 hours to sand the block down with the belt sander.
then I use 320 grit sand paper on the flat stone and hand sand the block for another 8 hours until the block is actually flat.
then I use 800 grit sand paper on the flat stone and hand sand the block for another 4 hours until the block is ready to be polished.
this I use a polishing bit for a drill and some Brasso for about 30 minutes until the block is polished.
it takes awhile but I have 3 done now, and only 3 left to go.

 

[rubix_1011]

You could always setup the blocks to have water pumped through them to keep them cooler like they would be under load.

 

[toolmaker_03]

you mean while sanding them with the belt sander have a running water loop connected to it? while I do like the idea, I think it might be rather hard to control, as it takes all I have now to keep the block from traveling why I am sanding it, on that sander.
well, if I put both the sander and the block in a vice, clamped them to a work bench and used a adjustable work bench top to bring to two together, it might be doable.

 

[computers_are_freaky]

It could cause problems with the fittings though. I assume you would use soft tubing, because the vibration would probably risk damaging any hard tube, but soft might slip off the fitting with constant vibration.
While it would probably be hillarious for anyone watching, to see the tube suddenly disconnect and water go everywhere, I thing it would probably be a bad thing.

 

[4Ryan6]

You could always setup the blocks to have water pumped through them to keep them cooler like they would be under load.

That idea is so simple it is genius!

 

[4Ryan6]

It could cause problems with the fittings though. I assume you would use soft tubing, because the vibration would probably risk damaging any hard tube, but soft might slip off the fitting with constant vibration.
While it would probably be hillarious for anyone watching, to see the tube suddenly disconnect and water go everywhere, I thing it would probably be a bad thing.

All it would take is straight barbed fittings, cheap clear vinyl tubing, and stainless steel hose clamps, at the water block, and it would not disconnect.

 

[4Ryan6]

IMO the greatest danger to applying the fully constructed water block to a belt sander is it would literally cook the rubber ring seal between the block body and the copper base plate, but it may stay cool enough by running tap water through the block, that way the water running through the block would be constant cold.

I would definitely not create a loop so the water could be heated but just use straight from the house tap water and let the output side be waste water, or catch it in a barrel for watering plants.

 

[toolmaker_03]

yea, I'll stick with my 80 seconds at a time, it works, and I am not on the clock, so its not like I have a boss to please.
I do like the idea, but I will not be doing that.

 

[toolmaker_03]

4ryan6
When I tested water cooling the hot side I used a 240 Black Ice GTX and it was overloaded with a single 200W/20A peltier (Actual measured draw), and the radiator got so hot I could not hold my hand on it.

thinking back to my first TEC setup it was a direct die to a K7 CPU the system was made by swiftech and it came with a 254W 24V TEC a power supply for it and a insulation kit for the TEC.

I needed to get the insulation for the motherboard, and I should have got a temperature controller for it.

the CPU block water loop it was attached to, consisted of 2 X 240mm radiators, I did not have water temp sensors back then, but I can remember the radiators being very warm to the touch, if I was to guess I would say they where between 40C and 50C, and closer to 50C.

 

[toolmaker_03]

after reading through a few threads, I started thinking even though the DYI reservoirs are inside the ice chest it might be a good idea to insulate them anyway, to try and help eliminate some of the radiated loss of coldness that happens in this way. I will probably insulate the 3-way connecting tubing on top and bottom of the reservoirs, but not the section of tubing going from the bottom of the reservoirs to the bottom GPU block, or the section of tubing going from the bottom of the GPU block to the outside of the ice chest.

ok so is 6 X 360mm radiators enough for this build?

well here is how I look at it, two 360mm radiators where able to keep the hot side of one TEC with a load on the cold side at 40C.
I have three TEC's and I would like to keep my hot side at 40C or below, with a load on the cold side, so I am guessing that I will need six 360mm radiators to do that.

another way that I look at it is I have at least 900 watts of TEC heat that needs to be dissipated, as well at the clocks I plan to run this system at, there is another 900 watts of heat that needs to be dissipated, that's 1800 watts total. I have at least 1800 watts worth of water cooling potential, in the hopes that it is enough to dissipate the 1800 watts of heat that is being generated.

 

[rubix_1011]

What fans are you going to run on those? I'm thinking you won't want to run anything slower than 2000rpm and I would run them in Pull.

Also, did you get your temp controllers figured out for your TEC power draw? I know you were wanting to have them cycle off based on water temp, correct? Depending on Amps and Watts drawn, I'm just picturing in my mind how I am designing my electric brewing system which will be 5500w @240v and likely close to 40 Amps. A single controller could control all your TECs at once, but I'm thinking you might actually want to have a staged approach: multiple controllers with different thermocouples set at different thermal ranges. Obviously, you don't want them all tripping on and off all the time, but you'll want to know the threshold at which 1, 2 or 3 (or more) will want to trigger and turn on. Although, for my eHERMS controller, it cycled on and off every few seconds while it maintains exact temps, so your biggest concern might actually be cycled load with multiple TECs.

 

[toolmaker_03]

yea, the SP 120's can run at 2100RPM, and for now I will have a total of 6 80mm fans, they can run at 3400RPM, so I think it will work alright for now, I will upgrade the little fans later, for the SP 120's.
for now I will be using the manual switches, and I will upgrade the system later for automatic control, after I figure out what my idol and load temps will look like on this system.

 

[4Ryan6]

Sounds great!

So when is the official power up?

Looks like you've made way more progress than first anticipated?

 

[toolmaker_03]

well I am doing a lot of work, but I still have months worth of work left to do, Plus I still have a few things I still need to get.
for now, I have all the fittings to finish the TEC assembly, I will order the 2 X rads in 4 days, and I have already started building the frames for all of them.
next part is to make frames for the TEC power supplies, than I will start building all the stuff needed for the cold side of the system.

 

[toolmaker_03]

Radiator frames.
http://imgur.com/a/gRLKX