Exploring Below Ambient Water Cooling

[4Ryan6]

In 2 hrs of idle run time, being surfing the net and such, it will drop the water temperature in the cooler which is now a full 10 gallons with no ice, approximately 1.0c.

My present water temperature in the cooler is 14.1c and dropping.

While gaming playing the original Crysis fully maxxed out, it held the water temperature constant at whatever the temperature was when the playing began, I played a full hour at the same temperature, (Now that's impressive!).

It actually is performing beyond my expectations the past 2 days has been one serious learning adventure.

At this time through many changes (All first plans were thrown out the window!), and trial, effort, and error, to discover the best way to utilize the peltiers capabilities, I have finally settled on the most optimum setup.

Pictures will be coming with full details of all my discoveries, including the present setup.

ok so under full load does the water start to warm up?

how efficient are the Peltiers? im just trying to soak this device up.

Edit: it looks like after a few minutes of googling that these devices have a low efficiency so that goes back to one of my original questions, 4ryan6 have you discovered any situation where the cpu is heating faster then it can cool?

I've answered these questions of yours if you would just read what's already been written.

FYI:

Efficiency covers way more than just the peltier efficiency, and since the peltier is only getting 12v and is spec'd for 15v, that has something to do with it's performance and efficiency right there.

Efficiency wise how well is the water block transferring the cold from the peltier to the water?

How does water flow rate effect overall efficiency of the entire system?

Since the peltier is operating below it's spec'd voltage, how does cooling fan air flow, affect the efficiency of the heat sink cooling the hot side?

How efficient is my cooler at keeping water cool?

How does my long runs of insulated tubing affect efficiency.

There are many factors affecting efficiency of this setup.

Efficiency wise, the peltier sucks!

But I'm not using it as it was designed to be used in the first place.

However, so far, it is doing what I need it to do!

 

[4Ryan6]

A lot of variables come into play with this peltier added to the loop, water flow rate affects the transfer of cold to the water, too much flow and the cold is not transferred to the water as well as some of the lower flow rates do, so there's quite a bit of fine tuning this particular thing.

Also clamping pressure is a direct variable, not enough and the heat and cold transfer is limited, too much pressure and you can damage the peltier?

Water levels in the cooler affect how fast the total volume can be cooled but too low of a water level cools too quick and doesn't give a constant, because the water temperature range is steadily dropping.

I really wish I had a 15v power supply to give the peltier it's spec'd load of voltage but for now 12v will have to do, I'm sharing things as I learn them and until I have solid information gathered I can only share what's being learned the hard way. Ry

 

[4Ryan6]

Have you insulated your board/gpus?

No insulation is necessary with this setup as long as I don't go below 8c water temperature in the cooler.

Condensation begins forming at 8C where I live, (It will be different for each living environment and area you live in), so I make sure my bottom threshold is 10c and condensation is never a problem.

That still has me running 13c below my ambient room temperature of 23c at 10c.

 

[4Ryan6]

Danger! Will Robinson!, Danger!

In my flow rate experiments with the peltier I dropped the flow rate too low, ice formed inside the water block enough to block off the water flow, my machine started crashing, I first thought my overclock had suddenly become unstable, which is unusual to just suddenly happen.

Thankfully I have the see through Danger Den block and saw the water was not flowing, the return line had completely iced at the peltier, I shut down the peltier and cranked the flow rate wide open and melted the ice inside the water block.

There are so many variables, when I increased the clamping pressure I increased the peltiers efficiency, which means it got colder faster and got colder period, than what was previously happening.

This type peltier setup is not something you turn your back on and just leave it running, if I had not modified the water block to allow more flow through it and left it with it's high restriction it would be freezing up on a regular basis.

Though a high flow rate does not seem to be the best cold transfer from the water block to the water, it is the safest flow rate to keep ice from forming inside the block.

Additionally as I have conducted various water levels in the cooler, I've come to the conclusion that the insulated standard reservoir idea I had earlier would not be a sufficient amount of water to act as a buffer, the lower the amount of water in the cooler the faster the peltier chills it and drops the entire water temperature in the reservoir or cooler.

I'm presently running approximately 7 gallons of water which seems to be a sweet spot for this setup? (still testing though not a solid statement), however too low a water level has the peltier dropping my water temperature faster than desired and I do not want it going below 10c.

At this point I would like to share that this peltier cooling is not a set it and forget it solution, it could cost you your entire investment, thankfully Intels thermal throttling saved my machine this time, whatever you guys do overclocking, never ever disable thermal throttling! Ry

 

[4Ryan6]

This may seem unusual to be in this water cooling thread but I have to give recognition to the TRUE > Thermalright Ultra Extreme 120 air cooler, in all the testing I have done, (Thermal Compound Roundup, this testing etc.), in direct comparison to it's performance vs the Noctua NH-D14, it has out performed the Noctua on many tests.

The TRUE may be classed as old school, but IMO from actual tests, it is still one of the best air coolers you can possibly get your hands on, especially with 100cfm fans in a push/pull setup, it has one of the best clamping assemblies out there.

It originally came onto the market as a passive air cooling solution, you don't market something as passive without the confidence it will perform, and this cooler does perform!

Knowing what I know now, if the choice came to choose either a Noctua NH-D14 or the TRUE, I would choose the TRUE, no questions, no doubt!

That's just shared FYI. Ry

 

[cbrunnem]

Danger! Will Robinson!, Danger!

In my flow rate experiments with the peltier I dropped the flow rate too low, ice formed inside the water block enough to block off the water flow, my machine started crashing, I first thought my overclock had suddenly become unstable, which is unusual to just suddenly happen.

Thankfully I have the see through Danger Den block and saw the water was not flowing, the return line had completely iced at the peltier, I shut down the peltier and cranked the flow rate wide open and melted the ice inside the water block.

There are so many variables, when I increased the clamping pressure I increased the peltiers efficiency, which means it got colder faster and got colder period, than what was previously happening.

This type peltier setup is not something you turn your back on and just leave it running, if I had not modified the water block to allow more flow through it and left it with it's high restriction it would be freezing up on a regular basis.

Though a high flow rate does not seem to be the best cold transfer from the water block to the water, it is the safest flow rate to keep ice from forming inside the block.

Additionally as I have conducted various water levels in the cooler, I've come to the conclusion that the insulated standard reservoir idea I had earlier would not be a sufficient amount of water to act as a buffer, the lower the amount of water in the cooler the faster the peltier chills it and drops the entire water temperature in the reservoir or cooler.

I'm presently running approximately 7 gallons of water which seems to be a sweet spot for this setup? (still testing though not a solid statement), however too low a water level has the peltier dropping my water temperature faster than desired and I do not want it going below 10c.

At this point I would like to share that this peltier cooling is not a set it and forget it solution, it could cost you your entire investment, thankfully Intels thermal throttling saved my machine this time, whatever you guys do overclocking, never ever disable thermal throttling! Ry

why not lower the flow rate like you say is ideal and add something to the water to lower its melting point? like alcohol, antifreeze or salt? you could run a lower flow rate like you said you would like and not form ice.

 

[4Ryan6]

why not lower the flow rate like you say is ideal and add something to the water to lower its melting point? like alcohol, antifreeze or salt? you could run a lower flow rate like you said you would like and not form ice.

That is an option, I'll keep that in mind.

Even on flow level 2 it is steady dropping the temperature in the cooler, I had it on flow level 1 when the incident occurred, when I crank it up to flow level 5 the maximum flow rate for the Swiftech MCP655, the temperature stops dropping and holds steady and while gaming at that flow rate the water temperature does very slowly increase.

As far as temperature holding during gaming it has to be on level 2 or below.

Some type of anti-freeze would allow to drop back down to level 1, but any type of anti-freeze also raises the temperature conductivity of the water.

When I increased the clamping pressure, it seriously increased the cooling of the peltier, it came with no instructions what so ever, and I discovered the clamping pressure from shared info on the net.

The thing is much more durable than it looks if I estimated from past experiences of using this Thermalright clamping assembly that it's clamped with approximately 80psi plus or minus 10~15 pressure, I'm definitely way past the 50psi that IC Diamond thermal compound requires.

It's an ongoing learning experience man, how's things going for you?

 

[4745454b]

So knowing what you know now....

Would a smaller TEC be a good idea? What about lowering the voltage it gets? Would that make it cool less?

I understand its a flow issue that causes the icing problem and these won't fix it. I'm thinking about the volume of water "problem".

 

[4Ryan6]

So knowing what you know now....

Would a smaller TEC be a good idea? What about lowering the voltage it gets? Would that make it cool less?

I understand its a flow issue that causes the icing problem and these won't fix it. I'm thinking about the volume of water "problem".

When I decided to do this testing FrozenCPU had and still has 2 different peltiers, an 80w and a 245w, spec wise I could only assume what the lower powered 80w would do.

I am already supplying lower voltage than the 245w requires at 12v, my P/S has plenty of amperage at 80a single 12v rail to handle the peltier, but I am 3v below it's recommended voltage, I can only imagine what this thing could do if it actually was receiving it's 15v requirement.

Specification wise the 80w model requires 16v?

I chose the 245w model thinking I needed that much to counter the CPUs wattage heat load?

Would a smaller TEC be a good idea?

I don't know the answer to that question, it would be more convenient when it comes to fitting today's available heat sink coolers, because most all are not over 40mm base width, I had to come up with a heat spreader to even use the 245w model it's dimensions are 50mm x 50mm, where the 80w is 40mm x x40mm.

Additionally if you cannot give the 80w the voltage it requires it's not going to give you what you want from it, a dedicated 15v 25a switching dedicated power supply is expensive, and getting into the too high price range to be messing with, the power requirements is something learned the hard way, and something anyone considering this needs to be aware of.

What about lowering the voltage it gets? Would that make it cool less?

Yes

I understand its a flow issue that causes the icing problem and these won't fix it. I'm thinking about the volume of water "problem".

I'm still testing man, I just have another major project going on that's limiting my testing time, I'm building a screened in porch on the rear of my home, and it is so flipping hot here in South Carolina right now, that most of my experimenting is taking place on the weekends, during my inside the house under the AC recovery time from the weeks work!

I'll continue sharing as I discover these things. Ry

 

[Cjsparky]

why not lower the flow rate like you say is ideal and add something to the water to lower its melting point? like alcohol, antifreeze or salt? you could run a lower flow rate like you said you would like and not form ice.

Agreed right up till you said salt...

 

[4745454b]

Yeah, just tossing ideas out because you wanted to lower the volume of water. I don't think I would have gone 80W either...

 

[cbrunnem]

That is an option, I'll keep that in mind.

Even on flow level 2 it is steady dropping the temperature in the cooler, I had it on flow level 1 when the incident occurred, when I crank it up to flow level 5 the maximum flow rate for the Swiftech MCP655, the temperature stops dropping and holds steady and while gaming at that flow rate the water temperature does very slowly increase.

As far as temperature holding during gaming it has to be on level 2 or below.

Some type of anti-freeze would allow to drop back down to level 1, but any type of anti-freeze also raises the temperature conductivity of the water.

When I increased the clamping pressure, it seriously increased the cooling of the peltier, it came with no instructions what so ever, and I discovered the clamping pressure from shared info on the net.

The thing is much more durable than it looks if I estimated from past experiences of using this Thermalright clamping assembly that it's clamped with approximately 80psi plus or minus 10~15 pressure, I'm definitely way past the 50psi that IC Diamond thermal compound requires.

It's an ongoing learning experience man, how's things going for you?

im contemplating it because i have an old spare 250w power supply and a nh-d14 laying around. along with the water cooling setup i have now which prolly wouldnt be enough rad but rads are cheap....ish. really it comes down to if it can be installed within the case at which point i wouldnt need a hsf just use my current water block with mods BUT condensation......

Agreed right up till you said salt...

yeah i was jus spit balling. idk how bad the pitting that is would cause would be in low concentrations. i think that a little bit of antifreeze would be the best route out of the three cause it has the second best thermal conductivity(brine has just as good as water).

 

[4Ryan6]

im contemplating it because i have an old spare 250w power supply and a nh-d14 laying around. along with the water cooling setup i have now which prolly wouldnt be enough rad but rads are cheap....ish. really it comes down to if it can be installed within the case at which point i wouldnt need a hsf just use my current water block with mods BUT condensation......

Power supply wise you have to be sure it has the 12v rail amperage load capability to run the peltier in the first place, not just any power supply will do the job.

The molex cabling from the power supply is not designed for the constant power draw of high amperage the peltier requires, molex cabling are only 18g wire, that discovery is covered above somewhere.

If you would even contemplate installing this inside your case, you have more guts than I do.

 

[Motopsychojdn]

And brine? really? like saltwater brine? that brine?
Hmm, I won't be putting one of the most corrosive liquids possible in my loop tyvm hehe
I wouldn't trust a 250w box to power much more than lights either, especially as Cbr mentions that its an old one,
Really can't wait to sort some fundage out and get back into the game here hehe, redesign time for the Rider I think,
Moto

 

[cbrunnem]

And brine? really? like saltwater brine? that brine?
Hmm, I won't be putting one of the most corrosive liquids possible in my loop tyvm hehe
I wouldn't trust a 250w box to power much more than lights either, especially as Cbr mentions that its an old one,
Really can't wait to sort some fundage out and get back into the game here hehe, redesign time for the Rider I think,
Moto

yeah i was really just a spit ball idea.

but what harm could an old psu do in this application. it goes out water temps goes above what ive told the computer is too hot and the computer shuts down.

 

[Motopsychojdn]

I see the sense there but if I'm applying brainspace to a mod or idea, I engineer in suitable components to avoid situations like the shutdown protocol you describe, I'd rather 'waste' a new 600w unit to power the Tec and KNOW it will hold out, rather than gamble on a lower rated possibly suss unit and need a backup plan for the inevitable inbound death of it
I'm ott in most things though hehe, You'll get used to me 🙂
Moto

 

[4Ryan6]

A single 12v rail is the best power supply option to use to make sure wherever you pull your power from will have the amperage available, and you won't be getting your power from the 18g molex cables either.

You will need to get power from 14g mains, like from the 8pin or 4pin motherboard connector, those are usually 14g.

I used 2 yellow positive 14g wires soldered together as the positive primary, and 2 black 14g negative wires soldered together as the negative secondary to power mine.

When it comes to powering a peltier you have to get your power from an area of the power supply that can handle the amperage draw which is a fixed not variable amount of power draw, or you'll burn up your power supply!

Learned that bit of info, just before learning it the hard way, more details are above in this thread somewhere.

 

[4Ryan6]

Yeahh..I live in FL..humidity is routinely 80% or better..I still run a bong 😀
I have AC..cuts the air moisture some.

I live in South Carolina we're in the 75% to 90% humidity range every day here this time of year, if not for the AC in my office my condensation would be much worse than it is.

What are the details of your Bong Cooler, do you have any pictures you can share?, if you do by all means share them?

I have an idea to increase bong cooling efficiency if you're interested, I almost went the bong route myself.

How do you exhaust the evaporation out the top of the bong?

What size pipe is your cooling tower section?

What pump are you using?

Is cleaning it difficult, and how often do you have to do it?

 

[4Ryan6]

Wow! 😱

You used a cooler for your base!

That's freaking Ingenious!

So you have the same advantages as I do, (adding ice plus the evaporative cooling to boot)

Impressive!

So with just the evaporative cooling alone what temperatures can you run steady below your ambient room temperature?

Got to go right now, but I'll check back later, it's nice to meet someone as crazy as I am! Ry

 

[4Ryan6]

6-11c below ambient..ambient is 29c
I searched for years trying to find a better way to cool a cpu..Finally one night I found it..ended up reading all night :lol:
I had to do it

Why is there a 6~11c variant, if ambient is constantly 29C?

I stick an Intel stock cooler in the wye to draw air through the water.

So you're drawing air through the water instead of pushing it, that's different!

Why did you choose to draw the air instead of pushing as most have set theirs up to do?

Did you try both ways?

From my reading on the evaporative cooling solution the coolest water is the water flowing down the inside wall of the tube, simply because it stays in contact with the airflow longer than the droplets falling through the air.

Have you done any temperature measurements to see if that is correct?

 

[4Ryan6]

Do you have any larger pictures of your setup, the ones you linked are pretty small?

 

[pnightshade]

Do you have any pics of how you have the TEC hooked up to the power supply and how you are regulating the voltage to it. I'm seriously considering making a W/C rig very similar to yours

 

[4Ryan6]

Do you have any pics of how you have the TEC hooked up to the power supply and how you are regulating the voltage to it. I'm seriously considering making a W/C rig very similar to yours

No but I'll try and take some later today of how I'm hooked to the power supply if I have the time, I have another major project going on.

I'm not regulating the voltage it's pulling 12v as soon as the power supply is energized, but you could say 12v itself is actually regulating the voltage it's a 15v peltier, so it's actually under powered.

Amperage wise it can have all it digests, the peltiers draw is 26a at 15v DC, the 1000w Silverstone is an 80a single 12v rail power supply.

 

[Cjsparky]

not really..I noticed there was like..stagnant air trapped in the cooler by the waterline..so I turned the fan around and drilled some 1" holes.The variation is probably due to water running over hot cpu..or humidity,or both.

Yeah I would think relative humidity on any given day would cause evaporation rate to fluctuate...but I don't really know anything about a bong...need to read about this from a Bongs for dummies point of view. Got a link I could go to so I don't have to Google "Bong" from work ;-p

CJ

 

[rubix_1011]

"Evaporation tower cooler" = 'bong cooler'