Understanding radiators

[citronym]

RB, needing to cool the water before it enters your blocks is an old myth. The water temp likely changes less than 0.5º anywhere throughout the loop, so order of components does not matter.

 

[thorilan]

RB, needing to cool the water before it enters your blocks is an old myth

it is not a myth as i have proven hundreds of times. even though a system reaches equilibrium if you put a thermo right before your rad and right after you will see a diference and that should situate 1 thermo between your rad and cpu. if you put a third thermo after the cp you will see that its hotter than its input by more than 1C especially in games.

1c is 1c ,,2c is 2c . while it may not make a diference to the average person it does prove that it is not a myth and for those crazy people that like to squeeeeeeze every last bit out of thier system weather it be by fans that make your ears bleed or spending 300+ dollars on a water block its still valid.

 

[citronym]

I def will not argue that there is a difference, but we end up back in the whats worth what argument...
So I won't tread

 

[sumka]

I'm a sap for clean wiring. What can I say?

What are your temps and what blocks do you use? I almost popped for a Polarflo TT last time because of its flow characteristics, but I just HAD to try a Storm.

It's difficult to explain temps (though I monitor them closely) as they depend a lot on the ambient air temp in the room, rad fan speed, pump speed, whether I'm crunching at the time (usually am, all CPU), benching (CPU & OC'ed GPU) and 'thickness of dust on the rad' setting, LOL.
I'd have to run tests to give some sort of reliable idea of ambient versus core/water temps (delta-t is it called ?). I think they're in the region of about about 6-12C at load depending on the variables mentioned above.

The blocks are somewhat mediocre (I guess) Polarflo TT's. Actually I don't think the CPU block is even designed for being run on a bare-die Athlon like I'm doing. It's probably a bit of a weak link really which could be changed for something better/more suitable like an Apogee/Storm or something to bring improvements. Also I'm only using 2 of the 3 barbs for cooling so it's a pretty daft choice I made in purchasing it with hindsight (1st build with water, aka n00b ).

sumka isnt that d4 pump pulling water out of the cpu block and into the rad? should it not flow the other way?

Yeah, it pulls coolant from the top barb on the block and pumps upwards into the rad (bottom barb is a drain). I think it might have been better to have the pump pushing coolant directly into the middle barb of the block though if memory serves me right it wasn't easily doable due to the way the pump is mounted (suspended by the rad tube). Anyway, as it's a closed loop I didn't imagine pumping up to the radiator would be of any concern as the pump is also pulling coolant down from the rad through the blocks at the same time, just depending on which way you look at things. Regarding correct order of componentry, I'm not sure whether ordering makes any real difference overall (pump before/after blocks/rad, etc.) as I've read allsorts of conflicting opinions over time. I just thought about it, built it, watched how things went temp-wise and all seems well enough for me (I'm not an hardcore overclocker or anything, just wanted a nice quiet & efficient system really).

full res here

 

[Otter]

I did, BTW, read the thread title before deciding to participate. I didn't imply or infer that the laminar effect wasn't worth discussion.

Oh! I didn't mean to imply that you hadn't read the title or that you were implying or infering anything about what was worthy of discussion. I was just trying to be clear about why I brought it up lest someone get the idea that they absolutey had to have a double pass rad to keep their cpu from melting a hole in the floor.

 

[Otter]

it is not a myth as i have proven hundreds of times. even though a system reaches equilibrium if you put a thermo right before your rad and right after you will see a diference and that should situate 1 thermo between your rad and cpu. if you put a third thermo after the cp you will see that its hotter than its input by more than 1C especially in games.

I don't doubt you performed this experiment, but if your data is accurate, I don't understand the physics.

Consider a 200W CPU in a system with a flow rate of 1GPM.

200W dumps 3.33 watt hours, or 11.4 BTU or into the water each minute.

Water weighs 8.3 pounds per gallon, hence 1 GPM = 8.3 lb/min.

One BTU is the energy required to raise the temperature of one pound of water one degree fahrenheit. Hence, the temperature differential across our 200W CPU is
(11.3 BTU/min) / (8.3 lb/min) = 1.36 F = 0.76 C

A more likely 125W CPU would heat 1GPM only .48C, and would produce a temperature differential of more than 1C only if the flow dropped below .5GPM.

Unless your Vcore is insanely high or your flow rate unusually low, I suspect one of your thermometers might be miscalibrated. But you say you've done this hundreds of times?

Most people are more concerned about cooling their CPU, as the GPU is relatively easy to cool. If you put a 50W GPU and a 20W pump between the rad and CPU, and all the pump's energy turns to heat before it reaches the CPU (which it wouldn't), the CPU temp will be only one quarter C higher than if you put the rad right before the CPU block.

Of course, the water temp will rise as it goes through each block and drop as it goes through the rad. The "myth" is that this temperature differential is enough to matter, and I think for most people, .5C at the GPU or .25C at the CPU wouldn't be very important. It might even be less than the bump for taking the long winding road around the case to hit all the components in the ideal order.

 

[Dice]

Using That logic, the watercooling loop would only remove .48c from the 125 watt CPU. There is the flaw. The water will hold more heat than its actual temperature would suggest and it is that heat, the unmeasured heat, that the radiator is removing before the water gets to the CPU block.

 

[greenmaji]

Useing that logic thats all the heat the CPU generates for the rad to remove.

 

[Otter]

Using That logic, the watercooling loop would only remove .48c from the 125 watt CPU.

No. 0.48C is how much the coolant temperature rises as it flows through the CPU block. It will be less confusing if you think about the water removing BTU's or watts from the CPU block instead of degrees.

There is the flaw. The water will hold more heat than its actual temperature would suggest and it is that heat, the unmeasured heat, that the radiator is removing before the water gets to the CPU block.

Huh? If you add heat to liquid water, the water temperature goes up. The only exception to this occurs at the boiling point, and if your loop is running that hot, you've got far worse problems than the order of your components.

 

[Perseus]

Oh! I didn't mean to imply that you hadn't read the title or that you were implying or infering anything about what was worthy of discussion.

It sure seemed like it, but okay.

I was just trying to be clear about why I brought it up lest someone get the idea that they absolutey had to have a double pass rad to keep their cpu from melting a hole in the floor.

I doubt there was any real danger of that happening. It isn't too hard to understand that there are tradeoffs associated with differing design methodologies.

 

[Dice]

No. 0.48C is how much the coolant temperature rises as it flows through the CPU block. It will be less confusing if you think about the water removing BTU's or watts from the CPU block instead of degrees.

You are probably correct. It would be less confusing. While I concede that you are far smarter than I, I don't see your math accounting for these constants and variables:

~Specific gravity of distilled water (90% by volume) + Liquid Ethyl Glycol additive (10% by volume) as is accepted practice.
~Specific heat of both of the above.
~Thermal capacity of both of the above.
~Temperatures of coolant samples. (Which, as I am sure you know, affect the thermal capacity of the coolant.)


I only say this because you are using your math and physics against Thorilan's thermometers, and suggesting his equipment may be faulty.

Since the Pump-rad-block-x-pump loop order 'myth' is easily debateable by both sides of the fence, and since the final results of a through examination would be very helpful to all newcomers, and some of those with more experience, I'd venture to guess that a "Definitive loop component order" thread may get stuck.

Huh? If you add heat to liquid water, the water temperature goes up. The only exception to this occurs at the boiling point, and if your loop is running that hot, you've got far worse problems than the order of your components.

With a 1:1 ratio of heat units to water units? That still doesn't account for the thermal capacity, does it?


Edited for gender-bending reference.

 

[TreeNode]

... and suggesting her equipment may be faulty.

HAH, is this considered a flame at Dan?

 

[greenmaji]

HAH, is this considered a flame at Dan?

Or refering to Dan's rig as a woman, much like alot of guys do their cars

 

[Dice]

Or refering to Dan's rig as a woman, much like alot of guys do their cars

Nice try at saving me, but I ate foot on that one.

 

[greenmaji]

Nice try at saving me, but I ate foot on that one.

Nobody can say I didn't look at the bright side.. (sorry you didn't take the lousy excuse I so convently provided for you )
Now.. How about them rads

 

[Otter]

You are probably correct. It would be less confusing. While I concede that you are far smarter than I, I don't see your math accounting for these constants and variables:

~Specific gravity of distilled water (90% by volume) + Liquid Ethyl Glycol additive (10% by volume) as is accepted practice.

That's a good point. I didn't include the antifreeze, but I know Thorilan uses 5% or less in most of his rigs, hence my numbers will be close. They might be 3% low, but the measured value shouldn't be twice what I calculated.

~Specific heat of both of the above.
~Thermal capacity of both of the above.

The specific heat and thermal capacity of water are included in the definition of a BTU. Again, Thorilan is savvy enough to know that antifreeze can't compete with water as a coolant, and his experience gives him the confidence to use as little as possible. Do a search of his posts with antifreeze as a keyword and you'll see he often points out that people use too much antifreeze in their systems.

~Temperatures of coolant samples. (Which, as I am sure you know, affect the thermal capacity of the coolant.)

Not enough to matter. Pure water is only .43% less dense at 30C than it is at 4C. Hence we can neglect the effect of temperature on the thermal capacity of water for estimates like this.

I only say this because you are using your math and physics against Thorilan's thermometers, and suggesting his equipment may be faulty.

I also suggested my understanding might be faulty. As the two don't agree, at least one of them has to be off. But there is no fight here.

Huh? If you add heat to liquid water, the water temperature goes up. The only exception to this occurs at the boiling point, and if your loop is running that hot, you've got far worse problems than the order of your components.

With a 1:1 ratio of heat units to water units? That still doesn't account for the thermal capacity, does it?

That's why I said it would be less confusing if you considered BTU's (heat) or watts (heat/time) instead of degrees. Celsius degrees are units of temperature, not heat. What's a water unit?

 

[thorilan]

ok to help with some insite. many of you are trying to use a formula that can only approximate the heat change and what happens in a loop. there are so many variables that it would take a lot more than our basic calcs to do so i go by real world .

keep in mind thermos that have .02 accuracy in conjunction with tim and mounting of blocks and ambient and every other variable you can think of will effect this. now my data comes from real world builds where i consistantly get a better temp. when i build some systems i will try more than 1 plumbing route just to see since my systems bleed so fast this is easy.
and what i have seen is that the way i do it i see about 2c better temps at load and they drop faster when load is reduced .
so taken on the whole i have tried to design my systems using good solid principles and it pays off. now the price for the pay off is in case size and shape. generally the case is the number 1 influencing factor in the loop as it determines the loops components and layout. hence if you look at the how to ask questions sticky you will see the pattern of how i chose the questions

 

[Otter]

It sure seemed like it, but okay.

In hindsight, I can see how it might seem that way. But when I wrote the post, that was the furthest thing from my mind. And because I wasn't thinking it or saying it, it didn't occur to me to be diplomatic about it. No offense intended.

I was just trying to be clear about why I brought it up lest someone get the idea that they absolutey had to have a double pass rad to keep their cpu from melting a hole in the floor.

I doubt there was any real danger of that happening. It isn't too hard to understand that there are tradeoffs associated with differing design methodologies.

True. But we'd just been talking about how these small differences in performance can take on exaggerated importance for someone still absorbing the basic concepts. See where I was coming from now? Though it was part of a reply to you, the bit about the thread title was intended as a general comment rather than a reference to anything you'd said.

 

[Otter]

and what i have seen is that the way i do it i see about 2c better temps at load and they drop faster when load is reduced .

Hmmm. 2C better than what? Do you have enough data to separate the effect of other things, like having the pump at the bottom of the loop, from having the GPU between the rad and CPU?

If component order really does make 1-2C difference, do you have any idea about the physics behind it? Do you think there is uninterrupted laminar flow from one block, through the tubing and all the way through the next block, for instance, making the effective flow rate far less than the actual GPM?

 

[thorilan]

its just over all design in order of components and tubing length and 100% bled system ( you would not believe how many people have incomplete bled systems)

when i make a system with other than my standard it increases the tubing length and i can see a visual diference in flow in the res