Overclock.net › Forums › Cooling › Specialized Cooling › Peltiers / TEC › Things I have learned about TEC’ing over the past year and why I took a vacation...
New Posts  All Forums:Forum Nav:

Things I have learned about TEC’ing over the past year and why I took a vacation...

post #1 of 10
Thread Starter 
Apparently despite numerous attempts by several forum members to try and stress the fact that the only real way to determine what a given TEC setup will do is to actually run it people still feel there is some magical logical linear formula to express this mathematically. Well I am here to tell you to get your head out of your bum.

There is no magical TEC calculator. There is no magical formula. What’s more even the charts are a rough estimate at best.

That being said it is becoming more and more common for posts to be made to buy product X. All I have to say to that is let the buyer beware. Do not be foolish and decide simply because someone provided you with mathematical formulas, renderings, and/or simulations that what should happen actually will. If you purchase something and it does not perform the way you think it should be aware it is your problem. There are so many different variables involved that speculating on performance with no actual testing data to support it in the end you will end up chasing yourself in circles because you will probably be told you are doing something wrong and as a result of your foolishness have no real way to refute it. Doesn’t make much sense does it.

Personally, I wouldn’t buy a car that was rated 5 stars for crash testing if it was never actually crash tested in the first place. This is however exactly the case when using TEC’s. Life is like a box of chocolates you never know what you are going to get. If a said TEC item was pretested it would be a different story.

People tried to show the forum how a given TEC setup would perform in a real world live testing scenario and well so much has been deleted at this point I am not going to elaborate on it anymore. Suffice it to say this practice was attacked and eventually smothered out to the extent that in the end it really wasn’t worth the amount of time it takes to actually get a TEC setup assembled and then tuned properly. It takes at least 8 hours of testing if not more to properly “tune†a TEC setup. The old 67.77% max volts is just a barometer. You can see as much as a 4c swing simply adjusting a given voltage by as little as .5 volts. So in this case if Vmax is 31.5 21.3 volts is the starting point. From there you’ll then need to adjust in .1 volt increments and test the setup on a processor for at least an hour per .1 volt increase/decrease. So you see just given the latter it would take you 7 hours just to get to 22 volts. Ideally you would have thermocouples hooked up to the hot side, coldside, water (resevior), and ambient recording temperatures accordingly. CPU core temps are a mixed bag depending on software, the temperature sensors on the CPU itself, and the motherboard in question. Best bet is yet another thermocouple mounted to a groove (yes you would have to make one in the HIS) on the CPU’s IHS (integrated heat spreader). Thermocouples and meters are not all that cheap when you need them to go down to around -40c or so which is why no one is currently doing all of that. Instead we theorize what we think should happen. I at one point was willing to go to this extent but to be honest the constant bickering of theory versus actual results as if the people actually testing things were somehow daft just got way too tiresome followed by the outright deletion of a certain members complete existence which really was the final straw for me. At the same time I can’t help but look at the test bed now and again and want to try something new it is simply that sharing the results will result in nothing but a half arsed argument.

People can buy whatever TEC’s they want but I have yet to see a Chinese manufactured TEC tested on any modern up to date system that out performed an American TEC set up. You should also note that Chinese TEC’s are cheap for a reason but are in no way harder to install than any other TEC. So if a TEC from china costs $10USD and it takes literally 15mins to install (I have done it in less) base what you are willing to pay for that installation service on that. For example, if you want a $10 TEC installed and the prospective installer wants $100 for that service run away like a little girl because they are charging you what they think you will pay based on your ignorance as opposed to what is fair and proper. Nothing you need to do to install a TEC would demand a 90% markup. You’d be better served sourcing your own TECs and spending the $30 on a torque screwdriver off of Amazon like I did and install them yourself.

Going forward with the newer P67 2600k/2500k CPU’s and soon to be released AMD Bulldozer chips TECing is looking up. They don’t require extreme cold to get great results. I have my 2600k on water (no TEC) running at 5.0GHZ 24 x 7. They just don’t get that hot until you go past this point and start applying the big voltage. Then again that’s the whole point of freezing chips. The voltage isn’t really what kills a chip it’s a combination of heat and voltage. Control the heat and the more voltage the traces in the cpu can withstand without damage. In general silicon performs better cold.

If you’re still interested in trying a TEC set up just do what I did. It works surprisingly well and it will allow you t get your feet wet as well as getting familiar on insulation techniques (I have a thread on how to do this about here somewhere). Grab yourself two small pieces of ¼â€(or metric equivalent) aluminum (cut a sheet if necessary) sized about ¼†larger than the TEC/TECs (start with one trust me ) you are trying to use. Get a mirror and some 600/800/1200 grit sandpaper from the auto shop and lap both sides of each plate. Drill 4 holes in the corners and bolt the TEC in between with the correct torque. There is a formula for this but even that isn’t exact for various reasons. If need be I am sure someone can supply the formula. Then all you need do is get a high end (EK Supreme HF full copper/ Koolance 360/70 class and bolt that down on top of your TEC sandwich and the CPU.

Lastly, remember it’s really all about fun and experimentation in the end and be careful TECing can be and is addicting.

Regards,
Elloquin
    
CPUMotherboardGraphicsRAM
4930k EVGA X79 Dark EVGA GTX 980 Hydrocopper GSKILL 
Hard DriveHard DriveOptical DriveCooling
6 x 300 GB Western Digital Velociraptors Raid 0 Western Digital Velociratpr 10k LG Custom Water Cooled 
OSMonitorKeyboardPower
Windows 7 Ultimate 64 AOC - 27" Widescreen Flat-Panel IPS Razer  EVGA 1300w G2 
CaseMouseAudio
Danger Den Torture Rack Razer Creative Sound Blaster X-FI Titanium HD 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
4930k EVGA X79 Dark EVGA GTX 980 Hydrocopper GSKILL 
Hard DriveHard DriveOptical DriveCooling
6 x 300 GB Western Digital Velociraptors Raid 0 Western Digital Velociratpr 10k LG Custom Water Cooled 
OSMonitorKeyboardPower
Windows 7 Ultimate 64 AOC - 27" Widescreen Flat-Panel IPS Razer  EVGA 1300w G2 
CaseMouseAudio
Danger Den Torture Rack Razer Creative Sound Blaster X-FI Titanium HD 
  hide details  
Reply
post #2 of 10
+REP

Excellent post.

I gather you don't agree with ultrasonics TEC calculator?
post #3 of 10
I'm guessing this is somewhat in response to my thread...


Don't get me wrong , I understand how wrong the maths can be, I've seen acceptable calculations with over a 100% margin for error.


And If I'm going to get into TEC cooling I want to do it right first time since I'm looking at spending several hundred pounds, I can't afford to build custom blocks, buy 3 separate liquid loops, buy a specialised PSU and then have it fail.



I understand where you're coming from, I'm guessing a lot has been deleted from the stickies, since this is the first mention of specific pressure requirements I've heard of.

Hell I've learned quite a bit more just from reading this post, I don't mean to be too cheeky, but if you know any decent resources for reading about peltiers, I'd really like to read more on them.

All I can seem to find is either hugely out of date or academic based.
    
CPUMotherboardGraphicsRAM
e5645 @ 4.2Ghz P6X58D-E 5450 x2 6Gb Corsair Dominator 1600 
CoolingOSPowerCase
ALPENFOHN K2 Windows 7 Pro Antec Truepower Trio 650W Dell Optiplex 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
e5645 @ 4.2Ghz P6X58D-E 5450 x2 6Gb Corsair Dominator 1600 
CoolingOSPowerCase
ALPENFOHN K2 Windows 7 Pro Antec Truepower Trio 650W Dell Optiplex 
  hide details  
Reply
post #4 of 10
Thread Starter 
Quote:
Originally Posted by xtremetechuk View Post
+REP

Excellent post.

I gather you don't agree with ultrasonics TEC calculator?
It's not specific to anyone really. It is what I said.

Quote:
Originally Posted by Crabid View Post
I'm guessing this is somewhat in response to my thread...

You're fine mate. Was just venting off things that have been building up over time really.

Don't get me wrong , I understand how wrong the maths can be, I've seen acceptable calculations with over a 100% margin for error.


And If I'm going to get into TEC cooling I want to do it right first time since I'm looking at spending several hundred pounds, I can't afford to build custom blocks, buy 3 separate liquid loops, buy a specialised PSU and then have it fail.

It does absolutely add up rather quickly. That said if your resourceful you can do it economically (not meaning cheap).

I understand where you're coming from, I'm guessing a lot has been deleted from the stickies, since this is the first mention of specific pressure requirements I've heard of.

Hell I've learned quite a bit more just from reading this post, I don't mean to be too cheeky, but if you know any decent resources for reading about peltiers, I'd really like to read more on them.

All I can seem to find is either hugely out of date or academic based.
Feel free to ask questions. It may take a bit for someone to get back to you but people like Zipdogso, Xtremetecuk, Un-nefer and others including my self will surely help you. Ultrasonic has a site which I am sure someone can point you to. He has an excellent all in one post about TECs in general what they are and the various formulas like torque settings.

Regards,

Elloquin
Edited by Elloquin - 4/2/11 at 7:28am
    
CPUMotherboardGraphicsRAM
4930k EVGA X79 Dark EVGA GTX 980 Hydrocopper GSKILL 
Hard DriveHard DriveOptical DriveCooling
6 x 300 GB Western Digital Velociraptors Raid 0 Western Digital Velociratpr 10k LG Custom Water Cooled 
OSMonitorKeyboardPower
Windows 7 Ultimate 64 AOC - 27" Widescreen Flat-Panel IPS Razer  EVGA 1300w G2 
CaseMouseAudio
Danger Den Torture Rack Razer Creative Sound Blaster X-FI Titanium HD 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
4930k EVGA X79 Dark EVGA GTX 980 Hydrocopper GSKILL 
Hard DriveHard DriveOptical DriveCooling
6 x 300 GB Western Digital Velociraptors Raid 0 Western Digital Velociratpr 10k LG Custom Water Cooled 
OSMonitorKeyboardPower
Windows 7 Ultimate 64 AOC - 27" Widescreen Flat-Panel IPS Razer  EVGA 1300w G2 
CaseMouseAudio
Danger Den Torture Rack Razer Creative Sound Blaster X-FI Titanium HD 
  hide details  
Reply
post #5 of 10
Missed you Elloquin! About 2 weeks into seeing all the stuff that was going on in the forums of bickering, and of Practice vs Theory, I realized experimenting myself is probably the best bet. Because in the end, a setup like this is customized to suit me, and no one else. The person the system is for is the one who has to say "That's good enough." and you will never get that with numbers unless you are truly ignorant. I hope to see you around more often dude, and I'd love to see what you can get going.
Whitey!
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II x6 1090T BE Crosshair IV Formula HIS Radeon HD 5870 2GB 4x2GB Corsair Dominator DDR3-1600 
Hard DriveOSMonitorKeyboard
Corsair 32GB SSD, Seagate 1TB, Western Digital 1TB Windows 7 Ultimate SyncMaster 245B and SyncMaster 2343 Logitech G15 v2 
PowerCaseMouseMouse Pad
Antec 850W True Power, Meanwell 600W 24V Painted CoolerMaster Half 932 Logitech G5 Weighted Mouse Steelseries Frosted Glass 
  hide details  
Reply
Whitey!
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II x6 1090T BE Crosshair IV Formula HIS Radeon HD 5870 2GB 4x2GB Corsair Dominator DDR3-1600 
Hard DriveOSMonitorKeyboard
Corsair 32GB SSD, Seagate 1TB, Western Digital 1TB Windows 7 Ultimate SyncMaster 245B and SyncMaster 2343 Logitech G15 v2 
PowerCaseMouseMouse Pad
Antec 850W True Power, Meanwell 600W 24V Painted CoolerMaster Half 932 Logitech G5 Weighted Mouse Steelseries Frosted Glass 
  hide details  
Reply
post #6 of 10
I have been trying to say for a long time that charts and formulas are just ballpark figures, trying to be accurate is just foolhardy and time wasting.

Nobodys set up is the same, just the quality of the copper/aluminium you use for plates will affect things. And few go as far as getting a torque screwdriver to mount them properly. As elloquin said even testing is awkward because absolutely everything including the ambient temp has to stay exactly the same through testing......of course it won't.

All the attributes of a TEC change during running so results you have carefully formulated go straight out the window with regularity. The only attribute of a TEC you can be 90%+ sure of is the physical size and thats it.
post #7 of 10
Quote:
Originally Posted by zipdogso View Post
I have been trying to say for a long time that charts and formulas are just ballpark figures, trying to be accurate is just foolhardy and time wasting.

Nobodys set up is the same, just the quality of the copper/aluminium you use for plates will affect things. And few go as far as getting a torque screwdriver to mount them properly. As elloquin said even testing is awkward because absolutely everything including the ambient temp has to stay exactly the same through testing......of course it won't.

All the attributes of a TEC change during running so results you have carefully formulated go straight out the window with regularity. The only attribute of a TEC you can be 90%+ sure of is the physical size and thats it.
While I agree, In-theory I find Ultrasonics TEC calculator to be just as accurate as manufacturer charts, and are more than just ball-park figures. Especially considering the calculator allows users to change Th temperature, opposed to choosing between 25C, 27C or 50C.

Ambient temperature will not change drastically, at least not to the point where calculations become useless. However thermal resistance from metals and thermal compound will effect performance, which is why in many of my posts regarding calculations, I state that these figures do not taken into account of inefficiencies.
post #8 of 10
Nice post Elloquin and very imformative.

But like xtremetechuk above, I also find Ultrasonic's TEC calculator to be pretty accurate and I do not think it should simply be ruled out because of Ultrasonic himself.

What some need to understand though is that the TEC calculator is only as good as the information entered, and the results are very specific.

So TEC coldside is the TEC coldside, not what the final temp of the CPU or GPU chip will be - and I have tried to ensure I state that in every post where I use the results it provides. To work what the actual CPU or GPU temps will be, you'll need to move away from the theory and do some hands on tests - like what you and others have tried posting on OCN in the past, and for which I am very greatful I had a chance to view them before they were trashed by some members and then removed by a particular forum mod.

But anyhoo - so what the TEC calculator is good for is working out how the TECs in a TEC setup will perform and how efficient and how much heat the TEC setup can move BEFORE you spend all your money on that TEC setup.

It will show that TEC setup A might have a much colder TEC coldside then TEC setup B before you go and spend you money on TEC setup B, and at the end of the day, if the TECs are colder, then you can summize that your CPU or whatever you are going to cool with it will also be colder - and that is really the whole point IMO.

We want to know what will cool better and what is needed to do that, we want to know which setup is best for a specific temp or a specific cost or a specific thermal load - and that is something that Ultrasonic's TEC calculator can help out with.

Before I stumbled upon Ultrasonic's TEC calculator I was working with convoluted calculations in a spreadsheet, that took a lot of effort and a lot of time to get a ballpark idea of how a TEC would perform, but now I can quickly do calculations at any voltage and at any thermal load and get an instant result - and I'll take that result every time now.

Lastly, Ultrasonic's TEC calculator is pretty good, but to be honest, it is lacking some things and has not been updated with the latest TECs - so I've actually started using a different TEC calculator now
Edited by un-nefer - 4/2/11 at 3:51am
Crimson Basilisk
(25 items)
 
  
CPUMotherboardGraphicsGraphics
Intel i7 2600K @ 5.1Ghz AsRock Extreme4 Gen3 Palit GTX 680 Jetstream 2GB Palit GTX 680 Jetstream 2GB 
RAMRAMHard DriveHard Drive
G.Skill F3-12800CL9D-8GBXL G.Skill F3-12800CL9D-8GBXL Sandisk Extreme 120GB SSD Sandisk Extreme 120GB SSD 
Hard DriveOptical DriveCoolingCooling
WD Caviar "Black" 1TB SATAII LG WH14NS40 Blu-ray Burner XSPC Rasa CPU (Acetal) Black RX360 Triple Fan Radiator 
CoolingCoolingOSMonitor
XSPC DDC Two BayRes BMaverick Laing DDC (MCP350) Windows 7 Ultimate x64 Philips 201B 21" CRT 
MonitorKeyboardPowerCase
ViewSonic VX922 19" LCD Diamond Digital Black Antec True Power 750w Antec LanBoy Air Red 
MouseMouse PadAudioOther
Epic Gear Meduza (By GEIL) Steel Series QCK+ XL Asus Xonar DX ROCCAT Kave 5.1 Surround Sound Gaming Headset 
Other
Sennheiser hd555 
  hide details  
Reply
Crimson Basilisk
(25 items)
 
  
CPUMotherboardGraphicsGraphics
Intel i7 2600K @ 5.1Ghz AsRock Extreme4 Gen3 Palit GTX 680 Jetstream 2GB Palit GTX 680 Jetstream 2GB 
RAMRAMHard DriveHard Drive
G.Skill F3-12800CL9D-8GBXL G.Skill F3-12800CL9D-8GBXL Sandisk Extreme 120GB SSD Sandisk Extreme 120GB SSD 
Hard DriveOptical DriveCoolingCooling
WD Caviar "Black" 1TB SATAII LG WH14NS40 Blu-ray Burner XSPC Rasa CPU (Acetal) Black RX360 Triple Fan Radiator 
CoolingCoolingOSMonitor
XSPC DDC Two BayRes BMaverick Laing DDC (MCP350) Windows 7 Ultimate x64 Philips 201B 21" CRT 
MonitorKeyboardPowerCase
ViewSonic VX922 19" LCD Diamond Digital Black Antec True Power 750w Antec LanBoy Air Red 
MouseMouse PadAudioOther
Epic Gear Meduza (By GEIL) Steel Series QCK+ XL Asus Xonar DX ROCCAT Kave 5.1 Surround Sound Gaming Headset 
Other
Sennheiser hd555 
  hide details  
Reply
post #9 of 10
I don't think anyone was ruling out Ultrasonics TEC calculator because of himself, and like Elloquin said the post was not specific to anyone. I highly respect Ultrasonic for the help and TEC guide, and Ultrasonic deserves more credit.

Elloquins post was implying that TEC calculators in general are only good for ball-park figures at best, which may be the case if used incorrectly. TEC calculator accuracy greatly depends on setup, direct die or waterchiller. TEC waterchillers are subject to additional thermal resistance and other inefficiencies from coolant to CPU waterblock, and so the results from Ultrasonics TEC calculator will not be as accurate. Which is a no-brainer; of course there will be a difference.

Direct die setups are subject to less thermal resistance and inefficiencies, and so the difference between Ultrasonics TEC calculator and real-world results will be smaller. We all know TEC charts and calculators can’t compete with real-world testing, however I believe Ultrasonics TEC calculator to be a very good indicator of performance, and should not be disregarded.
post #10 of 10
Quote:
Originally Posted by xtremetechuk View Post
I don't think anyone was ruling out Ultrasonics TEC calculator because of himself, and like Elloquin said the post was not specific to anyone. I highly respect Ultrasonic for the help and TEC guide, and Ultrasonic deserves more credit.

Elloquins post was implying that TEC calculators in general are only good for ball-park figures at best, which may be the case if used incorrectly. TEC calculator accuracy greatly depends on setup, direct die or waterchiller. TEC waterchillers are subject to additional thermal resistance and other inefficiencies from coolant to CPU waterblock, and so the results from Ultrasonics TEC calculator will not be as accurate. Which is a no-brainer; of course there will be a difference.

Direct die setups are subject to less thermal resistance and inefficiencies, and so the difference between Ultrasonics TEC calculator and real-world results will be smaller. We all know TEC charts and calculators can’t compete with real-world testing, however I believe Ultrasonics TEC calculator to be a very good indicator of performance, and should not be disregarded.
extremechuck..

I don't think his calculator works because if you can't cool a tec to its max then you cant get max dt. and so everything is scewed with that screwed up calculator..

regards
lich
LKTEC
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 2600K @ 4.2ghz GA-P67A-UD4-B3 mobo 2x ATI 5770 in crossfire ripjaw 8gig 1.5v 8cas 
Hard DriveOSMonitorPower
69gig ssd Windows 7 Ultimate Old Hp f2105 Thermaltake 850 
Case
Haf 932 
  hide details  
Reply
LKTEC
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 2600K @ 4.2ghz GA-P67A-UD4-B3 mobo 2x ATI 5770 in crossfire ripjaw 8gig 1.5v 8cas 
Hard DriveOSMonitorPower
69gig ssd Windows 7 Ultimate Old Hp f2105 Thermaltake 850 
Case
Haf 932 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Peltiers / TEC
Overclock.net › Forums › Cooling › Specialized Cooling › Peltiers / TEC › Things I have learned about TEC’ing over the past year and why I took a vacation...