MCT saved my system!

[Liquid3D]

I want to share a story i feel will be helpful. Last year in January we reviewed (http://www.madshrimps.be/?action=getarticle&articID=292) MCT-5 and MCT-40 from Midwest Cooling Technologies made exclusively for Danger Den (http://www.dangerdenstore.com/product.php?productid=144&cat=11&page=1). The fluid performed as advertised and had been in my systems for the better part of this year. Then something horrific happnened. As the one of the thumbnails below shows the fluid had spilled onto my Sapphire X800 XT and into the PCI-ex slot of a Asus P5AD2-E Premium mobo. Even though the spill was substantial, there was no artifiacting, no anomalies, no indication anything had happened. What does this mean?

Simply put MCT did it's job. Given the time which elapsed, residue or addtives in the fluid actually "baked" onto the card's hot spots. I estimate the spill went unoticed for about 3-4 days. When a fluid spills onto a graphics card what destroys the card and can even fry the system is roughly a form of electrolysis (http://www.chem1.com/CQ/ionbunk.html). In such cases the fluid (albeit distilled H20 which has aquired TDS) acts as electrolyte between positive and negative graphic card SMD-leads. These leads are normally isolated and the same can occur to pins carrying current in the PCE-ex slot once an liquid spills over them.

From my experience, Distilled H20 as the liquid breaks down and looses it's non-conductive properties Galvanic corrsion begins to occur. This is also known as dissimilar metal corrosion which can occur where an aluminum radiator and a copper waterblock come into electrical contact once a corrosive eletrolyte is introduced. So while you think your safe and then a spill occurrs it's this eletrically conductive fluid which comes into contact with live components. The end result being a severe arc or short. When this occurs it's usually followed by immediate artifacting then a system-reboot. Eeven attampting to enter the BIOS is usually impossible as the artifacting gets worse and the system continues to freeze. Another indicator is the BIOS warning the graphic card's 6-pin power connector is unplugged (among other anomalies) even though it is plugged in. This happened to me about a year ago with distilled water, water weter and anti-alge mixture I'd concocted.

Most recently I only discovered the MCT spill after I pulled the card. I was about to install another Asus board (P5ND2-SLI Deluxe) and a BFG 6800GT it was at that point I'd noticed the residue remaining from the spill. I completely freaked out even though there was no prior indication the spill had occurred. I simply thought I lost a $200 mobo (got it new)and $600 card (got it new) and my life was saturated with self-pity for a few moments. Bottom line the card and slot cleaned up real nice using 92% isopropyl achohol, the Sapphire X800 XT and Asus mobo are both working fine! I not only highly recommend this product I'm indebted to Danger Den and Midwest Cooling Technologies. By the way they called me and were very concerned and I've found out their releasing a claning fluid specifically designed to remove their product soon. It's wetting agent (glycol?) is really wet, I imagine that's also the anti-corrsion agent as well. Finally our data indicates (along with all the other reviews) that MCT out-performs distilled H20 as temps rise, and that gap widens as the temp rises further MCT becomes more effective...

[saratoga]

I've actually soaked a board in water before while its running. The results are interesting. Of course noncoductive water is a pipe dream thanks to CO2 in the air acting as an electrolyte, but the voltages are also so low that theres not enough current to actually short anything. Rather you get this odd sort of corruption as some of the traces leak current between one another. Eventually it crashes, but it works for a while.

At least thats my experience.

[n00b 0f l337]

I've had plenty of leaks onto hardware that I havent noticed. Eventually I do, clean it up, let it dry and it works just fine. mct becomes conductive and no different that most other liquids quite quickly upon contact.

[MaxxxRacer]

I have had similar experiences as n00b and saratoga.. If you are careful, even a spill on the video card or mobo wont kill it..

Ive soaked my dfi mobo, my msi mobo, several video cards, and even got a hdd wet. All of which were un dammaged once i dried them off..

so considering this, it is no surprise that you didnt kill your hardware when the MCT spilled on it..


BUT atleast the MCT wasnt actually conductive and it didnt dammage your hardware.


oh one thing.. Water doesnt break down.. It gets contaminated with particulates and other chemicals, which then ionizes the water. The H20 ions in the water release electrons, which in turn makes the water conductive.. when you start with deionized or distilled water there are no such ions and thus it is non conductive.

[Holst]

Where as ive had just a few drops kill a motherboard (they went under the NB package on an 8K3A+ and it died instantly)

Glad to hear that this stuff works L3D, rather have you test it than me :P

Just out of interest how did you get a leak?

[moonlightcheese]

this post is just a commercial... lol

[Bloody_Sorcerer]

I've dumped an entire can of rootbeer on a running computer and it was fine. Pulled the plug, sopped it up with paper towels, booted it again and it was happy. Smells like rootbeer and is quite sticky though.
I've also had mountain dew dumped in the top of my computer (through a fan) and it splattered everywhere (including INSIDE my powersupply). After towelling it off a bit everything still works just fine.

[thelostrican]

I have had similar experiences as n00b and saratoga.. If you are careful, even a spill on the video card or mobo wont kill it..

Ive soaked my dfi mobo, my msi mobo, several video cards, and even got a hdd wet. All of which were un dammaged once i dried them off..

so considering this, it is no surprise that you didnt kill your hardware when the MCT spilled on it..


BUT atleast the MCT wasnt actually conductive and it didnt dammage your hardware.


oh one thing.. Water doesnt break down.. It gets contaminated with particulates and other chemicals, which then ionizes the water. The H20 ions in the water release electrons, which in turn makes the water conductive.. when you start with deionized or distilled water there are no such ions and thus it is non conductive.

ufff same here

[BRiT]

Quick, someone change the channel... There must be something better on than this infomercial. ;)

[nikhsub1]

Yeah, commercial it is... In that whole post, you never once pointed out how the leak happened? Do you use hose clamps? I'm also LMFAO at the review... MCT-5 outperforms pure water? Better alert NASA ROFL.

[ZX7891]

i spilled water all over my cpu and mobo and it lives today

[afireinside]

Yeah, commercial it is... In that whole post, you never once pointed out how the leak happened? Do you use hose clamps? I'm also LMFAO at the review... MCT-5 outperforms pure water? Better alert NASA ROFL.


I had a DFI NF3 SIT IN A PUDDLE OF WATER for 2 hours while I tried turning it on and wondering why it wouldn't work. Was working good as ever 2 days later.

[eBoy0]

I run a chiller and the only way my card shows any form of failure is it moister gets into the PCI-E slot, then it gets all goofy looking. Water doesn't AUTOMATICALLY fry your stuff, i ran on a soaked card while running 05 until i noticed dripping.

[Liquid3D]

Trolling for dollars.

The leak was casued as I was installing brand-x into brand-x and brand-X leaked. When I awoke before going to class I re-secured the hoses in brand-X which is why the leak seemed intermittent.

[Holst]

Was this using push fittings then?

Ive allways been a bit wary of them myself.

[Tulatin]

Push fits almost never leak unless you've put them in wrong. I've soaked plenty of boards over time (and killed a Rad 9000 this way), not to mention had a BIP pee all over the inside of my case due to a loose pushfit.

[MaxxxRacer]

Trolling for dollars??? eh?


I guess I didnt read the part about it performing better than water.. As nik said, we should call NASA as they need to start using something else besides water

[Liquid3D]

I've been an advocate of "push" or quick-connect clamps for about a year. My thoughts are that female hoses over male "barbs" by definition stretch the hose material. Therefore the width of most included zip-ties or clamps is simply inadequate to prevent leakes with any security. This varies depending on the hose material and ID, for example 1/2" is not only heavy it's usually associated with "high flow rate" systems creating a further propensity for spillage. 1/2" tubing are both obtrusive, bulky as well as difficult to work with without loosening the connections compared to push/lock connectors and 8mm internal (for example) hoses.

Screw bands are about the only real logical alternative for 1/2" or greater diameter tubing, however; most dislike their ergonomics. I've pretty much converted to Euro (German) watercooling blocks and systems based on their performance, build quality, ergonomics and the wide selection which is about 20x greater then USA made blocks. They've simply been at it much longer and have evolved much further in watercooling systems as a whole.

Below are photos from a current system I'm testing Alphacool (http://www.alphacool.de/xt/) whose designs currently dominate Caseumbau's (http://www.caseumbau.de/index.php?PHPSESSID=25c428da331cd38040eb5b87f3a7a8 82) top-20 waterblock cooling chart (http://www.caseumbau.de/index.php?page=charts/charts&kat=watercool&PHPSESSID=25c428da331cd38040eb5b87f3a7a882) which is based on this teststand (http://www.caseumbau.de/index.php?page=test150/test150). The only block which currently jousts for 1st place on that chart is Cool-Cases Masterfreezer M-II (http://www.cool-cases.org/) waterblock. I'll be recieving this block to compare against Alphacool's NexXxos XP, NexXxos BOLD, and CAPE KC-42 X2 PRO (http://www.alphacool.de/xt/product_info.php?products_id=325&cPath=5_18_66_261) in the next week. Alphacool (and Cool-Cases) also continually dominate Watercoolplanet's (http://www.watercoolplanet.de/index.php) 50 waterblock cooler chart (http://www.watercoolplanet.de/index.php?open=4) (their teststand is included in that link). I consider these site's to be the some of the very best based on the consistancy of their teststands, the number of blocks they test continually and that they test blocks from around the globe.

[saratoga]

Screw bands are about the only real logical alternative for 1/2" or greater diameter tubing, however; most dislike their ergonomics. I've pretty much converted to Euro (German) watercooling blocks and systems based on their performance, build quality, ergonomics and the wide selection which is about 20x greater then USA made blocks. They've simply been at it much longer and have evolved much further in watercooling systems as a whole.

Nonsense. Spring clamps will seal more evenly then the screw on type even at 5/8 and beyond. I've completely switched to them in my system with great results.


Below are photos from a current system I'm testing Alphacool whose designs currently dominate Caseumbau's top-20 waterblock cooling chart which is based on this teststand. The only block which currently jousts for 1st place on that chart is Cool-Cases Masterfreezer M-II waterblock. I'll be recieving this block to compare against Alphacool's NexXxos XP, NexXxos BOLD, and CAPE KC-42 X2 PRO in the next week. Alphacool (and Cool-Cases) also continually dominate Watercoolplanet's 50 waterblock cooler chart (their teststand is included in that link). I consider these site's to be the some of the very best based on the consistancy of their teststands, the number of blocks they test continually and that they test blocks from around the globe.

Maybe I'm forgetting something, but aren't you the guy who claimed that using the motherboard's thermistor was good enough for a test bench?

Regardless, be careful what you believe! watercoolplanet's methodology is deeply flawed from what I've seen, and as such its been widely critisized. You can't condense a waterblock's performance down to one number. If you could we wouldn't need to care about pumps and flow restriction and barb sizes.

I don't know if you've ever gotten serious about testing or not, but they're not the best example of how to go about producing meaningful data. Do some searches on these forums. MaxxxRacer has done a lot of research lately on these forums into testing. His experience may be of interest to you.

[MaxxxRacer]

I've looked at that testbed a while back and its OK, and better than all of these mobo thermistor guys, BUT its not anywhere close to being what is required for acccurate and useful testing of waterblocks..

that would be dereks or robo's setup..

[nikhsub1]

And watercooledplanet decided to change the test die and never bothered to retest blocks tested with the old die... utter garbage.

[MaxxxRacer]

u serious?? lol.. absolutley and utterly shockingly stupid..

sigh..

[Liquid3D]

I guess this did read like an infomercial and I apologize if it seemed like spam. The truth is living on SSDI $700 per month (and now starting classes again finally) I just happened to be pretty excited the stuff worked. In the past I've lost about $575 in hardware when (failry new) distilled H20 spilled onto a PCI-ex slot damging mobo and vid-card. I don't have a lot of money so I was VERY happy not to loose that card.

Saratoga; I'm not aware of "spring clamps" could you post a picture please?

Thank you for the info about the die-change. That does seem odd since I've never seen nor read where they made that change? Can you link me please? There should be a qualifier in the CPU Kulher Chart indicating which blocks were tested under the new die.

I would honestly appreciate it if someone could tell me "how" I'm explaining this wrong. I don't want to mislead readers. What I gathered from each of the MCT review data (not just mine) was that when temps began to climb above a certain level (in most cases at LOAD) systems with MCT showed lower temps compared to the same systems with distilled H20?

Finally just for the record, I'm just about positive I never said a mobo-thermistor was ever an acceptable replacement for a well constructed sim-die. What you may be referring to were my comments: "...on-die thermal-diodes were getting fairly accurate compared to socket-thermistors." Even using the best simulated-die you can't place the measurment device (thermistor or what have you) "inside" the core which is ultimately where were trying to reduce temps isn't it? While a thermal-diode isn't the most accurate device in and of itself, I would guess a CPU manufacturer spending billions on a technology such as thermal throttling would need a failry accurate device in the core given the nature of thermal throttling. Not to mention shut-down thermal-diodes which have to accurate enough to save a CPU before it's damged. I know damage occurs so rapidly no socket thermistor could ever save a CPU (please correct me if I'm worng (I know you will hehe)) Eeven the PII and AMD XP had internal thermal diodes (an early artcile from AS (http://www.arcticsilver.com/diode_calibration.htm) on calibration). Of course these examples aren't pertinent, at least in Intel's case the temps we see in the BIOS or in temp monitoring software are taken froma thermla diode supposedly placed off center and up near the surafce of the IHS (the coolest spot they could find to avoid panic). In AMD's case It seems the shut-down thermal-diode and temps we see in software such as MBM are one in the same, which is why I've compared many of my readings between the CPU's and comparing an AMD dissipating less wattage then a Pentium the AMD runs hotter at Idle?

Even when I can afford a simulated-die testbed I'll continue to test blocks mounted in systems after the simulated dies tests, because there's still valuable data such as how the case temps will effect cooling when the majority of the system components (pump etc.) sit in higher case temps, ergonomics, ease of installation, build-quality and even the philosophy behind the design, etc. Insofar as "simulated dies" if I can't do it right what's the point, i.e. an authority in the H20-world ProCooling monitored pins from a CPU, meaning they were limited in what most modern CPU's dissipated in wattage with each block they tested. I don't know if he has his sim-die built yet. Will they repeat all tests? Of course Watercoolplanet has tested hundreds of blocks comparitively, although that's no excuse not to re-test.

Finally if your motivation in such passionate replies is based on your concern to stifle bad-science and not something more personal, I would be very grateful if you could provide some links. Also photos of spring clamps and respond (with some civility if possible) to these questions. Thank you all very much.

PS: Insofar as NASA maybe we should call them, they haven't had the best of luck lately. Could be another uniquely American example where "Red Tape Politics" displaced the best science?

[nikhsub1]

Thank you for the info about the die-change. That does seem odd since I've never seen nor read where they made that change? Can you link me please? There should be a qualifier in the CPU Kulher Chart indicating which blocks were tested under the new die.

http://forums.procooling.com/vbb/showthread.php?t=8570
http://forums.procooling.com/vbb/showthread.php?t=8005

[moonlightcheese]

PS: Insofar as NASA maybe we should call them, they haven't had the best of luck lately. Could be another uniquely American example where "Red Tape Politics" displaced the best science?
got a point there... how many space projects have they trashed due to dumbass errors like forgetting to convert from standard to metric measurements... (so proud to be american :rolleyes: )

[MaxxxRacer]

The whole thing about NASA was referring to the fact that other than a few highly expensive liquids, water is the best coolant for its thermal transfer and thermal capacity properties.. This has been known for a very long time and is the reason that we generally laugh at those who claim they got better temps by adding something to their water..

Short of adding copper molecules (on the molecular basis) and having them bind to the water molecules, you arnt going to get better heat transfer.



One would think that AMD and Intel would spend a good chunk of change on making the on die sensor work better. BUT.. they dont.. As long as it gives numbers that are not more than 5 to 10C off, then what does it matter to them.. And who cares if it is entirely consistent.. They most surely dont..

I briefly talked to an AMD engineer/intern who explained the woes of the on die thermal sensor to me and that AMD would be making strides to improve its pitiful accuracy in socket M2. If this actualy happens.. well as the Romans said. its up to the gods.


With all of that said, all the tests done on MCT have been done so on non calibrated mobo sensors which as we all know are as accurate as me pulling a number out of my bum.

[saratoga]

Saratoga; I'm not aware of "spring clamps" could you post a picture please?

I can't find my picture, but Wikipedia has a nice link:

http://en.wikipedia.org/wiki/Image:Two_spring_Hose_Clamps_-_small.jpg

They're used on automotive fuel tanks and have an enormous amount of tension. At least thats been my experience. YMMV.

I did a casual experiment where I filled a tube with rubbing alcohol and then tossed it in my sock drawer to see how long it would hold the alcohol as I rummaged through the drawer. Its been 3.5 years and counting, with no visable loss of fluid :)


Insofar as "simulated dies" if I can't do it right what's the point, i.e. an authority in the H20-world ProCooling monitored pins from a CPU, meaning they were limited in what most modern CPU's dissipated in wattage with each block they tested. I don't know if he has his sim-die built yet. Will they repeat all tests?

IMO there are really only two acceptable ways to test: either build a sim or build a diode reader. Either way, what counts is that you can show that your results are correct. Generally this means either getting close agreement with existing data, or by extensive testing and calibration or better yet, both. I don't think either of these are very expensive, but they are very difficult and require some mechanical ability (die sim) or some electrical skill (diode reader) which makes them problimatic for most casual users. However, I think you will be hard pressed to get the accuracy and repeatablity you need out of anything else.


Of course Watercoolplanet has tested hundreds of blocks comparitively, although that's no excuse not to re-test.

Well aside from the questions of their integrity after they evidently modified a Whitewater in such a way to compromise it's microchannel design (at least according to Cathar), my main compliant is that they don't test correctly. AFAICT they test at one flow rate and don't test what that flowrate is. As procooling, Overclockers, Swiftech, and CoolingMasters have shown, flowrate is EXTREMELY important. And they don't evidently understand that, which makes me question their competance.

That said, I only speak English, so if I've missed some extra data on their site, I'd be happy to have it linked so that I may retract that statement. However, if it is there, I didn't see it, and I found their use of a single irrelevent number to characterize waterblock performance extremely misleading.

[Liquid3D]

Thank you for the thoughtful and informative answers. My edited signature may show my desire to build freindships here not to conflict. I do apologize because I tend to REACT taking things much too personally being a highly sensitive person.

I re-read the post and it does read like an info-mercial, I can understand with all the spamming and inaccurate reviews (some of mine included) that are found on the Internet why more informed people tend to be skeptical. I was honestly happy to not loose a $600 graphics card when it would take me three months to save that much, and by then they release an XT version or even a new GPU. It's frustrating to keep up with so any products released so often.

Anyway I have no affiliation with MCT, nor Danger Den other then requesting products from time to time. I'm fortunate they and others have given me the opportunity to review products, I'm still learning how to write a review that is attractive to the reader and factually informative. I think what they appreciate is my desire to learn. I'm struggling with simply concentrating on the facts over the fodder, there are comflcits as to which test methodlogies to use and the right teststand is currently way out of my price range. I will never claim to know so much as to reject the experience of others. I also know I have a long way to go.

Ultimately I want to learn, for these reasons I will always (ultimately) remain open to the information others can give me. There are some highly intelligent people here, the knowledge in your answers reflects this.

Once again I honestly appreciate your help.

PS: can anyone lead me to an article or link which shows how to construct, or modify a socket-775 and/or 939 to get more accruate thermal measurements? For example 939 seems more likely since AMD is still using pins, but LGA is another challange? A link would help if you don't have the time. If someone would actually take the time to assist me I would be happy to give a waterblock raditor of pump I may have from a review sample for their time. Thanx

[MaxxxRacer]

liquid, www.procooling.com has a guide on how to build a diode reader for A64 and for AXP.. if your interested in building one I can help you out, so send me a PM.

[Liquid3D]

Thank's I appreciate that and I will. I wanted to ask a quick question though. I have a water-temp digital meter, I only have one so I need to place it where it'll do the most good (give me useable data) would you suggest placing just before or after the CPU? And why?

[moonlightcheese]

using just one won't yield any useful test data really. you need to measure dT (deltaT) or the data means nothing.

[saratoga]

PS: can anyone lead me to an article or link which shows how to construct, or modify a socket-775 and/or 939 to get more accruate thermal measurements? For example 939 seems more likely since AMD is still using pins, but LGA is another challange? A link would help if you don't have the time. If someone would actually take the time to assist me I would be happy to give a waterblock raditor of pump I may have from a review sample for their time. Thanx

AFAIK its not possible on 775 since Intel does not allow access to the diode via pins, so you're out of luck. On 939/940/754 its the same as Socket A. Check out the guides on procooling on how to tap into it. On 939 AMD actually did us a favor by putting the diode pins on the corner IIRC, so its much easier to get at them without damaging your CPU/socket.


I have a water-temp digital meter, I only have one so I need to place it where it'll do the most good (give me useable data) would you suggest placing just before or after the CPU? And why?

Once you get a diode reader, put them both in the same temperature water (this will be fun with a CPU!) and read the temperature. This should allow you to calibrate them. Then put the the meter in series just before the block. This will give you the coolant temperatures allowing you to calculate the dT.

What kind of meter? What probes you using?

[Bloody_Sorcerer]

for the 939 mod, cruise on over to www.procooling.com
kinda on topic here: today i discovered a leak from the outlet barb on my swifty storm. It's been dripping on my 6800 probably since install. Everything is still working like a charm. Running distilled + prestone coolant stuff (NOT antifreeze)

[Liquid3D]

It's a thermaltake (http://www.thermaltake.com/watercooling/cl-w0033/cl-w0033.htm) meter, I doubt it's the most accurate on the market compared to professional types, however; it's what i have and I'm sure it's better then my past idea of inserting a thermometer through the tubing such as seen below hehe.

[MaxxxRacer]

the temp meter should be right before the cpu to give you a dT measurment when used with the core temperature.

You should really calibrate that thermaltake unit.. I can assure you it wont be very accurate, but with some known measurement points you can see how far it is off.

Do these two things.

1. Get a gallon jug of distilled water
2. Freeze some of it in ice cubes
3. get a insulating cup.
4. fill the cup with crushed distilled water ice cubes. Fill in the remaining space in the cup with water, filling it ONLY to the to top of the ice level and no further.
5. Read the temperature. If done correctly it will be 0C. If its not you know how far it is off at 0C..

6. Take some of the distilled water and boil it.. I reccomend washing out the pan/pot with distilled water prior to doing this for the best purity.

7. submerge the sensor in the boiling water.. the water will be exactly 100C. if its not.. again you know how far you are off..

If the amounts you are off at 100 and 0C are different, this means that the temperature curve of the unit is not linear. I would say contact thermaltake and ask them for the curve, but its probable they would not even know what you were talking about.


BTW, if cathar or Bill sees that first thing you did.. well Bill will (would have rather) rip you a new one, and stew will probably get a good chuckle like i did..

[phextwin]

AFAIK its not possible on 775 since Intel does not allow access to the diode via pins, so you're out of luck.Surely you could solder the wires onto the appropriate lands.

[MaxxxRacer]

I dont think its possible phex.. mainly becuasey ou would have to remove part of the cpu underside to allow room for the wires.. just a bad situation all together.

[Ouchy]

7. submerge the sensor in the boiling water.. the water will be exactly 100C. if its not.. again you know how far you are off..

Water doesn't always boil at 100C. It depends on the current pressure ie. Height from sea level. You'll want to find out at what temperature it boils at where you live. Also, you'd probably want more then just two extreme measurements to calibrate a sensor for any sort of accuracy. Finally, it's going to be hard to gauge when the boiling actually begins.

Then again, calibrating a sensor that is bound to be inaccurate probably doesn't warrant the time needed to design an experiment that will properly calibrate it. I must just be getting picky. :rolleyes:

[phextwin]

I dont think its possible phex.. mainly becuasey ou would have to remove part of the cpu underside to allow room for the wires.. just a bad situation all together.
The contacts on the back of the board then :P (they still have these on 755 i hope)

Already pre tinned =)

[MaxxxRacer]

the contacts on the back of the board wont work.. pH found this out with A64 and AXP.

the problem is that when the sensors are connected to the mobo they dont allow accurate readings through the SM reader. this was the whole reason for pH going to the lengths he d id.


ouchy: thanks i forgot about the pressure factor.. as to when it boils.. that is pretty easy.. when it starts bubbling and giving off lots of steam.. wait a few minutes from then to make absolutely sure.. and then stick it in.

I agree that the extremes arnt the best, but to get the temps at other points is near impossible afaik without thermocouple or RTD probes to compare the thermaltake to.

[saratoga]

Surely you could solder the wires onto the appropriate lands.

There aren't any. According to pH or Cathar (i forget who) there is no externally accesible diode on the P4. You have to trust the software reader if you want any core temps. Obviously this isn't acceptable for high accuracy measurement. I also flipped through the P4 spec sheets and didn't see anything else suitable, but I suppose its possible we all missed something if you've got time to kill and want to have a look for yourself.


If the amounts you are off at 100 and 0C are different, this means that the temperature curve of the unit is not linear. I would say contact thermaltake and ask them for the curve, but its probable they would not even know what you were talking about.

Well thats probably a thermistor unit, so its already extremely nonlinear (exponetial actually) if its working correctly :) Testing over such a wide range of temperatures probably isn't going to work well even in the best circumstances since thermistors have very high sensitivity, but very narrow range of operation, unlike a thermocouple. You'd need to test over very close temperatures ideally. I suppose you could try it like that, but even if its a good unit, you might still get dissipointing results.

Anyway, my take is that the absolute numbers don't matter much. Remember we just want to know how much hotter the die is then the water. So you just need to see that the probe is linear and reproducable over the narrow range of interest; say 20C to 30C. Then you can compare its reading to the CPU die's reading at a given temperature or better yet, a range of temperatures and come up with a calibration that way. Its not a big deal if it hits the steep part of the exponetal at 50C and becomes useless above that.

[phextwin]

There aren't any. According to pH or Cathar (i forget who) there is no externally accesible diode on the P4. You have to trust the software reader if you want any core temps. Obviously this isn't acceptable for high accuracy measurement. I also flipped through the P4 spec sheets and didn't see anything else suitable, but I suppose its possible we all missed something if you've got time to kill and want to have a look for yourself.Either i am thinking about the completley wrong thing, or you have all missed the two THERMDA and THERMDC 'pins' on the 755 p4. Complete with explicit electrical specificatrions of said diode in the same .pdf.

http://www.intel.com/design/pentium4/datashts/306382.htm

[MaxxxRacer]

well that is the right thing.. but as i pointed out, you cant solder the leads onto the back of the mobo and expect good results..

saratoga... I also would prefer doing measurments in the 20 to 35C range, but he doesnt have a high accuracy thermocouple, thermistor or RTD setup to compare the temps with..

BUT what can be done is this.

take both the cpu (use pH's method of sealing it up) and the thermaltake sensor and put them in a waterbath.. start around 20C and add a little warm water. when the temperature stabalizes take measurements again. If the temperatures are the same around 20 to 35C then you are in good shape.. if they are off, you can adjust the appropriate direction to fix it up..

[phextwin]

well that is the right thing.. but as i pointed out, you cant solder the leads onto the back of the mobo and expect good results..So you did, i missed that post. My bad.

[MaxxxRacer]

lol.. its ok..

[Weapon]

Short of adding copper molecules (on the molecular basis) and having them bind to the water molecules, you arnt going to get better heat transfer.

non-electrically-conductive copperized water....we need that.

of course, a few weeks later someone would come out with new and improved non-electrically-conductive silverized water. :D

[MaxxxRacer]

Rod..

there is a chemical that uses copperized water.. It was alluded to over at procooling.. I think someone priced it out and it was over 1000 dollars to fill up your watercooling loop.. but it might be conductive which would lead to it allowing galvanic corrosion.

[Weapon]

Rod..

there is a chemical that uses copperized water.. It was alluded to over at procooling.. I think someone priced it out and it was over 1000 dollars to fill up your watercooling loop.. but it might be conductive which would lead to it allowing galvanic corrosion.

lol - I was being entirely facetious....and the discount version would be aluminumized water. :P

Anyway, with regards to copperized water (or any other H2O that has been combined with any form of metal), I would think it would be electrically conductive out the wazoo (common copper alloys range from 85-101 on the handy electroconductivity chart which equates with being hella conductive) but that would not necessarily lead to galvanic corrosion provided that copper + copper or copper + brass was the makeup of all of the blocks in the loop. No dissimilar metals (in your water or otherwise) = no galvanic corrosion...hence the synomous name for it - dissimilar metal corrosion.

Since I have it handy anyway, a quick rundown for those who are not familiar with galvanic corrosion:
Galvanic corrosion is the accelerated corrosion of a metal because of an electrical contact with a more noble metal or nonmetallic conductor in a corrosive electrolyte. Galvanic corrosion potential is a measure of how dissimilar metals will corrode when placed against each other in an assembly. Metals close to one another with regards to nobility generally do not have a strong effect on one another, but the farther apart any two metals are separated, the stronger the corroding effect between them [The metal that is higher on the Electrochemical Series will be the cathode. The other metal will suffer anodic reactions and will corrode], however, the actual corrosion in each application is difficult to predict. Typically, the presence of an electrolyte (eg. water) is necessary to promote galvanic corrosion.

[phextwin]

I cant imagine this particulate fluids being to freindly on the pump bearings. Unless the particles were insanely small.

But if it is conductive you could use one of those solid state electro-magnetic pumps.

[saratoga]

Either i am thinking about the completley wrong thing, or you have all missed the two THERMDA and THERMDC 'pins' on the 755 p4. Complete with explicit electrical specificatrions of said diode in the same .pdf.

http://www.intel.com/design/pentium4/datashts/306382.htm

Looking through that, I'm clearly mistaken. With some effort you could probably get at those pins and test just as with an Athlon.