• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Monitoring, loop tuning and expansion

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
I resisted the urge to break out calculus for an example equation...someone needs to get radeon28 in here, isn't he in thermodynamic engineering?
 
1) Ambient + cwdT + awdT = chip temp. If you can make your cwdT smaller then, for the same ambient temp, you will get lower chip temps.

Perhaps I should clarify this by saying that simply heating up your water a bit to get it closer to chip temp is NOT what I mean by reducing cwdT. (In fact, this would simply increase your chip temps.) To reduce cwdT, you would need to improve the heat transfer efficiency of your water block.

... you've seen how a bad cpu mount can affect temps right? Well that actually DECREASES your water temp as well (albeit very little) a good cpu mount will increase your water temp (once again, nominally).

I'd be interested to see data on whether this last point actually holds true. I'm only theorising here, but I reckon that your water temps should stay about the same when you improve cwdT by (e.g.) remounting your block. Here's why:

The chip will consume the same amount of power (wattage) under load, no matter what you do to your loop. (Let's set aside any possible power consumption increases caused by increases in electrical resistance, due to higher heat, for the moment.) For a given awdT and fan speed, your rad will dissapate the same amount (wattage) of heat. The only thing that changes with a better block mount is that the amount of cwdT required to transfer all of the heat given off by the chip is reduced. So, the chip need not reach such a high temp in order to push all of its excess heat (watts) through the block and into the water.

Of course, it is probably more complicated than that. It could be that more heat might be radiated to the air around the chip, rather than the water, when there is poorer conductivity to your block. This extra wattage that is radiated to the air might end up in the water when block conductivity is improved. I'm guessing this will be negligible, though.
 
As I did some testing recently I have the figures for the system in my sig to hand although my sensors aren't particular sensitive but have been 'calibrated' with a known accurate one.

This is a very important point. Calibration is key. Also, most commonly available instrumentation is more accurate in specific ranges (response curves). Too hot or too cold and they can become inaccurate. Overall instrumentation accuracy needs to be considered. A lot of times I'll use the 1/2 least unitl method for a very coarse margin of error. Ex: if you have one inch units then the acuracy is at most +/- 1/2 an inch. I'm also pretty sure you can get intel data on the accuracry of their internal diode's temp measurement. I've gotten these from AMD before.

After a point, this all becomes very academic because of the temperature ranges we typically see and can measure cheaply. Personally, I think we need much better temp measurement before any hard conclusions can be made. So we tend to use trial and error or at least educated guesses for our system sizing etc. Still I suppose good enough to get a general idea what to troubleshoot.

Good discussion everyone, thanks! :attn:
 
Last edited:
evolvedpc - Thank you, I also understand that now.

As I did some testing recently I have the figures for the system in my sig to hand although my sensors aren't particular sensitive but have been 'calibrated' with a known accurate one. With GPU at idle I get:

Ambient = 23ºC
Load Water temp = 28ºC
CPU Load temp = 63ºC

This was after cleaning my loop and is roughly a degree cooler than it was.

Hmmm, that's not what I wanted to hear! You have a cwdT of 35C, which is a good 7 degrees worse than my own. I was hoping that other people would report CPU cwdT figures more in line with my GPU results. (i.e. <15C)

Interesting to note that your air/chip dT is exactly the same as my own - 40C - despite your higher cwdT, which is offset by your lower awdT. Given your higher O/C, you must have a fair bit more rad/fannage than me.
 
After a point, this all becomes very academic because of the temperature ranges we typically see and can measure cheaply. Personally, I think we need much better temp measurement before any hard conclusions can be made. So we tend to use trial and error or at least educated guesses for our systme sizing etc. Still I suppose good enough to get a general idea what to troubleshoot.

Yeah - the trouble with this stuff is that there are a huge number of variables to take into account. Comparisons between different systems will not be nearly so useful as comparisons between different iterations of the same system. (e.g. Did that remounted block lower my temps, any?)

However, having an idea of what makes a good value for awdT and cwdT is one thing that we should be able to compare. That will at least give us an indication of whether we've mounted our blocks properly, which is worth knowing!
 
I know in my case it's unlikely to be a problem with the block. I've switched mine round, firstly to see the effect of lapping a block (before and after) and also to see how an old block compared to a recent one (badly, although not as bad as I expected). I'm assuming it's my particular CPU.
 
I know in my case it's unlikely to be a problem with the block. I've switched mine round, firstly to see the effect of lapping a block (before and after) and also to see how an old block compared to a recent one (badly, although not as bad as I expected). I'm assuming it's my particular CPU.

Could also be your higher O/C. If you're running a higher voltage, you will be consuming (and shedding) more power. For a given cwdT, heat (power) will only transfer so fast, so a higher cwdT will be required to transfer a higher wattage.
 
Yeah - the trouble with this stuff is that there are a huge number of variables to take into account. Comparisons between different systems will not be nearly so useful as comparisons between different iterations of the same system. (e.g. Did that remounted block lower my temps, any?)

However, having an idea of what makes a good value for awdT and cwdT is one thing that we should be able to compare. That will at least give us an indication of whether we've mounted our blocks properly, which is worth knowing!
I agree. As you said each system is different, but I think the over all approach you have is sound.

I've used C/W ratios in the past - C here is usually waterblock or heatsink temp but we tend to use die temp and W is in CPU wattage. Getting a "true" wattage of a CPU is tough when overclocking and overvolting. And not many of us have instrumented our waterblocks so CPU temp is used instead.
 
I agree. As you said each system is different, but I think the over all approach you have is sound.

I've used C/W ratios in the past - C here is usually waterblock or heatsink temp but we tend to use die temp and W is in CPU wattage. Getting a "true" wattage of a CPU is tough when overclocking and overvolting. And not many of us have instrumented our waterblocks so CPU temp is used instead.

That raises an interesting point, actually. Each of my loops has 2 in-line temp sensors in it. They tend to read within 0.5C of each other at all times, so I have a pretty good idea of my water temp. As my flow rate decreases, I start to see differences between water in and water out of my rads. Which water temp should we use for our dT values?

I suspect that water in/out of any given block will be so close as to make the question irrelevant for any loop with a workable flow-rate, but for a big rad, with a low (but still workable) flow rate and a high awdT, there is going to be a difference between water in/out. Which should we use for awdT? Is one more correct? Maybe it doesn't actually matter, so long as we are consistent, but if we are comparing my figures to yours, then we need to make sure we are comparing apples with apples.
 
That raises an interesting point, actually. Each of my loops has 2 in-line temp sensors in it. They tend to read within 0.5C of each other at all times, so I have a pretty good idea of my water temp. As my flow rate decreases, I start to see differences between water in and water out of my rads. Which water temp should we use for our dT values?

I suspect that water in/out of any given block will be so close as to make the question irrelevant for any loop with a workable flow-rate, but for a big rad, with a low (but still workable) flow rate and a high awdT, there is going to be a difference between water in/out. Which should we use for awdT? Is one more correct? Maybe it doesn't actually matter, so long as we are consistent, but if we are comparing my figures to yours, then we need to make sure we are comparing apples with apples.
I agree that we need to use as similar of methods/data points as possible. I have found that similar to what you stated water temp in and out of the rad is very close. We'd need temperature measurements with a .01 C resoluiton (or better) to really see the inlet to outlet difference. Again another reason to use CPU temp to ambient as you suggest.

In this case with a closed loop steady state system (need to use only conservation of energy not mass or momentum), the energy equation can be simplified down to heat rejected Q = Mdot * Cp * deltaT, where Mdot is the mass flow rate of the water Cp is the constant pressure specific heat of water (which can be hard to determine if you don't use pure water) and that small inlet to outlet temp change is deltaT. I tried to measure all of this before and the numbers are hard to work out because of the smalll temp difference. And of course Q is tough to determine on a PC especially when overclocking.
 
Last edited:
Well, I know the way intel monitors the surface temps on their ihs is that they mill a channel into which they put their temp probe...hardocp monitors the same way...I can't think of anyone else who does it like that, probably because you have to put a couple hundred dollars worth of gear in the hands of the man operating the mill. I feel like that would be the most accurate way to measure processor temps...if someone could ask skinnee what he does for his water temps, his seem to be pretty consistent, even if they aren't accurate (which is to say, I can't speak to their accuracy, only their consistency...I know one is accuracy and one is precision, by accuracy i mean on an absolute scale), and air temps are easy. I feel like my above statement about better mounts meaning higher water temps would have to be true. From a physics standpoint, all energy in a system must be conserved, and we can limit the system to the room where the computer is being run.

Electricity is used inefficiently to run the processor, this inefficiency is expressed as heat.
This heat goes 3 places from the transistors, into the ihs, and to the back of the board and into warming the cpu (mostly into the ihs, since it has less thermal resistance)
From the IHS it goes two places - air above the board and into the block
From the back of the board it will go into the air or warm the board, depends...i'm not really sure on what
From the block it can move into the water or into the air, most goes into the water due to lower thermal resistance - some also heats up the block because copper will pull more heat away than water...theoretically, if you ran liquid copper through the loop you'd be :thup:, but that crap's hot too...maybe at low density, i digress
the walls of the tubing are insulated, so thermal loss out the tubing is nominal
Water enters the rad, some water warms the rad, the rest is dissipated by the rad into the air over a huge surface area...
Room temps increase or stay the same, depending on ac/w.e, where the energy goes is no longer relevant to us unless it's raising ambients

so, if we go back to the very beginning, this is how we can get solid temp decreases. We can't make the processor run more efficiently, and when we can, it loses the oc, so that goes out the window. However, we can hit step 2, that's where processor choice comes in, the soldered ihs's will give better thermal transfer. From there, we can indirectly control the amount of heat dissipated into the air by having better thermal transfer into the block (aka, a better mount). This does, however, mean that more total heatload from the cpu will enter the block, and if you follow the chain, more will enter the water. This is theoretical, I cannot say if it will work exactly like this irl, but in terms of theory, better mount = more energy dissipated into the water, which means higher water temperatures, and the increase in rad performance at that higher temp means that the water's equilibrium will only increase slightly. This warmer water is not reflected in cpu temps and actually will be reflected in LOWER cpu temps because the cpu is able to disperse more heat into the water.
 
@m0r7if3r: thanks for the shoutout, I agree with the formulas that you have posted so far. Physics I can be used for energy equations, and I would hate to recommend fluid dynamics or heat transfer, cause that involves calculus, diff eq, and tables which will scare people away. Basic thermodynamics can be tackled though with nothing more than Algebra, so its not to scary.

I detest writing out formulas and using long winded explanations though, so I won't ramble, but if someone had a specific mathematical question about the HT behind a component, I'd gladly do it.

One of these days I might write a "mini-textbook" for watercoolers, which will give all the thermo, HT science, and math they need to understand their systems better, but still make the math simple so heads don't explode... i smell a sticky in my future haha.

Back on track though:


That raises an interesting point, actually. Each of my loops has 2 in-line temp sensors in it. They tend to read within 0.5C of each other at all times, so I have a pretty good idea of my water temp. As my flow rate decreases, I start to see differences between water in and water out of my rads. Which water temp should we use for our dT values?

I suspect that water in/out of any given block will be so close as to make the question irrelevant for any loop with a workable flow-rate, but for a big rad, with a low (but still workable) flow rate and a high awdT, there is going to be a difference between water in/out. Which should we use for awdT? Is one more correct? Maybe it doesn't actually matter, so long as we are consistent, but if we are comparing my figures to yours, then we need to make sure we are comparing apples with apples.

I would use the flow rate reading before the mass flow goes into the radiator, because most physicists or engineers that measure "radiator efficiency" use the mass flow rate (also known as m-hat or m-dot) before the fluid goes through a heat exchanger.

Regarding deltas, it gets a little more tricky, the temp reading before will give you the value that can determine your maximum HT, whereas the temp after the waterblock will give you the value showing your ACTUAL HT.
 
@m0r7if3r - All valid points IMHO. :D

In the best case scenario - all the heat goes into the water and then all the heat from the water goes into the air leaving the CPU cooled the maximum.

Of course at the end of that chain the coolest the CPU can get is determined but the all that heat that was put into the air going into the world to warm it. :shock: lol
 
well, I live in nome, ak...sooo, I don't know bout you guys ;) :D :D :D

I would hate to recommend fluid dynamics or heat transfer, cause that involves calculus, diff eq, and tables which will scare people away. Basic thermodynamics can be tackled though with nothing more than Algebra, so its not to scary.

bah, break out the calculus, lets separate the men from the engineers :clap:

One of these days I might write a "mini-textbook" for watercoolers, which will give all the thermo, HT science, and math they need to understand their systems better, but still make the math simple so heads don't explode... i smell a sticky in my future haha.

I think that could clear up some misconceptions and be tremendously useful...
 
Last edited:
That's a rare statement, someone asking for calculus (even in the engineering community). I'll reserve my math response until my finals are over on the 16th (Solid Mechanics and Design... worthless class imo) and then I promise I'll break out calculus lol...
 
That's a rare statement, someone asking for calculus (even in the engineering community). I'll reserve my math response until my finals are over on the 16th (Solid Mechanics and Design... worthless class imo) and then I promise I'll break out calculus lol...
Solid Mechanics is that like Continuum Mechanics? My favorite ME course was Computational Fluid Dynamics, but I'll stick with my Thermalhydraulics. Double the phases, double your fun! But I digress.

I aggree that "Straight" thermo is more usefull here. You hardly ever need the more advanced stuff in "real world" applicaitons. We even used thermo for my gradutate powerplant design course.

I had a link to write up on the basic thermo of watercooled electronic components. I'll have to look.
 
man, glad to know there're people in here who're competent to keep me in check...I've got the physics knowledge for low level stuff, but outside of applying physics laws to stuff, I'm useless (ask me how to wire something though...I can do that!)
 
I don't wanna super hijack this thread, but Solid Mechanics is more along the lines of studying the moments and forces that occur within an object under load or torsion. I'm glad to hear you enjoyed fluid mechanics, cause I'm taking Fluid Dynamics II this fall.

Maybe once I get a "math explanation thread going" I can run the math and theories by you and m0r7if3r... I want it to encompass fluid dynamics, heat transfer, and thermodynamcis, as well as little things as to why, with mathematical backup, stacking radiators don't work and other common questions that can only be truly explained through some math.
 
I don't wanna super hijack this thread, but Solid Mechanics is more along the lines of studying the moments and forces that occur within an object under load or torsion. I'm glad to hear you enjoyed fluid mechanics, cause I'm taking Fluid Dynamics II this fall.

Maybe once I get a "math explanation thread going" I can run the math and theories by you and m0r7if3r... I want it to encompass fluid dynamics, heat transfer, and thermodynamcis, as well as little things as to why, with mathematical backup, stacking radiators don't work and other common questions that can only be truly explained through some math.
Not sure your curriculum equivalence but I had incompressible (mostly water/liquids) and compressible fluid dynamics (gas flow shock waves etc) which may be like the Fluid Mechanics I and II? CFD is actually taking the formulae from these "Fluids" courses and developing computer codes to model them. Although mostly moot in todays world as there are off the shelf software codes to do all of this.

Before we overly complicate things (as only math can do ;)). What do you think about the OPs original idea? I tend to like simple approaches even if they are approximations because they are easy to use and apply. For me overclocking and optimization/effective use are a good pair and I like the OPs approach.
 
Newby here, just wanted to poke my head in and say Thanks! This is a great thread and I am learning a lot! Rock on guys! :rock: I can't wait till the wife lets build out my water-cooled system. :comp:
 
Back