Inlet and outlet temps = no difference. How can that be?

[DDogg]

Seems like when you think you are starting to understand something, a cold fish slaps you in the kisser to keep you humble.

I added another thermistor on my outlet so as to have one on inlet and outlet. Before that I made sure both agreed when exposed to several different temps. Basically, and I am really confused about this, there is no difference between inlet and outlet. If I really throw serious VCore like 2.2 at it there is a 1C difference.

How could this be? Gee, I actually thought I was starting to understand this stuff.

Would really appreciate some comment and help to understand this.

[jaydee]

In and out temps only very by .5C or less. Probably not enough accuracy in the equipment to measure it. Hence the difficulties to properly test blocks.

[bigben2k]

Yep.

http://www.nordichardware.com/articl...vattenkylning/

Go to page 2 for the formulae:
P*t/(C*m)=DT

Where:
P=power
t=time
C=constant 4190 J/(kg*K)
m=mass of water
DT=temp difference

[redleader]

Water has an astronomical specific heat capacity. The change in temperature is simply too miniscule for your equipment to detect.

[OneMadPoptart]

Hurrah for the laws of heat transfer and thermodynamics.

Now if I could only figure out how to solve those damn differential equations.....

[DDogg]

Originally Posted by bigben2k

Yep.

http://www.nordichardware.com/articl...vattenkylning/

Go to page 2 for the formulae:
P*t/(C*m)=DT

Where:
P=power
t=time
C=constant 4190 J/(kg*K)
m=mass of water
DT=temp difference

Thanks for the replies, especially the link. bigben2k, it seems like you are a near librarian for many of us. A personal thanks for that. I've learned a lot from some that you have posted.

Along those lines, and since I have liberally applied the butter, have you run across any in depth testing of the popular pumps where the the amount of heat dumped into the circuit was measured in lab or semi-lab conditions?

I'm particularly interested in the 12 volts pumps where the cooling effects of flow changes, via overvolting or undervolting, might be negated by changes in the waste heat.

My question may not be clearly stated, so let me clarify a little.

If you increase flow via overvolting, you would also increase the heat dumped into the system which may negate any cooling gains. Inversely, lowering the voltage might decrease flow but also decrease heat dump. I'm basically trying to find the X spot.

People talk about using 2 Swiftech pumps in series, yet they consume 25 watts each so that is 50 watts. Since I don't know how much heat is dissipated from the case to air and how much is dumped into the loop, it is hard to judge whether that would be a good idea. 50 watts into the loop seems self defeating?

I'm sensitive to this because I installed a groundloop and added a larger pump rated at 70 watts to replace my 10 watt fountain pump. It increased my flow like crazy, yet drove up cpu and inlet temps. That was a real eye opened and I don't think I would have noticed if I did not have the constant temp of the groundloop constantly monitored in front of me.

Any information you can point me to would be appreciated.

[bigben2k]

The heat added is relatively negligeable, and IMO, poorly understood. Myv65 and I had a lengthy discussion on this, some time ago.

Let me know if digging up the thread would help: I'll be more than happy to look it up.

[DDogg]

Originally Posted by bigben2k

The heat added is relatively negligeable, and IMO, poorly understood. Myv65 and I had a lengthy discussion on this, some time ago.

Let me know if digging up the thread would help: I'll be more than happy to look it up.

Yes, please, and although I'm no pro at this, I not so sure it is negligible. Clearly, the pump I used perhaps doubled my flow, yet increased temps in the ground loop. When I went back to my smaller pump, the GL temp returned to what it had been. Perhaps there is another explanation. I thought at first the increased flow was better scavenging the heat, but the CPU temp also went up so that could not be it. Comments?

[redleader]

Originally Posted by DDogg

Yes, please, and although I'm no pro at this, I not so sure it is negligible. Clearly, the pump I used perhaps doubled my flow, yet increased temps in the ground loop. When I went back to my smaller pump, the GL temp returned to what it had been. Perhaps there is another explanation. I thought at first the increased flow was better scavenging the heat, but the CPU temp also went up so that could not be it. Comments?

Radiators are generally well over speced verses waterblocks simply because makeing a big, powerful radiator costs almost nothing (heatercore). Conversely extracting extra performance from a block is very hard because you are limited by the size of the die.

Furthermore relatively little energy is added by the pump. 10 or 20 watts means very little when a CPU is putting out 100, and a GPU 60.

Combine the excess radiator capacity with low heat input, and the result is generally a negligable change in temps in all but the most extreme cases.

[DDogg]

Originally Posted by redleader

... Furthermore relatively little energy is added by the pump. 10 or 20 watts means very little when a CPU is putting out 100, and a GPU 60.

Yes, I have seen that type of statement before (many times), but never any hard facts or data to back it up. Saying it doesn't make it so.

10, 20, or 50 watts is just that - watts. Heat is heat, isn't it? Each system has an ability to sink a certain amount of heat. If you add more the system has to compensate by finding a higher temp to achieve equilibrium again, so I think it could be said that the potential extra heat dumped into a system by a pump could cause that shift in equilibrium.

Let's look at it in another way. When you increase VCore you are increasing heat output. What? 10 watts maybe per .1 increment (just pulling that completely out of the air to illustrate the point). If a pump dumps just 10 watts into a system, in effect it would be the same as increasing VCore by .X

At that stage it is not insignificant anymore because it is actually costing you higher VCore potential. At least that is how it strikes me as an OC'er. As said, I'm no pro or engineer. Just a country boy tinkerer, but I wonder if this is not a subject that could bear further scrutiny by those that do have this stuff down cold from a proper math and engineering standpoint (and hopefully those that do not lovingly embrace the concept of "conventional wisdom").

/Add: BTW, redleader, that is not a barb directed at you. Rather one directed at the general BS that gets handed back and around over and over. This subject may not be one of those, but then, it may be.

Something I can say from absolute certainty. There is not one member of this forum who would not appreciate all pumps being reviewed with the added information of how much fluid temp is increased above ambient in a closed loop for a 10 hour cycle, or something similar. Then you could make a better buying decision by knowing:

Pump A is 450 GPH 14 ft head - 11C rise on closed loop cycle/ 55 watts
Pump B is 350 GPH 12 ft head - 2C rise on closed loop cycle/ 12 watts

I know I would sure not be interested in Pump A, would you?

BTW2: I don't have a radiator, I have an external fixed ground loop/sink. I can't turn up the fan to make it better.

[bigben2k]

Here it is:
http://forums.procooling.com/vbb/showthread.php?t=3756

[DDogg]

Originally Posted by bigben2k

Here it is:
http://forums.procooling.com/vbb/showthread.php?t=3756

Thanks, very nice of you. Lot of stuff in that thread to sort and wade through.

I'm coming to really appreciate the Swiftech's performance on 25 watts. Given the specs of the pump, it must be very efficient. I suspect the pump that gave me trouble was a very inefficient one, Heck, like I said, you could not hold it. Literally, it would have burned your fingers after 5 seconds or so.

I wonder how much heat the MAG5 pumps dump into the loop. People may not be getting the great deal they think they are.

[redleader]

10, 20, or 50 watts is just that - watts. Heat is heat, isn't it?

Not really. A watt is a unit of power, not heat. Joules are heat. Generally we just deal with power and temperature since its easier that way. But I get your point about thermal equilibrium. Its just that the change in coolant temps isn't all that significant. Remember the majority of the thermal resistance in a loop is in the block, and increaseing pump capacity decreases the thermal resistance in the block even if it does heat the water more.

Sort of the whole big piece of a small pie or small piece of big pie thing.

Let's look at it in another way. When you increase VCore you are increasing heat output. What? 10 watts maybe per .1 increment (just pulling that completely out of the air to illustrate the point). If a pump dumps just 10 watts into a system, in effect it would be the same as increasing VCore by .X

The effect of increaseing vcore on coolant temps is generally immeasureably small, so I don't see your point. Have you ever tried measureing it? I doubt 10w is going increase water temps more then a few tenths of a degree even in systems with small exchangers.

At that stage it is not insignificant anymore because it is actually costing you higher VCore potential. At least that is how it strikes me as an OC'er. As said, I'm no pro or engineer. Just a country boy tinkerer, but I wonder if this is not a subject that could bear further scrutiny by those that do have this stuff down cold from a proper math and engineering standpoint (and hopefully those that do not lovingly embrace the concept of "conventional wisdom").

How is it significant? What kind of numbers are you estimateing here? I'd say significant would be when its heating the water substantially more then the magnitude of decrease of the temps across the block.

/Add: BTW, redleader, that is not a barb directed at you. Rather one directed at the general BS that gets handed back and around over and over. This subject may not be one of those, but then, it may be.

Something I can say from absolute certainty. There is not one member of this forum who would not appreciate all pumps being reviewed with the added information of how much fluid temp is increased above ambient in a closed loop for a 10 hour cycle, or something similar. Then you could make a better buying decision by knowing:

Pump A is 450 GPH 14 ft head - 11C rise on closed loop cycle/ 55 watts
Pump B is 350 GPH 12 ft head - 2C rise on closed loop cycle/ 12 watts

I know I would sure not be interested in Pump A, would you?

Generally the max power consumption is listed on pumps. Unfortunately this is often far in excess of the real world value, but it gives you at least a very pessimistic upper bound. If you don't trust it, you could always get out a meter and get the exact value yourself. Thats what I did on my 1250, ~9w when hooked up to my loop. A bigger pump will be more, but not so much more that its impractical to cool IMO.

BTW2: I don't have a radiator, I have an external fixed ground loop/sink. I can't turn up the fan to make it better.

I don't have any experience with such systems, but I'd think a bigger pump would make the most sense here since a reasonably powerful pipe exchanger is probably going to be either restrictive or have a lot of slow moveing parallel passages. Either way more flow would probably help a good deal.

Edit: Sorry I had to leave the library I was typeing this post in and walk a few blocks to another building with a computer. Hence the edit.

[DDogg]

Not really. A watt is a unit of power, not heat. Joules are heat.

Ok, Joules it is ;-) Heat, by whatever name, put into the loop is is still heat that has to be dealt with and it can drive up your inlet temp. The higher your inlet temp is, the higher your CPU temp.

Given that, isn't it then factual to say the amount of heat put into a loop by the pump directly effects the inlet temp which directly effects the CPU temp?

The effect of increaseing vcore on coolant temps is generally immeasureably small, so I don't see your point. Have you ever tried measureing it? I doubt 10w is going increase water temps more then a few tenths of a degree even in systems with small exchangers.

Hmm, I'm sensing a dissconnect between us here and so I can't completely understand your point either. On my machine CPU temp is around 20C above inlet temp and rises about 2C for ever bump of ~.03 VCore.

Presently my inlet temp from the gloop is 28C.

VCore/CPU Temp= 2.03/48C, 2.06/50C, 2.075/52C, 2.10/53C, 2.125/54C, 2.150/56C So yes I have measure it to a point. Nothing real scientific. IIRC, when I was on the radbox and opened the window on a chilly morning where ambient air to the rad was around 20C, I could run 2.125/48C. I did not have the inlet sensor installed at that point so I can't quote inlet temp directly.

Generally the max power consumption is listed on pumps. Unfortunately this is often far in excess of the real world value, but it gives you at least a very pessimistic upper bound. If you don't trust it, you could always get out a meter and get the exact value yourself. Thats what I did on my 1250, ~9w when hooked up to my loop. A bigger pump will be more, but not so much more that its impractical to cool IMO.

I'm not terribly interested in the power consumption. I'm interested in how much heat it dumps into the loop because that increases inlet temp where there is a direct correlation to CPU temp.

Remember my semi-rant (sorry for that) started when many good and knowledgeable people insisted that more flow would be better without ever mentioning the negative effects of pump heat dump. This is a glaring omission in my book and one that should be considered as important as GPH and Head.

I grabbed a 300 GPH pond pump and flow increased very substantially, yet inlet temps and the corresponding CPU temps increased. I think the pump I tried was particularly inefficient and so effected my temps more than if I had bought a proper pump more designed for our use. I have a feeling I monitor my inlet temps a little more intensely than many. Because of this I picked up on the problem quickly. I do wonder how many folks are out there that thought, like I, that a pump is just a pump and that it was all about GPH and Head. That is clearly not so. Pump heat dump is an important part of the equation and I don't think that can be disputed. I'm thinking there may be a lot of folks that may actually have higher CPU temps because of buying low efficient pumps.

My central point is that GPH and Pump Head, without the corresponding information on heat dump, makes it impossible to evaluate a pump purchase. There is certainly a point where the increased flow can not offset the increased heat input into the loop by an inefficient pump. That's just an indisputable fact, I believe. Without factual information on the pump efficiency/ heat gain, one is just taking a potshot at when buying a no name pump.

[AngryAlpaca]

Your new, higher flow pump offered higher temperatures because of the higher heat. Estimates put the heat output (inline) from a pump to be 50-60% of the power drawn. Try your overclock with the big pump that offers higher temperature. It may in fact be better despite higher temperatures.

Since the energy is constantly being transferred, we're talking joules, not watts. Joules are for one time things, like combustion.

[DDogg]

Originally Posted by AngryAlpaca

Your new, higher flow pump offered higher temperatures because of the higher heat.

Huh? Having a little trouble understanding your point. My whole point was that the higher heat was caused by the waste heat being dumped into the circuit by the pump. Are you making a different point, or the same?

Originally Posted by AngryAlpaca

Estimates put the heat output (inline) from a pump to be 50-60% of the power drawn.

I've read somewhere where efficiency can be up around 75% on an extremely well designed pump, don't know if that is documented anywhere.

Originally Posted by AngryAlpaca

Try your overclock with the big pump that offers higher temperature. It may in fact be better despite higher temperatures.

Huh? (again) - Sounds like more bad science to me. Maybe I just don't understand what you are trying to say.

Originally Posted by AngryAlpaca

Since the energy is constantly being transferred, we're talking joules, not watts. Joules are for one time things, like combustion.

Yep, I already took that instruction from redleader ;-)

[AngryAlpaca]

Huh? Having a little trouble understanding your point. My whole point was that the higher heat was caused by the waste heat being dumped into the circuit by the pump. Are you making a different point, or the same?

Confirming.

I've read somewhere where efficiency can be up around 75% on an extremely well designed pump, don't know if that is documented anywhere.

Hard to find. You can find pumps that turn 75% of their power into heat that goes into the water, though.

Huh? (again) - Sounds like more bad science to me. Maybe I just don't understand what you are trying to say.

Cathar's testing showed that with more flow and higher temperatures, one's overclock may increase. He also beat a Vapochill with plain watercooling. Just take a go at it.

Yep, I already took that instruction from redleader ;-)

Oh shit! I meant to say the opposite. I was trying to argue with redleader!

It should be, "Since the energy is constantly being transferred, we're talking watts, not joules. Joules are for one time things, like combustion.

[DDogg]

Cathar's testing showed that with more flow and higher temperatures, one's overclock may increase. He also beat a Vapochill with plain watercooling. Just take a go at it.

Maybe so - I did notice when I added the second mini pump in series that I could carry one more notch of VCore and generally things seemed more 'well mannered'. Not a scientific term, but I imagine you know what I mean.

I think you are speaking of the pinprick hot-spot stuff that I read about somewhere a while back. It made the point that while CPU temps may not increase, lower flow allowed 'mini-pockets' or 'pinpricks' of heat to develop in the die surface. These mini-pockets of heat buildup were not enough to raise the CPU temp noticeably, but they would cause CPU failure due to 'junction failure', or something like that. Makes some sense I guess. :-)

Got a direct link laying about for Cathar's stuff you mentioned? If not, maybe I can find it. Thanks.