I'd bet those are really really low powered TEC's.
A quick way to tell, get a 12v battery (or any 12v source that can handle some current) and put your meter inline with one power lead at time, a good meter can handle 10 amps and I can't see that plug taking a full 20 (ten per line). Or you can use a amp clamp if you have one available....
I can't see them putting high power TEC's on that setup, especially with four of them! More than likely there are four 12704's or something along those lines, or basically around 10 amps total, for all four. (looks like they have two wired in parallel?)
Still, will be interesting to see what kind of performance you can get out of it!
I'd bet those are really really low powered TEC's.
A quick way to tell, get a 12v battery (or any 12v source that can handle some current) and put your meter inline with one power lead at time, a good meter can handle 10 amps and I can't see that plug taking a full 20 (ten per line). Or you can use a amp clamp if you have one available....
I can't see them putting high power TEC's on that setup, especially with four of them! More than likely there are four 12704's or something along those lines, or basically around 10 amps total, for all four. (looks like they have two wired in parallel?)
Still, will be interesting to see what kind of performance you can get out of it!
i'm going to need [strike]one point twenty one gigawatts[/strike] a kilowatt PSU.
The first thing that catches my attention is they used spring pressure to sustain the sandwich clamp, they should have just used bolts because you need about a minimum of 100psi clamping pressure and you'll never get anywhere close to that with springs.
Even clamping pressure around 100psi does make a huge difference in the peltiers thermal transfer in both directions hot and cold.
maybe 1000w is too much... the original had two 8800s in SLI and i'll be using one relatively low powered GPU. when i have more info i can see if 800 or even 750 would be enough. i have been considering a server power supply, with one rail delivering over 80 amps... overkill much?
the spring bolts/clamps can be replaced, but one at a time so i'm not disturbing the sandwich. what's 100psi (lb/ft?) on a metric torque wrench? and will i need high-tensile bolts?
maybe 1000w is too much... the original had two 8800s in SLI and i'll be using one relatively low powered GPU. when i have more info i can see if 800 or even 750 would be enough. i have been considering a server power supply, with one rail delivering over 80 amps... overkill much?
the spring bolts/clamps can be replaced, but one at a time so i'm not disturbing the sandwich. what's 100psi (lb/ft?) on a metric torque wrench? and will i need high-tensile bolts?
That original setup was designed to run it all from 1 power supply, do not scrimp on the power of the power supply more is better when it comes to powering peltiers, as my peltiers are on their own power supply but they are higher power peltiers than those in that configuration. (Because my setup cannot be run from just one power supply)
I would suggest a minimum of 1000w single 12v rail around 80a capability so you would only be using a percentage of the power supplies load handling capability.
Regarding the high tensile bolts that is not necessary we're talking PSI (Pounds Per Square Inch) not Ft/Lbs of torque.
There's really no way you can accurately measure the exact pressure produced by the 4 bolts on the flat contacting surfaces without using contact pressure paper and sending it off to a laboratory to be analyzed for actual mounting pressure. (that's just not necessary)
Regular #8 x 32 threaded screws is all I am using, and I do not know exactly how much pressure I have it clamped down to but it is way past spring pressure.
Most heat sinks clamp at about 25psi ~ 35psi pressure, some like the Thermalright TRUE could be additionally tightened to about 50psi, and most water blocks at around 35psi as well, and they are all under spring tension.
Using the screws directly tightened will increase the pressure, do it evenly and equally as you tighten them, just don't go to Gorilla tightening when doing it.
Caution: Gorilla meaning not to tighten the screws until they cannot be tightened anymore as you could damage the peltiers themselves, they will take more pressure than you may think they can, and their performance does improve with more pressure, but do not tighten until the screws cannot tighten anymore.
Peltiers usually like around 8 in/lbs or as high as about 15 in/lbs of mounting pressure. There is a calculation, I've done all the research on it and laid it out in this post: Peltier mounting procedures
Check the post just below that for the manufacturers mounting diagram.
Peltiers usually like around 8 in/lbs or as high as about 15 in/lbs of mounting pressure. There is a calculation, I've done all the research on it and laid it out in this post: Peltier mounting procedures
Check the post just below that for the manufacturers mounting diagram.
This is to address the cold pickup water block setup this thread actually covers and explains, it is not covering tech-wrecks supplied picture of the aluminum heat sink peltier sandwich that springs were used on, as he can safely add more pressure to that setup.
Looks like according to the manufacturers mounting diagram mine are mounted wrong however there is some missing information regarding my mounting setup in relation to the manufacturers mounting recommendations.
That missing information is, they were not mounting those peltiers to modified flow standard CPU water cooling blocks.
The water blocks in the mounting diagram have the inlet and outlet on the side of the water block, the metal is twice as thick as the copper base is of a standard water block so those type water blocks can take much more pressure than a modified flow water block can take.
Taking those manufacturers suggested mountings into consideration maybe that's why my own experiments with the block setup above were unsuccessful, because I could have mounted on that water block, exactly as they had suggested.
That monstrous copper block was not designed for cold pickup, it was designed to cool the hot side.
When the traditional CPU water block is flow modified you remove the center support structure column of the original design and that basically limits the raw pressure you can clamp down without warping the copper base plate of the modified flow water block.
So depending on the mounting pressure the peltier should be sized as close to the coverage of actual water pickup area of the water block vs the actual mounting possibilities, because too much pressure can actually warp a non modified water block, so you can definitely warp a modified one.
Many have actually warped water blocks that were not flow modified attempting to get 50psi of mounting pressure that a certain thermal compound company required for their guaranteed thermal compound performance. Many complaints this company received and at one time there was a linked thread right here at Toms Hardware of complaints against the company of users of their product complaining that their suggested pressure had warped their water blocks.
The company refused to take responsibility as that owner has a host of lawyers backing him, but it did force them to change their website from their suggested 50psi mounting pressure to showing pressure charting of various mounting pressures vs the gained increase in performance. That Toms thread by the way got so nasty it had to be dealt with. http://www.innovationcooling.com/key_to_performance.html
Note: If you desire more information regarding the spoiler simply Google it, there is a lot of information to glean from the spoiler regarding mounting pressure vs thermal conductivity. Also the company in the spoiler not only had to deal with warped water blocks, but deep scratches on the CPUs heat spreader causing in some cases a loss of CPU warranty coverage.
My estimated 100psi mounting pressure is exactly that estimated taking into consideration that I cannot just go Gorilla torquing on mountings to a modified flow CPU water block, sometimes it is possible to just have too much information!
I mounted as was conveniently possible and actually regarding mounting pressure torque some of the mounting pressures I saw suggested at various websites there's just no way the modified flow block copper base plate could take that kind of pressure.
So at some point you have to realize that manufacturers mounting suggestions do not fit every possible mounting a peltier could be put into, so what then?
Scrap the entire project?
Not the MaMa!
I'll be the first to admit that, #1 I am not mounting as the manufacturer recommends, and #2 I am not torqued down to any spec'd torque.
Which obviously means that if I were torqued to the proper spec's I possibly would get a little better thermal conduction both ways hot and cold, and also possibly warp the blocks base plate and not make full contact with the peltier and that would be bad.
I am glad you brought this up because this is something anyone duplicating this cooling needs to be aware of.
Thank You! Ry
Edit: I would like to add that one of the main reasons I decided to use a traditional CPU water block and modify it's flow through was because the thinner copper base thickness would be a faster cold transfer to the coolant flowing through the block.
Reasons for that is faster chilling of the coolant in the insulated reservoir and less copper mass contacting the peltier to retain and hold cold, it would logically seem you would want to retain and hold the cold, but from my testing when the mass cold is retained is stalls out the peltier.
You do not want to create stall points in the loop because if you do the system does not store enough useable amount of cold in the reservoir to use to cool the CPU, and when that happens the chilled system cannot hold a constant temperature under a CPU full stress load.
Every part of the cooling loop has to be contributing it's full thermal capabilities without stalling to be in constant cold coolant production.
That's why the modified flow water block works so well as it is constantly picking up what the peltier is delivering leaving the interior surface free of ice buildup yielding the needed below ambient cooling possibility.
The uncovered cold side of the peltier will ice over under supplied power, how fast that cold output translates to useable cold depends on the thickness of what is mounted to it.
The faster that raw cold from the peltier can get to the coolant, the faster cold can build up in the insulated reservoir, that is why the thinner copper water block base plate works better than the traditional thick ones do, they transfer the cold faster.
This type of chilled water cooling is new Territory and falls outside of any manufacturers mounting specifications or requirements.
Peltiers usually like around 8 in/lbs or as high as about 15 in/lbs of mounting pressure. There is a calculation, I've done all the research on it and laid it out in this post: Peltier mounting procedures
Check the post just below that for the manufacturers mounting diagram.
This is to address the cold pickup water block setup this thread actually covers and explains, it is not covering tech-wrecks supplied picture of the aluminum heat sink peltier sandwich that springs were used on, as he can safely add more pressure to that setup.
Looks like according to the manufacturers mounting diagram mine are mounted wrong however there is some missing information regarding my mounting setup in relation to the manufacturers mounting recommendations.
That missing information is, they were not mounting those peltiers to modified flow standard CPU water cooling blocks.
The water blocks in the mounting diagram have the inlet and outlet on the side of the water block, the metal is twice as thick as the copper base is of a standard water block so those type water blocks can take much more pressure than a modified flow water block can take.
Taking those manufacturers suggested mountings into consideration maybe that's why my own experiments with the block setup above were unsuccessful, because I could have mounted on that water block, exactly as they had suggested.
That monstrous copper block was not designed for cold pickup, it was designed to cool the hot side.
When the traditional CPU water block is flow modified you remove the center support structure column of the original design and that basically limits the raw pressure you can clamp down without warping the copper base plate of the modified flow water block.
So depending on the mounting pressure the peltier should be sized as close to the coverage of actual water pickup area of the water block vs the actual mounting possibilities, because too much pressure can actually warp a non modified water block, so you can definitely warp a modified one.
Many have actually warped water blocks that were not flow modified attempting to get 50psi of mounting pressure that a certain thermal compound company required for their guaranteed thermal compound performance. Many complaints this company received and at one time there was a linked thread right here at Toms Hardware of complaints against the company of users of their product complaining that their suggested pressure had warped their water blocks.
The company refused to take responsibility as that owner has a host of lawyers backing him, but it did force them to change their website from their suggested 50psi mounting pressure to showing pressure charting of various mounting pressures vs the gained increase in performance. That Toms thread by the way got so nasty it had to be dealt with. http://www.innovationcooling.com/key_to_performance.html
Note: If you desire more information regarding the spoiler simply Google it, there is a lot of information to glean from the spoiler regarding mounting pressure vs thermal conductivity. Also the company in the spoiler not only had to deal with warped water blocks, but deep scratches on the CPUs heat spreader causing in some cases a loss of CPU warranty coverage.
My estimated 100psi mounting pressure is exactly that estimated taking into consideration that I cannot just go Gorilla torquing on mountings to a modified flow CPU water block, sometimes it is possible to just have too much information!
I mounted as was conveniently possible and actually regarding mounting pressure torque some of the mounting pressures I saw suggested at various websites there's just no way the modified flow block copper base plate could take that kind of pressure.
So at some point you have to realize that manufacturers mounting suggestions do not fit every possible mounting a peltier could be put into, so what then?
Scrap the entire project?
Not the MaMa!
I'll be the first to admit that, #1 I am not mounting as the manufacturer recommends, and #2 I am not torqued down to any spec'd torque.
Which obviously means that if I were torqued to the proper spec's I possibly would get a little better thermal conduction both ways hot and cold, and also possibly warp the blocks base plate and not make full contact with the peltier and that would be bad.
I am glad you brought this up because this is something anyone duplicating this cooling needs to be aware of.
Thank You! Ry
Edit: I would like to add that one of the main reasons I decided to use a traditional CPU water block and modify it's flow through was because the thinner copper base thickness would be a faster cold transfer to the coolant flowing through the block.
Reasons for that is faster chilling of the coolant in the insulated reservoir and less copper mass contacting the peltier to retain and hold cold, it would logically seem you would want to retain and hold the cold, but from my testing when the mass cold is retained is stalls out the peltier.
You do not want to create stall points in the loop because if you do the system does not store enough useable amount of cold in the reservoir to use to cool the CPU, and when that happens the chilled system cannot hold a constant temperature under a CPU full stress load.
Every part of the cooling loop has to be contributing it's full thermal capabilities without stalling to be in constant cold coolant production.
That's why the modified flow water block works so well as it is constantly picking up what the peltier is delivering leaving the interior surface free of ice buildup yielding the needed below ambient cooling possibility.
The uncovered cold side of the peltier will ice over under supplied power, how fast that cold output translates to useable cold depends on the thickness of what is mounted to it.
The faster that raw cold from the peltier can get to the coolant, the faster cold can build up in the insulated reservoir, that is why the thinner copper water block base plate works better than the traditional thick ones do, they transfer the cold faster.
This type of chilled water cooling is new Territory and falls outside of any manufacturers mounting specifications or requirements.
You are absolutely correct! I think that most important part of that procedure to take away is the mounting pressure, too little and you lose cooling ability, too much and you can damage the peltier and or water block, and I've read that even if you don't damage it that it also loses efficiency if it is over torqued. No idea how much or if it's even true, I'll try to find that article again.
I'm not to worried about warping on my blocks due the way that they are clamped, there is no pulling pressure on the block itself, the peltier is sandwiched between two flat surfaces and then an external clamp was applied. Gah, I'm not worried. I'm not taking it apart and checking! Damnit.... Now I want a way to check my blocks....
I think that the clamping points as shown by the manufacturer are that way so that you don't put pressure on the fragile corners. But again, that's just a guess on my part. It worked out conveniently for me to use some AL to make a clamp that worked for me, the blocks I used have a different mounting pattern than the heat sinks so I was kind of stuck using that method!
I also ended up modding one of the MC-80 swiftech block I have. It's base flow rate is something like 0.7 GPM so I modified it so that it was moving 1.6 GPM, the same as the other two MC-60 blocks. I'm not sure that they are quite as good as your CPU block, it's not the thickness in this case, but the water channels that I think aren't quite as efficient. Once all three were milled and lapped, they aren't much thicker than a CPU block now.
I'll be getting the fans you recommended in tomorrow so we'll be able to see what kind of a difference they make with those TR coolers!
One of my Swiftech MCP655 pumps yesterday started making a noise like playing cards on bicycle spokes after being in service for 2 1/2 years, I drained the loop and inspected the pump, the ceramic bearing was fine virtually no visible wear, but the impeller showed signs of hitting the housing so it has to be the rider inside the impeller.
Serious Edit: After looking back over my purchasing information that pump that failed has been in operation over 4 years, it is amazing how fast time goes by! 2 1/2 years is how long this peltier cooling has been in operation and I was thinking along those lines but the Swiftech pump was in service long before that, so kudos to the Swiftech pump!
I always keep 2 pumps on hand for just such a dilemma so it wasn't something I wasn't prepared for just very inconvenient so yesterday was blown as it was dedicated to flushing out the system and opening and cleaning.
This pump was running the GPU cooling loop using the Mighty MO radiator, (Watercool MO-RA3), the loop consisted of 400ml Bits Power reservoir, Swiftech MCP655, Mighty MO, GTX Titan Hydro Copper, and back to the Reservoir.
My disgust came when I opened up my GTX Titan Hydro Copper graphics card to clean it!
You would think that a graphics card that cost $1300 bucks that EVGA would have their best assemblers making sure they were 100% right going out the factory door, but instead either my card was assembled by a bunch of monkeys being paid with bananas, or at 4:59PM when the employee was getting off at 5:00PM.
I have seen some sorry work under the hood of plenty of graphics cards but this was the worst I've ever seen, the GPU die was not fully covered with thermal compound, and it is a bare die not one covered with a heat spreader, and one of the corners had nothing on it at all!
I would estimate about 90% of thermal compound coverage over the die, and that's bad! Really Bad!
Many of the memory chips had almost zero contact with the thermal padding on the front of the card and the same for the VRS, and on the rear of the card the thermal padding did not even touch any of the memory chips.
The water block on the EVGA Hydro Copper is made by Swiftech and IMO it is the sorriest designed water block I have ever seen!
Compared to my past Heatkiller GPU water blocks, inside this Swiftech block looks like it was designed by the diaper rash kid!
Swiftech makes some good CPU water blocks but IMO if they cannot do any better than this, they should stop with making GPU water blocks all together!
First off it is chrome plated, the contacting plate is all one piece machined to fit the actual card face, but the flow side of the block is completely flat, there is zero water channel grooving on that side of the block, all the water channel grooving is in the acetyl top.
That means for the VR and Memory chips, they do not get advanced cooling from the metal coverage, but the thermal distance to coolant flow is the full thickness of the water block plate which is about 3/16th of an inch.
Plus the chrome plating is conveniently smooth and a step saving manufacturing process for Swiftech, but sucks for cooling performance when the coolant has to have time to pickup the heat to get it out of the water block.
Whatever you do, Do not buy a Swiftech GPU water block!
Knowing what I paid for the graphics card with water cooling and looking at the quality literally made me sick to my stomach!
I now know exactly why I was having problems overclocking it, #1 bad thermal compound and thermal padding installation.
#2 The way the water block is designed, there is way too much contact plate thickness over the VRs and Memory chips!
Anyhoo, I got it all back together last night, fixed everything wrong inside the card I could, new thermal compound and thermal padding, but there's just no fixing a piss poor designed GPU water block!
The pump failure was expected that's why I have spare pumps, but the crap I discovered with the graphics card was not, and it really wasn't even dirty inside, I'll have to hand that honor to PT Nuke BioCide the (Copper Sulfate), type. (It Works )
Aw man, that sucks about the cards condition. I'd agree with you, that should be assembled by their elite assemblers... the people who have a little badge for excellence on their chest or something. At that price it definitely should not be made by the new hire who's been working there for two days.....
Yesterday was the day for failures.... I was benching my video cards (so a heavy draw on the PCIe lanes) and suddenly my computer froze, a light blue almost teal color on my screens. I couldn't get it to do anything, so hard reset it. During post I noticed that the PCIe M.2 Raid drive said RAID OFFLINE. But it flashes so fast I missed the rest, then the screen flickered and went straight into BIOS..... ooooohhhhh crap.
So restarted, watched carefully and read the message.... one of the two M.2 drives on that card is D.E.D. dead. Son of a!!!!!!!!! So I pulled it, and last night started the process to load windows on my SSD drive. Le Sigh. Install from scratch again!
Learned my lesson, I kept putting off making a backup image. I'm making one tonight as soon as I have everything installed...... Got the OS and most of the drivers done last night, so just a bit more work to go. But then I have go back and install all the games / programs gah! What a pain!
I'll probably see if I can exchange the PCIe drive with the manufacturer for a couple SSD's instead. If not, I'll sell the damn thing after the RMA and get a couple SSD's that way. The technology might just not be mature enough for heavy usage yet, so I think I'm going to stear clear for another couple of years. I won't notice a speed difference with two or more SSD's in raid anyway.
Got my fans this evening. Got them installed, not a huge change in res temp. The heatsink temps did drop quite a bit you were 100% right there! They are no longer hot enough to cook on! I'd be that they are probably around 35-38C°.
The res temp has stabilized about 14C°. I just don't understand what is going wrong for me. These water blocks aren't that thick, I'm running around 600W of cooling with all my peltiers, three are 13v 200Qmax, running at 11.9, one is a 145Qmax at 15v and is running at 5v (using it as my 5v load). All lines are insulated, the Res is super insulated. Unless my 5820 is just that much higher of a TD than your CPU. I dunno, I'm grasping at straws at this point.
With ALL of them running I can get to about 14C.
Maybe it's the smaller surface area of these TEC's. They are focused more directly over the cooling plate on the coolers, but they don't cover the entire thing.
It's crazy the amount of heat this thing puts out into my office, lol. My thermostat for the house is directly across from the office and keeps going off
Now to determine if it's time to fish or cut bait.... I've got a TON invested in this project at this point, but without being able to drop to my dewpoint (which is 8C° tonight and as low as 2C° in my house during winter) I'm not 100% convinced that the heat exchange is worth it.
As it sits I can't up my over clock past what I was already achieving with just plain water cooling, I would stabilize around 18-20C° to get any higher. I don't really want to drop another couple hundred onto this project. For what I've spent so far I could have SLI'd a couple of 980ti's! Granted, I've learned a ton and had a good time doing it, so for sure no regrets.
I might go back this weekend and drain everything and hookup my old loop and see what the actual difference is. I also reseated my CPU cooler, which I had found the mounts had loosened up on the back of the board, which would have been raising my water cooling temps.
I can use those 110CFM fans on the top radiator and probably help quite a bit with stabilizing temps there, and possibly put the noctua fans in the front radiator and help out there also. Possibly replace my reservoir with a different one (or use my new found skills to make one) and hook up the swiftech pump for a better flow rate.
I'd bet that I can come pretty darn close to what I'm achieving with TEC's with just a water loop with all these fans I've got now! I mean I'm only achieving a res temp of 7-8C° below ambient. According to NZXT CAM my CPU is 16C° right now, so at an idle, no load, I'm only 5-6C° below ambient. If I sell off as much of this as I can, buy the new heatkiller iv, I would probably only be 3-4C° over ambient with straight WC, that's only a 8-9C° drop for this.
Don't get me wrong, the cooler is working! I'm just not achieving my goals with it. As I said in my thread, short of sending you one of my TEC's to throw into your setup to see if they just can't handle it, I'm not sure what else to do. To replace these TEC's will run upwards of $150, and even then I'm not guaranteed to get the performance I was hoping for, I would only guaranteed to add even more heat to the office...
I'll have to think on it this week, maybe throw my WC'ing loop back together with the new fans and test the temp difference before I make up my mind.
One other Toms forum member has duplicated the setup this thread covers using the exact same peltiers and the Thermalright coolers running 110cfm fans, he is also running the same sized and insulated reservoir, and modified flow water blocks, his results are almost identical to mine within 1c ~ 3c difference, even though he changed the routing of the loop.
His results are back on page 4 a little over half way down the page and his name is arthurh.
Oh, I'm not doubting that it works. Mine is able to cool to about 14C with the chip OC'd to 4.4Ghz @ 1.25vcore with the Cache OC'd as well, that's a TDP of close to 225-250 watts under full load.
If I were to turn my OC off and run this at stock speeds I bet I could get down to the 10C mark (which coincidentally with no OC on my chip would match the TDP of your chip over clocked).
I think part of it is the smaller surface area of the TEC's, and I think part of it is my higher wattage CPU. My stock wattage is as high as a 3770k OC'd to about 5Ghz. OC'd I am close to double your 5Ghz TDP!
In order to compensate for the higher TDP I think I would have to ditch my current TEC's, go with the TEC's you are running for the larger surface area, and run three of them all the time, two just won't cut it with this CPU.
I'm not debating whether the system works or not, it obviously does! What I'm debating (internally, not with anyone else...) is if the overall heat output and extra cost will be worth it? I think to reach anywhere close to 5Ghz I would have to run at least three of your TEC's with a fourth switched for load temps to be acceptable. This is just rough off the top of my head figuring based on what I've seen so far.
I ordered a GTX Titan Heatkiller water block to change out the piece of junk Swiftech on my Titan, I paid too much for that GPU which was classed as the Ferrari of GPUs when it was released and it surely did not deserve being cooled by a Jalopy POS water block!
Instead of a back plate to the card that really doesn't cool worth a crap either, I am leaving the back in the air flow and mounting RamSinks to the memory on the back side of the card, I will get the beast running cooler!
It's not king of the hill anymore but it plays all the games I play with maxxed out eye candy settings, and I do not plan on any future upgrading until Skylake shows us what it can do when the unlocked CPUs arrive.
I followed your lead on the reservoir. I think the loop total holds about 0.65 gallons, the res is just about 0.5 gal. I got three full flushes out of 2 gallons and still have 1/4 gallon left.
There is no 'data sheet' that I can pull up for these peltiers, they were a custom unit made by kryotherm in Russia for a company in Cali.
Qmax (W) = 196.0
Imax (A) = 23.1
Umax (V) = 13.7
dTmax = 69.0
R (Ohms) = 0.45
Dimensions = 35mm x 40mm x 3.1mm
I got these because they are close to the same wattage at 12v as yours, but with a smaller size that didn't require a plate on the hot side, the Heat Sinks cover 100% of the hot side, so a more efficient heat removal.
I followed your lead on the reservoir. I think the loop total holds about 0.65 gallons, the res is just about 0.5 gal. I got three full flushes out of 2 gallons and still have 1/4 gallon left.
There is no 'data sheet' that I can pull up for these peltiers, they were a custom unit made by kryotherm in Russia for a company in Cali.
Qmax (W) = 196.0
Imax (A) = 23.1
Umax (V) = 13.7
dTmax = 69.0
R (Ohms) = 0.45
Dimensions = 35mm x 40mm x 3.1mm
I got these because they are close to the same wattage at 12v as yours, but with a smaller size that didn't require a plate on the hot side, the Heat Sinks cover 100% of the hot side, so a more efficient heat removal.
OK 1 difference is the actual size mine are 50mm x 50mm = 2500mm total contact area vs your 35mm x 40mm = 1400mm total contact area, I have 1100mm more contact area per peltier than you do. (Just the total contact area alone is seriously affecting your cooling output.
Come on Ruwed, between you and I from what I've seen, you are the math expert, you can't see the difference in cold production the surface area alone is making, excluding the other differences
My Qmax is 243.5w @ 15.4v but at 12v is a full actual wall outlet load tested 200w.
Your Qmax is 196.0w @ 13.7v but what is it's actual load wattage at 12v have you measured it?
I'm guessing about 175w ~ 180w at 12v and when you use a voltage controller on that and force recovery time you don't even have that.
Ponder over this information, it's not that it cannot be done, but you are half stepping your way to it, and wasting money in the process, trying to save money.
Until you see the results from the right combination of peltier, air cooler, water block pickup, wattage, amperage, voltage etc. you will stay a little defeated, but once you see what the right combination can actually yield in cooling production, then you'll know 100% for sure and there will be no stopping you.
Unless you just give up and quit after all the time and effort you've already invested?
Heh, I'm no math expert! I just like to play sometimes, real math experts baffle me, I have no idea how people can grasp some of those abstract concepts!
There is no recovery time, I did away with the voltage controller setup and went straight 12v.
I had hoped that using a smaller surface area with very very close the the same output as your's, that I would be able to focus the cold to directly over the water channels (which we had discussed earlier, I think maybe back on page 8 or 9?) and that the heat would be more focused to the heat sink and be easier to remove.
Our systems are so close that they *should* be interchangeable.
What I may do is go in and completely unclock my CPU. Go all the way back to stock settings. If I am able to reach 10C° there and the third peltier has to switch off to prevent dropping below the dewpoint, then we will know that the only real difference is in the TDP of the CPU's and that I need an increased cooling capacity to handle the OC.
If that's the case I'll have to drop the project. I don't have room to add another cooler to the system, and I'm not sure I would be willing to put up with the heat even if I could! And to go in and change over to a smaller WC setup would be atrociously expensive.
It's not a matter of giving up, it's a matter of setting a limit to how much I'm willing to spend and realizing when the cost has far outweighed the gains that I can see. I can continue to dump cash into the project, but when is it too much? At a thousand? Two thousand? I'd bet that I'm probably pretty close to that thousand mark already, there comes a point where you just have to say "MY ASS IS ON FIRE FROM THE MONEY LEAVING MY WALLET SO FAST!!!!" and call it good
I'm willing to test with what I have, see if we can find the root cause of the temp difference (which I'm thinking is just the CPU difference) and what the cost would be to overcome it.
In either case, last night I dropped my ambient temp in my office to 22C° and saw my res temp drop to 13C° and my CPU was running 15-16C°. I mean come on, that's by no means a failure! That's a pretty substantial temperature difference! I think that at idle under WC I'm at 26C°, and load is similarly dropped by about 10C°, it's just that 10C° isn't enough to get me to 5Ghz. I'll try and play with settings, see where the saturation limit is and how high I can push the OC under this cooling setup, I still have lots of testing to do before I call it quits!
Plus like I said, I had a ton of fun with the project either way! It got me chatting with some very cool people, I learned a bit of programming, I can make custom fan controllers with touch screens, I spent HOUR's going back to the basics for electronics. I'm really enjoying the journey, which as trite as it sounds, is true.
That is not the peltier I am running, I did not buy any peltiers from them, each of my peltiers draws 200w tested from the wall socket that's actual wattage increase when the peltier is energized.
Running it at 12v instead of 15.4v drops 43.5w off the Qmax.
Also drops my amperage draw from 26a down to approximately 20a. (Calculated)
Maximum operating temp: 125 C
Imax: 26 Amps
Qmax: 243.5 Watts
Vmax: 15.4 Volts
Delta Tmax: >68 (C)
Size: 50mm X 50mm X 3.10mm
Ruwed :
There is no recovery time, I did away with the voltage controller setup and went straight 12v.
That's excellent and will avoid stalling out the peltier.
Ruwed :
I had hoped that using a smaller surface area with very very close the the same output as your's, that I would be able to focus the cold to directly over the water channels (which we had discussed earlier, I think maybe back on page 8 or 9?) and that the heat would be more focused to the heat sink and be easier to remove.
The smaller surface area cannot in no way output as much cold as a larger surface area can, surely you can see that?
Ruwed :
Our systems are so close that they *should* be interchangeable.
Only in your mind!
Not in raw output cooling performance!
How can you make that statement?
Look at the differences in the peltiers spec'd capabilities, the thermal cooling footprint size differences, modified flow water blocks, etc.
Ruwed :
What I may do is go in and completely unclock my CPU. Go all the way back to stock settings. If I am able to reach 10C° there and the third peltier has to switch off to prevent dropping below the dewpoint, then we will know that the only real difference is in the TDP of the CPU's and that I need an increased cooling capacity to handle the OC.
If that's the case I'll have to drop the project. I don't have room to add another cooler to the system, and I'm not sure I would be willing to put up with the heat even if I could! And to go in and change over to a smaller WC setup would be atrociously expensive.
It's not a matter of giving up, it's a matter of setting a limit to how much I'm willing to spend and realizing when the cost has far outweighed the gains that I can see. I can continue to dump cash into the project, but when is it too much? At a thousand? Two thousand? I'd bet that I'm probably pretty close to that thousand mark already, there comes a point where you just have to say "MY ASS IS ON FIRE FROM THE MONEY LEAVING MY WALLET SO FAST!!!!" and call it good
I'm willing to test with what I have, see if we can find the root cause of the temp difference (which I'm thinking is just the CPU difference) and what the cost would be to overcome it.
Well like ole Clint Eastwood said in Dirty Harry, "Every man has to know his limitations." jking
Ruwed :
In either case, last night I dropped my ambient temp in my office to 22C° and saw my res temp drop to 13C° and my CPU was running 15-16C°. I mean come on, that's by no means a failure! That's a pretty substantial temperature difference! I think that at idle under WC I'm at 26C°, and load is similarly dropped by about 10C°, it's just that 10C° isn't enough to get me to 5Ghz. I'll try and play with settings, see where the saturation limit is and how high I can push the OC under this cooling setup, I still have lots of testing to do before I call it quits!
Plus like I said, I had a ton of fun with the project either way! It got me chatting with some very cool people, I learned a bit of programming, I can make custom fan controllers with touch screens, I spent HOUR's going back to the basics for electronics. I'm really enjoying the journey, which as trite as it sounds, is true.
Your cooling is not a failure it is just not quite there yet, look what you have learned to get to this point, in the beginning you thought the CMH212EVO was going to handle the cooling load of the hot side and it did not.
I told you way back it wasn't going to handle it, but you had to try it out yourself and discover for yourself.
You thought higher CFM cooling fans would not make that much difference on the Thermalright cooler but you should know they do by now.
By the way they won't make much difference at all with the CMH212EVO as it is already paired with the best fan for it's cooling capabilities. (Been there, done that, tested it myself)
The unknown becomes known, and then you acted from that information to make improvements.
You are a learn the hard way individual, and I respect that as I am as well.
Edit: Do you have any pictures of the water block flow modifications you have done?
EDIT: How do I size pictures so I'm not taking up your entire thread with three pics? Sheesh, those suckers are HUGE!
4Ryan6 :
That is not the peltier I am running, I did not buy any peltiers from them, each of my peltiers draws 200w tested from the wall socket that's actual wattage increase when the peltier is energized.
Running it at 12v instead of 15.4v drops 43.5w off the Qmax.
Also drops my amperage draw from 26a down to approximately 20a. (Calculated)
Maximum operating temp: 125 C
Imax: 26 Amps
Qmax: 243.5 Watts
Vmax: 15.4 Volts
Delta Tmax: >68 (C)
Size: 50mm X 50mm X 3.10mm
Ah, ok, I pulled the pn you gave earlier in the thread and compared it that part number. The TEC pn's don't change through multiple manufacturers so I'd just used the specs listed from that page.
So math again so that we can have a straight comparison since I don't have a wall comparison
Using the same calculations I used last time, which is just dividing max Qmax by Voltage, it's not perfect, but gives us a representation at least:
Your cooling is 189 watts per peltier, you run two all the time - 378 watts of cooling all the time with a turbo mode of 567 watts of cooling.
I'm running three peltiers at 171 watts, plus a fourth smaller TEC at 5v which is adding 48 watts. = 561 watts of cooling.
4Ryan6 :
The smaller surface area cannot in no way output as much cold as a larger surface area can, surely you can see that?
It's outputting just as much cold as a larger TEC at the same wattage, the difference is the parts of my blocks that are exposed to ambient temps and could have a 'heating' effect due to cold loss to ambient air. I will add some insulation around (and over) the blocks, which should equalize the temp difference. Possibly the MCW-60 blocks aren't as efficient as your CPU blocks also. That's a pretty penny to get CPU blocks though.... Might be worth testing before I move on though. It IS one of the major differences. I could just start with testing two of them if I can find a couple used for cheap. See if it makes a temp difference.
4Ryan6 :
Ruwed :
Our systems are so close that they *should* be interchangeable.
Only in your mind!
Not in raw output cooling performance!
How can you make that statement?
Look at the differences in the peltiers spec'd capabilities, the thermal cooling footprint size differences, modified flow water blocks, etc.
See above math Our systems are within touching distance for raw cooling performance. I have the same coolers as you, the same fans, one Noctua and one EVO, the EVO sits at room temp as the TEC on it is only at 5v. The Noctua is actually cooler than both of the TR coolers. The difference is partly in insulation (I need to insulate my blocks) and mostly in CPU.
EDIT: there is also a potential difference in the MCW-60 blocks as I mentioned above.
4Ryan6 :
Well like ole Clint Eastwood said in Dirty Harry, "Every man has to know his limitations." jking
Lol, absolutely right! I'm still more than happy to do testing before I pull the chiller out of service. Maybe I'll find something that I missed in the build during the testing phase!
4Ryan6 :
Your cooling is not a failure it is just not quite there yet, look what you have learned to get to this point, in the beginning you thought the CMH212EVO was going to handle the cooling load of the hot side and it did not.
The EVO does handle the cooling of the smaller TEC's, which is what I'd bought it for Where I didn't listen is when you said the smaller TEC's wouldn't be enough.
4Ryan6 :
You thought higher CFM cooling fans would not make that much difference on the Thermalright cooler but you should know they do by now.
Never said that one though! I trust you man, you haven't given me bad advice yet! I may go through numbers and need to 'see' for myself, but you've given good advice all the way through!
4Ryan6 :
You are a learn the hard way individual, and I respect that as I am as well.
Yeah, sometimes it hurts being that way, but if it hurts it's usually remembered!
4Ryan6 :
Edit: Do you have any pictures of the water block flow modifications you have done?
The modifications were basically the same modifications that Swiftech did. This was the MCW80, now it's the MCW82 That's the only picture I have of the mod, I took it before it was done. The middle has also been clearanced by a couple of thousandths. The Copper block looks like a CPU block for this model.
Mod
Stock
And I am also using two of the MCW60's
The one on the right:
And the fourth 5v block is a chinese copper block that is similar to a TEC block, it has water just flowing through a open copper block with a channel that snakes a couple of times before exiting.
The more I think about it, the more all the differences are meaningless until I remove the OC from my CPU. There is no comparing the cooling capabilities based on the data we have currently.
By voltage and cooling wattage available, I am right there with you. The wider TEC won't make a difference if I can insulate the blocks to not be losing cold to ambient air.
The blocks... I'm probably losing a degree or two there I'd bet, there's just no way that they are as efficient as your blocks, but they should be pretty darn close.
But until my CPU is running close to the same TDP as yours, there's just no way to say that the current temp difference is due to any hardware related differences.
I'll remove the OC tonight and put my system back to factory specs and see where we end up there. It should be an enlightening experience. That will be the true test for the difference between our systems. If I am able to reach 10C° at an idle and the third TEC has to shut off AND I can maintain the 10C without the third TEC, we'll know that there isn't much of a difference between the chiller setups. If I am still stuck at the 13C°, then we'll know that your chiller is insanely more efficient than mine! (I'll still have the fourth TEC running, should come close to equalizing the TEC cooling difference, your 378 to my 390 and any gain is surely lost by the chinese water block.....).
So apparently my thermistor is a bit off in my Res.
So I un-clocked the CPU. Reached 10C in about 20 minutes, third TEC turned off, but I was seeing a discrepancy. My CPU was reporting 9C. Huh? Hmm, must be software not being accurate.
So I've let it run for about 20 more minutes at an idle, the temp crept back up to 11, then slowly rolled over to 12... but my CPU is still reporting 9-11C°.... hrmm.
So off the kitchen I ran, I cleaned a food thermometer in boiling water, took it back and dropped it in the Res. Res is now reading 12C°, thermometer reads 9.1! Ok, it just rolled to 9.2, it's been running without out the last TEC for about 25 minutes now.
I've got to go pick up the kid, so I'm going to let it run and see what happens, but I think that the water temp will stabilize just under 10C°, it's taking about 3 to 4 minutes to move 0.1°.
So it's possible that even with the OC on my CPU that I am managing to keep the Res temp at 10C°! This CPU just puts out so much freaking heat when OC'd that it runs 6-8C° over the res temp! But at an idle... with only a TDP of 120 watts, oh yeah baby. The cooler does exactly as advertised!
I'll be back to update in about 45 minutes. Wish me luck that the damn cooler doesn't blow up while I'm gone......
EDIT:
So after letting windows update do it's thing, finally got the system up to desktop. I'm not stable yet, still dropping, 9.9 now and going down about 0.1 every 3 minutes or so. (9.6c°)
So at roughly the same TDP as a 3700k @ 5Ghz, the cooler performs pretty close to yours Ryan. I'm pretty sure that your's is still more efficient at moving heat quickly than mine is, I know that your insulation on your tubing is better, and I'm losing cold around the exposed water blocks on the TEC's, but all in all, it's pretty close.
This tells me that it's just the extra heat my chip outputs when OC'd that is causing the difference.
I'll find out where the saturation point is for OCing this week and see where we end up.
Facebook
Twitter
Instagram