Cpu low usage but high temp

[OkannT]

Hi everyone,

i have this issue where im not doing anything yet stil my cpu is running hot on idle. we are talking about 40/50 degrees when it should be around 27/35.

My cpu is an i7 6700k (skylake)
motherboard is msi z170 gaming M5
AIO liquid cooler is corsair H100i v2

when gaming the smallest games my cpu easily reaches max temps like 90/100 degrees.
the real issue i have is that my core voltage is good around 0,7... and 1,2... (i have it on auto) but stil my temps are way to high. also in task manager it says that my cpu is just using 4/5 procent for standard windows purposes. furthermore my cpu is not overclocked its running auto between 800 and 4200 MHz so that should be fine.

i attached some picture so you guys can see for yourself.
i realy hope you guys can help me out to get normal temps!

 

[milewski1015]

Hitting 90-100C on water while gaming is a telltale sign something isn't right. I would start by double checking that the AIOs CPU block is properly mounted and has good contact with the CPU's IHS.

 

[Caring1]

Hitting 90-100C on water while gaming is a telltale sign something isn't right. I would start by double checking that the AIOs CPU block is properly mounted and has good contact with the CPU's IHS.

Based on those core temps the mount looks good, pretty even across the board and no large difference between cores that would indicate an improper mount.

We also need to know what Case you use, and other cooling solutions, how many fans and location of radiator/s and if the fans blow in or out of the case.

 

[Bill_Bright]

I agree with Caring1. The fact the core temps are consistent suggests the block is properly mounted (at least with even pressure).

Did you apply TIM (thermal interface material)? And if so, were the mating surfaces thoroughly clean of old TIM and dirt and did you use as little as possible of the new TIM while still achieving total coverage? A common mistake is to use too much TIM. TIM only needs to fill the microscopic pits and valleys in the mating surfaces to push out any trapped and insulating air. Any excess is in the way and counterproductive to the efficient transfer of heat.

Are you sure the impeller is spinning?

And as Caring1 also suggested, are you achieving good air flow through the case and through the radiator?

What is your ambient (room) temp?

 

[milewski1015]

@Caring1 @Bill_Bright My mistake. My thought process was that if the block wasn't properly mounted we'd be seeing higher idle temps (40-50C for idle on water is higher than it should be, no?), but a drastically improperly mounted block like I think my mind jumped to would be causing the chip to get much hotter.

 

[cucker tarlson]

how old is that cooler ? maybe you need to refill

 

[milewski1015]

how old is that cooler ? maybe you need to refill

Aren't AIOs closed-loop and therefore can't be refilled?

 

[cucker tarlson]

Aren't AIOs closed-loop and therefore can't be refilled?

some can and all lose liquid eventually

I think all alphacool aios are refillable

 

[Bill_Bright]

My mistake.

Maybe, maybe not. That's why I stipulated "at least with 'even' pressure". "Proper" mounting includes proper cleaning of the mating surface. A proper amount and distribution of TIM. And proper a proper mount means an even and the correct amount of pressure on all 4 corners. Now typically, a 1/4 turn on all 4 mounting screws achieves that so hopefully that is not a problem.

And I will note on several occasions with conventional heatsinks and water blocks, we have seen where that little plastic protective film was left on the heatsink/block!

Aren't AIOs closed-loop and therefore can't be refilled?

Any time there are flexible tubes and fittings, leaks can develop. There may be a manufacturing defect, puncture due to mishandling, or even over time, tubes can become brittle.

 

[WarthogARJ]

some can and all lose liquid eventually

I think all alphacool aios are refillable

I saw this by accident, and it got me interested.
When you say that "all coolers lose liquid eventually".

They are closed systems, and very low pressure, so unless someone screws up and has a leak, the only realistic way is by diffusion through the plastic pipes.

I was curious, so I calcul;ated what the losses would be for a few difference types of plastics.
Attached is a table with some ball park values for dimensions of a cooler with its exchangers and vessels made from either Plexiglass, Lexan or PVC.
And using tubing from either HDPE, PVC or TFE.

The losses can indeed be quite big if you make poor material selections, as it seems most PC cooler designers do.
And you can see the downside o this fetish of gamers to want to use LED's, transparent tubes, and coloured liquid: it pisses water out.

For purposes of illustration, I used a nominal surface area o 0.1m^2 for tubes (1mm wall thickness), and 1 m^2 for vessels from Plexiglass (2mm thickness).

So you can see that a cooler made from Plexiglass and LDPE tubes loses on the order of 1,000ml per year of water, compared to one made using PTFE tubes, and aluminium that would lose 0.3 ml/year....

These are ball park values to show the relative differences, but it does tie up with your comment that "all coolers lose coolant" (or at least the poorly designed ones).

A more sensible materials selection would be to use aluminium vessels and either a very good grade of HDPE, or better, PTFE, for your tubes.
You can get thin walled PTFE that would be fine.

Switching from Plexiglass to Lexan (polycarbonate) for the vessels isn't that much better, it's about half the permreability.
For the expense, you're better off using aluminium.
And aluminium vessels conduct at MUCH better than plastic, which is your end goal in any case.

I suppose as long as you don't mind continually checking and refilling your cooler it's not a big deal, but I think it's just poor engineering design to need to do that. And could be expensive if you don't realise you're below critical levels and you fry your over-clocked CPU....:-}

For small components you could use plexiglass, but you really don't want to go much above around 1,000 cm^2 surface area, which gives you around 100 ml water loss per year.

And seeing what the PC Coolant manufacturers charge for their stuff, you could afford to make your own using gold plated tubing!!

 

[Vya Domus]

It takes years and high liquid temperature (~45-50c which is unrealistic) to lose enough water such that it becomes a problem.

I can only see a couple of reasons for your issues :

- radiator is clogged with dust, no air goes through it and it gets heat soaked
- pump is not running at full speed / failing
- bad mount, yes you can have a bad mount even if temperatures are evenly spread out, remember that there is a heat spreader on the chip exactly for this reason

 

[WarthogARJ]

It takes years and high liquid temperature (~45-50c which is unrealistic) to lose enough water such that it becomes a problem.

I can only see a couple of reasons for your issues :

- radiator is clogged with dust, no air goes through it and it gets heat soaked
- pump is not running at full speed / failing
- bad mount, yes you can have a bad mount even if temperatures are evenly spread out, remember that there is a heat spreader on the chip exactly for this reason

Actually, that's not the case if you have enough permeable material in your system.
Work through the calculations and you can see for yourself.

In any case, I'm not saying that this is what's causing the OP's issue, I was showing that water loss over the course of a long enough time can indeed be an issue. And it does'nt need o be at high temperature either.

As Cucker Tarlson noted: "all cooling systems lose water".

In terms of your cooling system not working very well for any reason, it should have a temperature sensor on the WATER side, so you can measure how well the cooling water side is running. It's EXTREMELY inexpensive to add a sensor, and any commercial system without one is really not a good system.

In the short term you can easily monitor it by using a multimeter that has a plug-in thermocouple and just fasten it tightly to a pipe to get a relative value. And then clean your exchanger, check your flow rates etc etc and see what that does to it.


If you're interested in how to do these calculations, I can give you more detail.

The permeability of plastics to water vapour (I say vapour, because it's not LIQUID that is diffusing through, but the vapour) is well known in electronics sensors and related.

In terms of a "leak" per se, we're not talking much water loss: that's why I gave a column for water loss in terms of nanograms/day. But it adds up.

To be really accurate, in absolute terms, you'd need to know the actual values for the temperatures of the system, and the ambient relative humidity in your system, because that's what's driving the diffusion. This calc'n assumes a pressure differential of 1 atmosphere, which is on the high side, but then it assumes 25 deg C, which is low. So they cancel out to some extent.

But the idea is with something small like this, is to use overkill on material selection so you don't actually NEED to do a very complex design calculation. As in try to use materials that are very resistant to water diffusion. And that's easy, since you can buy PTFE tubing and aluminium is both easy to machine, and is made in suitable sizes or automotive cooling purposes.

I the objection is that you cannot easily bend PTFE into nice 90 degree elbows as you can LDPE, well.....I would ask you why you'd WANT to do that in any case?? It's much better to have a more gradual radius in terms of pressure losses.

As far as having transparent tubing so you can see the coloured water you use....well, if you don't add colouring, then there's no colour to SEE, right? If you TRULY need colour and flashing lights, then you can order one of these off eBay....:-}

 

[Vya Domus]

Actually, that's not the case if you have enough permeable material in your system.
Work through the calculations and you can see for yourself.

I don't know how well your calculations reflect the real world, what I do know is that AIOs do not become useless as early as you suggest, otherwise no one would be using them.

 

[WarthogARJ]

I agree with Caring1. The fact the core temps are consistent suggests the block is properly mounted (at least with even pressure).

Did you apply TIM (thermal interface material)? And if so, were the mating surfaces thoroughly clean of old TIM and dirt and did you use as little as possible of the new TIM while still achieving total coverage? A common mistake is to use too much TIM. TIM only needs to fill the microscopic pits and valleys in the mating surfaces to push out any trapped and insulating air. Any excess is in the way and counterproductive to the efficient transfer of heat.

Are you sure the impeller is spinning?

And as Caring1 also suggested, are you achieving good air flow through the case and through the radiator?

What is your ambient (room) temp?

In terms of solving the issue, these are good suggestions.
But before you start chan

ging stuff, you should benxhmark where you are now, so you know what the issue actually was, and how bad it was.
Because it can occur again, and maybe it's not what you thought it was, and you only "corrected" it accidently.

I'd buy a dedicated temperature sensor off eBay or Amazon like the attached. It's battery driven so you won't need to connect up a power supply. otherwise get a 5V DC one and splice it into your system.

If you measure the temperture sensor carefully to get the OD, you can make the correct sized hole into your water system AFTER the heat exchanger. Drill a hole in the same type mateial to check the fit: you want it tight, but not TOO tight since it's easy to crack some plastics. Then use some sealant to water-proof it.

A very nice method would be to use a bigger block of some plastic you can securely glue to a tube, and then drill a longer hole through that to accept the sensor. Then you can use more sealant over a larger area to make sure it won't leak.

I don't know how well your calculations reflect the real world, what I do know is that AIOs do not become useless as early as you suggest, otherwise no one would be using them.

You're missing my point.
I never said any specific coolant design becomes useless.
Although evidently this OP has an issue, and the system is poorly desined so it doesa't indicate what the issue is....hmmmm.....

My FIRST point was only to comment on Tucker Carlson's point that "all coolant systems lose coolant".
I was surprised, and then did a quick calculation, and saw that yes indeed, that with most mateials used by the designs I've seen, they CAN indeed "lose" water.

Whether or not it's SIGNIFICANT enough to cause any SPECIFIC one to become "useless" is really a function of that SPECIFIC type.

However, from the perspective of materials engineering, and electronics/control design, and chemical engineering, most of the designs I've seen are pretty poor. Especially for what is charged for them.

What do you pay for one? I it's anything over $50, then you're being robbed if they don't have a temperature sensor in it to indicate how well it's working.

I mean seriously, you're running some CPU at higher temps than it was designed to run at with OEM cooling, and if it gets too hot you fry it. Or at least it doesn't run as well as you'd like it to. So why not do the job properly? Do it right the irst time.

Personally I wouldn't buy one from a company, I'd make my own: very easy to get the components, and to design it to suit your own needs.

 

[Vya Domus]

You're missing my point.
I never said any specific coolant design becomes useless.
Although evidently this OP has an issue, and the system is poorly desined so it doesa't indicate what the issue is....hmmmm.....

My FIRST point was only to comment on Tucker Carlson's point that "all coolant systems lose coolant".
I was surprised, and then did a quick calculation, and saw that yes indeed, that with most mateials used by the designs I've seen, they CAN indeed "lose" water.

Whether or not it's SIGNIFICANT enough to cause any SPECIFIC one to become "useless" is really a function of that SPECIFIC type.

However, from the perspective of materials engineering, and electronics/control design, and chemical engineering, most of the designs I've seen are pretty poor. Especially for what is charged for them.

And I said they can lose water, it just takes a long time and high coolant temperatures to become a real issue. By the time you lose enough water you have to worry about other stuff such as the pump failing.

What do you pay for one? I it's anything over $50, then you're being robbed if they don't have a temperature sensor in it to indicate how well it's working.

I mean seriously, you're running some CPU at higher temps than it was designed to run at with OEM cooling, and if it gets too hot you fry it. Or at least it doesn't run as well as you'd like it to. So why not do the job properly? Do it right the irst time.

Personally I wouldn't buy one from a company, I'd make my own: very easy to get the components, and to design it to suit your own needs.

Making a loop yourself will cost way more than 50$, plus it's absurd to expect everyone can or will do that. AIOs are a cheaper and easier solution.

 

[WarthogARJ]

I dunno where you lads have worked, but you can classify cooling systems in two types: with poor instrumetation and with good instrumentation. That's going from computer cooling, to automative to industrial grade in refineries and such.

The former costs you about 10% less up front, and then costs you twice as much over it's lifetime in terms of inadequate coling, downtime, and frustrated trouble-shooting.
The latter costs you 10% more and .......works properly........

Because you can SEE when it starts to go bad, and be pre-emptive about it.

Maybe 50 years ago there was more of an excuse to be a skin flint on instrumentation, but nowadays with the plethora of very inexpensive (and not "cheap" in terms of quality) solid state sensors and circuits, it's VERY easy to have things properly instrumented.

And anyone who has any RBG lights on their liquid cooled computer, and doesn't have at least ONE temperature sensor on the water side......words fail me. Seriously.

If you REALLY want techno bling, then why not at least have PRACTICAL bling?
For peanuts you could have the following:
- Temperature sensors on water lines to and from the CPU, and whatever else it cools: cost <$5 of-the shelf as a self-contained unit, and <$1 per sonsor if you do it as a system connected to an Arduino etc
- Flow meter on the water pump: cost <$10 and VERY accurate
- Temperature sensor on heat exchanger inlet and outlet: as above for cost

I'm not saying you NEED this, but i you want BLING, then rather do this than have flashing disco balls and see through pipes and cases with coloured warer.

All you really NEED is one temperature sensor on the water going TO the CPU, but if you like RGB's and bling, and are into geek stuf, then the rest lets you do a continuous check on your cooling, and if you hook it up to an Arduino etc, you could have a high alarm to warn you of issues. And even log the results.

If this is not within the technical grasp of the average RGB gamer nerd then fine. I'm sure the owner of AIO is happy to take your $$$ for a really overpriced and poorly designed uninstrumented system.....I'm sure they learned their trade installing other poorly instrumented stuff in indusry.....:-}

And I said they can lose water, it just takes a long time and high coolant temperatures to become a real issue. By the time you lose enough water you have to worry about other stuff such as the pump failing.


Making a loop yourself will cost way more than 50$, plus it's absurd to expect everyone can or will do that. AIOs are a cheaper and easier solution.

I did'nt say it would cost you $50 to make a good one.
My point was that if you ARE indeed spending more than $50 then cutting costs without at least ONE temperature sensor is a really poor economy.

Does your have a sensor?
I suspect not.

However, have you ever had to trouble shoot a system with a thermal issue and wished you had at least ONE sensor?
On a car perhaps?

It's like asking an electrician or electrical engineer to fix an elecrical issue, and then say they can't use their multi-meter.
Temperatures and Pressures are like volts.
Flow rates are like currents.

 

[Vya Domus]

90% of people buying these probably have no damn clue what an Arduino is. I maintain my opinion that your idea is absurd, you don't need to monitor liquid temperature/flow rate.

 

[WarthogARJ]

Heh heh heh.....well, go ahead with that opinion......

I dunno what your background is, and no offence, if you've got a fundamental foundation in physics or engineering.
It's no problem if you don't, it just determines the level at which i can use to explain things.

In any case, I'm writing this for any others that pick up the thread and want to learn themselves about the issues.

The fundamental point here is that anyone who starts adding or modifying the CPU clock speed, and needs more cooling, so they don't fry it, is either totally at the mercy of people who DO know what they are doing, or CLAIM they do, or else they can learn the necessary to be able to help themselves.

I'm a fan of helping myself.

And it's not very hard to pick up enough to be able to see that it's very useful to be able to monitor how well your modified system runs.
Or doesn't run.
After all, that's where we came in: we have a guy asking why his modified sysatem is overheating.
And the syatem he's been sold by the people he trusted to do the job right, haven't given him enough to solve the issue.

Or even to be able to pass it to others who DO have the knowledge.

In terms of a LOT of extra thermal monitoring sensors, you've missed my point.
I did NOT say you need to add an extra microprocessor (e.g. Arduino based) thermal monitoring system.
Not did I say you NEEDED to monitor flow rate.
I said you COULD, and in terms of bling it's better than having some really useless flashing RGB's.

I DID say that any add-on cooling system like this needs to have some basic feedback to be able to tell you if it's working.
Basic.
Even just ONE temperature sensor.
If you dispute that, then I think it's best to stop reading now.

Otherwise....

Just looking at the CPU sensor is not much help.
It's like the red oil light on American cars that were called the "Idiot Lights".
When they turn on, you know there's a problem, and a pretty serious one.
But it gives you zero really helpful info about WHAT the problem is.

And some people tend to ignore it in any case: even put a bit of tape over it because it annoys them.....:-}
Sigh.

Better designed cars had more instrumentation, like oil pressure gauges and oil temperature gauges.
And modern cars have a LOT of sensors.

I'm not sure if you lift the hood of your car, but if you do, you can do a lot of the basic checking on your car yourself if you can access this instrumentation.
For $10 and a free app that runs on a smartphone you can see what the dealer does on their diagnosis instrument.

I'm not saying you need to be able to replace your cam belt yourself, but with a very minimal outlay, and a bit of reading, you can get on top of one of the most complex bits of machinery you're likely to see.
Be able to pick up issues BEFORE it fails.

And the reason for that, is being able to access the instrumentation.

So the parallel in a computer, is being able to access the various sensors in the hardware.
Assuming they exist.
And when Intel, AMD, Toshiba or Nvidia makes stuff, they add sensors.
For the critical issues.

So the obvious thing for a system critical component like cooling your over-clocked CPU is to have some instrumentation.
I presented a VERY inexpensive and very easy to install idea on how to retrofit a temperature sensor.

I'm sure you could do that if you wanted to.
I'm also sure that any coolant system should have one installed as an OEM feature.

So why don't they?
Well, because most people who jump on the overclock bandwagon aren't told it's a good idea, and don't know enough about physics or engineering to see it's a good idea.

Is there a risk of damage if you don't cool properly?
Yes.

Use Mr. Google, and have a look at all the advice that's given.
They ALL say you need to be careful about proper cooling.
And that you can fry your CPU if you don't have it.
You cannot rely on the CPU's sytems to save it from damage caused by poor cooling.

And use Mr. Google to see the number of times where people have cooked their CPU's.
And that's the low end, because I'm sure most who do that are too embasrrassed to admit it.

OK, so you read all this advice on how and why to overclock, and decide you want to overclock.
You then buy what seems to be a nice cooliong system.
Lots of transparent tubers, clear panels and coloured water.
Yee hah.

You then install it, and run your system.
And do whatever checks you read are needed.
Yee hah!
All good.

Ah....but is it?

It's only all good if your cooling system STAYS working properly.
If it starts to degrade, for any reason, your CPU isn't getting cooled.
And you can indeed fry it.
Very quickly.

And you might well not see the signs from the CPU Side for various reasons.
And without anything on your cooling water side, you're clueless about it getting gradually worse.

Is like driving down the highways, the idiot Light comes on, and before you can pull over....boommmmmm.......hsssssss.....and a cloud of smoke from in front.

Or sometimes that happens without the idiot light going on.


If you're going to cool your computer, you should do it right the FIRST time.
(1) Benchmark it BEFORE you start to fiddle with things.
(2) Look at upgrading the case BEFORE you start adding cooling.
More space makes it a LOT easier to add the cooling.
(3) Check that the existing fans are sufficient, and positioned right
Any liquid cooling system will work a LOT better if you get a good airflow through your case
(4) Don't get the cheapest or coolest looking cooling system, try to get something that's been designed for your specific situation.
And seriously, find one with a temperature sensor, or ask how you can add one if you don't know how to
(5) Benchmark your system in the new setup, and THEN start to mess with overclocking
And WRITE stuff down, in terms of loads and temperatures.
Clearly enough so that in a years time you know what it means.

 

[Vya Domus]

After all, that's where we came in: we have a guy asking why his modified sysatem is overheating.

And he got very clear and concise answers on why that might be, then you started to go onto these strange tangents with blocks of text that have basically no real relevance to the subject.

I dunno what your background is, and no offence, if you've got a fundamental foundation in physics or engineering.
It's no problem if you don't, it just determines the level at which i can use to explain things.

Do not worry about my background.

 

[ThrashZone]

Hi everyone,

i have this issue where im not doing anything yet stil my cpu is running hot on idle. we are talking about 40/50 degrees when it should be around 27/35.

My cpu is an i7 6700k (skylake)
motherboard is msi z170 gaming M5
AIO liquid cooler is corsair H100i v2

when gaming the smallest games my cpu easily reaches max temps like 90/100 degrees.
the real issue i have is that my core voltage is good around 0,7... and 1,2... (i have it on auto) but stil my temps are way to high. also in task manager it says that my cpu is just using 4/5 procent for standard windows purposes. furthermore my cpu is not overclocked its running auto between 800 and 4200 MHz so that should be fine.

i attached some picture so you guys can see for yourself.
i realy hope you guys can help me out to get normal temps!

Hi,
Isn't this a delid candidate chip ?

Either way most aio/ clc coolers like your corsair h110i the pump dies after 36 months
Only good thing is it has a five year warranty
If in doubt get a good air cooler to test with you'll need one sooner or later anyway dark rock has some nice ones

 

[CandymanGR]

Thats why i believe that AIO sucks (i've had two in the past. It is not that i havent tried them). The problem is that the pump WILL fail eventually (and suddenly usually), or the bearing will dettach from the water fan blade assembly, and you will get f......d royaly. Air cooler is the way to go in my humble opinion.

P.S. I suggest to also check with a good scanner for viruses. It might be a virus i encountered recently which doesnt load the cpu too much, but the temps get too high. Just a thought.

 

[Bill_Bright]

When you say that "all coolers lose liquid eventually".

They are closed systems, and very low pressure, so unless someone screws up and has a leak, the only realistic way is by diffusion through the plastic pipes.

"The only realistic way is by diffusion?" Ummm, no. Not true. That's in an ideal world and nobody lives in an ideal world.

You are assuming there are zero defects in the manufacturing and assembly of the cooler assembly and you are assuming zero impurities in the raw materials used. Until Man can create perfection 100% of the time, there will always be units that are not "perfect" even from the best manufacturers using the best manufacturing techniques and highest quality raw materials.

You are also assuming the product was not damaged during packaging, shipping, and installation.

Okay - let's assume 100% perfection, zero defects, and zero damage during installation.

You forgot about "time". While it may take centuries for Mother Earth to decompose plastics back to harmless materials , it may only take a year or two for plastics to age, become brittle and crack - especially after 100s or even 1000s of heat-up/cool-down cycles and the accompanying expansion/contraction of the materials that goes with those heat/cool cycles.

It may take only a year or two for hose fittings to work loose due to those same heat/cool cycles as well as vibrations in the case from impellor, drive and fan motors, not to mention footfalls, kicks and knocks to the case and more. And where glues and sealants are used, they can break down long before the plastics.

So I agree with the statement, "all coolers lose liquid eventually".

Some of us grew up on liquid cooling systems long before AiO coolers came about. I remember having to visit multiple fish aquarium supply stores and my local True Value and Ace Hardware stores to get the right pumps, reservoirs, hoses and clamps I needed - then hope the block I found in the Popular Mechanics and Computer Shopper "mail order" catalog really was compatible. I remember having to mix the right ratios of anti-corrosion and anti-mold and water formulas for the coolants. What a hassle

I remember leaks.

And I remember overcoming all those challenges is what made water-cooling fun. Today? Meh!

While AiO coolers today are sealed and almost foolproof, don't think for second leaks cannot or will not develop. They certainly will - sooner or later.

Another observation I noticed over the years is new alternative cooling users will diligently perform regular visual inspections to ensure no leaks have developed. They may even do this every week or so for a year or two. And that's great! But as "time" marches on, and inspection after inspection reveals no problems, the intervals between each inspection gets longer and longer - right when everything is aging and the inspections should be done more frequently! Then disaster strikes. And where do many radiators sit these days? Above the electronics.

 

[WarthogARJ]

And he got very clear and concise answers on why that might be, then you started to go onto these strange tangents with blocks of text that have basically no real relevance to the subject.

Do not worry about my background.

Actuallly.....no.
If you found them strange tangents, then I suspect you didn't read my reply very carefully.

Although if anything, I'm maybe a bit guilty of hijacking the thread a bit, but it was in response to Cucker Tarlson's comment that: "All coolers lose liquid".
Initially I didn't see how it could be true, but a small amount o analysis shows that the materials often used for liquid cooling are not very good.

And THAT then brought the point up that they aren't very well designed either.

"The only realistic way is by diffusion?" Ummm, no. Not true. That's in an ideal world and nobody lives in an ideal world.

You are assuming there are zero defects in the manufacturing and assembly of the cooler assembly and you are assuming zero impurities in the raw materials used. Until Man can create perfection 100% of the time, there will always be units that are not "perfect" even from the best manufacturers using the best manufacturing techniques and highest quality raw materials.

You are also assuming the product was not damaged during packaging, shipping, and installation.

Okay - let's assume 100% perfection, zero defects, and zero damage during installation.

You forgot about "time". While it may take centuries for Mother Earth to decompose plastics back to harmless materials , it may only take a year or two for plastics to age, become brittle and crack - especially after 100s or even 1000s of heat-up/cool-down cycles and the accompanying expansion/contraction of the materials that goes with those heat/cool cycles.

It may take only a year or two for hose fittings to work loose due to those same heat/cool cycles as well as vibrations in the case from impellor, drive and fan motors, not to mention footfalls, kicks and knocks to the case and more. And where glues and sealants are used, they can break down long before the plastics.

So I agree with the statement, "all coolers lose liquid eventually".

Some of us grew up on liquid cooling systems long before AiO coolers came about. I remember having to visit multiple fish aquarium supply stores and my local True Value and Ace Hardware stores to get the right pumps, reservoirs, hoses and clamps I needed - then hope the block I found in the Popular Mechanics and Computer Shopper "mail order" catalog really was compatible. I remember having to mix the right ratios of anti-corrosion and anti-mold and water formulas for the coolants. What a hassle

I remember leaks.

And I remember overcoming all those challenges is what made water-cooling fun. Today? Meh!

While AiO coolers today are sealed and almost foolproof, don't think for second leaks cannot or will not develop. They certainly will - sooner or later.

Another observation I noticed over the years is new alternative cooling users will diligently perform regular visual inspections to ensure no leaks have developed. They may even do this every week or so for a year or two. And that's great! But as "time" marches on, and inspection after inspection reveals no problems, the intervals between each inspection gets longer and longer - right when everything is aging and the inspections should be done more frequently! Then disaster strikes. And where do many radiators sit these days? Above the electronics.

Sure, good points about bad joints.
And you should check for leaks if you are using water in an expensive electronics system.

But even without a defect, or poor installation, or age, my point was that the plastics used for these coolers are poor choices.
They actually DO "leak" by a diffusion mechanism.
Enough to affect performance if there isn't a big reservoir.
I was surprised how much they CAN "leak".

As far as "age" goes, I don't see that these plastics will degrade from use in just a few years.
The temperatures and use-cycles aren't severe enough.

I won't go into too much detail, but these polymers degrade by five mechanisms: photodegradation (usually UV), thermo-oxidative degradation, hydrolytic degradation, hydrolysis and biodegradation by microorganisms. Welds would make it more likely, and high stresses.

And none of the conditions for these are going to be present unless something is REALLY unusual.
In the time frame.

If you want to confirm this, or to learn more about it, here are a few good references:
- ASM International: Chracterzation and Failure Analysis of Plastics (2003)
- Wright - Failure of Plastics and Rubber Products (2001)
- Shah - Handbook of Plastics Testing and Failure Analysis (2007)
- Moalli - Plastics Failure Analysis and Prevention (2001)

In terms of the instrumentation and control nd of things, I haven't looked at large numbers of "off-the-shelf" computer liquid cooling systems, but the ones I've seen have not been very impressive in terms of what you get for your money.

I started out as a chemical engineer in the oilfield and then in a refinery complex: the general principles of good design don't change as a function of scale.

If you look at what the computer OEM CPU/system designers have in terms of the OEM cooling instrumention and controls, why would you think some yahoo should get away with running the CPU's at extreme conditions with some 3rd party cooling and then use LESS instrumentation???
Seems pretty silly to me.

Which liquid cooling systems do you think are well designed? I'd be interested in seeing a good example.

As far as 3rd party fans are concerned, I've seen the ones made by Noctua: they look good. And they have a PWM speed controller that interfaces with the CPU, so you don't lose that functionality.

 

[Bill_Bright]

But even without a defect, or poor installation, or age, my point was that the plastics used for these coolers are poor choices.
They actually DO "leak" by a diffusion mechanism.
Enough to affect performance if there isn't a big reservoir.
I was surprised how much they CAN "leak".

As far as "age" goes, I don't see that these plastics will degrade from use in just a few years.

When it comes to alternative cooling in computers, your points are contradicting.

Yes, they do leak. But the average life of a computer is 3 - 5 years, 7 in extreme cases. But those who implement alternative cooling solutions are either enthusiasts or professionals. And they tend to upgrade and replace every 2 or 3 years - long before age or leakage over time becomes a problem - assuming all fittings and couplings remain properly intact.

Also, you are speaking about plastics in general. Not all plastics have the same characteristics. The plastics designed for use in hosing and tubes are not going to act as sieves! The plastics used are going to be selected because they are designed to block and contain the substances inside.

So today's AiO coolers - at least those from the reputable makers - are not going to employ tubing that allows the coolant molecules to leak past the plastic molecules in the tubes. Therefore, such leakage is NOT a concern as you are suggesting - assuming no physical damage to the system due to extreme age, or abuse.

As for reservoir size, in computer systems you don't use a large reservoir to compensate for leakage. You use a large reservoir because it takes a lot more heat from the components and a lot more time to heat up larger quantities of coolant. So a larger reservoir is used because it results in more efficient cooling. Not to compensate for coolant molecules escaping by worming their way between the tubing molecules.

 

[claes]

Today’s coolers? There’s really only one player in the CLC market, Asetek, and they aren’t innovating much. My third gen Asetek cooler dried out in two years, with audible slushing noises when I shook the radiator. It was strapped to a Fermi GPU though...

I’m sure more traditional uses wouldn’t dry out as quickly, but the h100i v2 is 4 years old or so. I’d give it a shake.