Swapped an Eheim 1048 for a Eheim 1250, Interesting results!

[nikhsub1]

Well as the topic says, I switched my pump to an Eheim 1250 from the 1048. The rest of my system is in my sig. So far, higher flow don't mean jack! I am actually seeing higher temps, about 3C. Of course the system is not 100% bled, probably 95% and the AS3 will need time to settle. With the 1048 temps were 41-42C load (on die reading) with a case temp of 26C. With the 1250 and the SAME case temp I load at 45-46C. The D-Tek Spir@l is a high flow block too (I think ) What do you think? Here is how my system is set up:



Here is a pic of where the pump is, this is BEFORE the rad/shroud re-do:

[Skulemate]

Those are interesting results. Have you confirmed that your flow is actually better? Your results could be due to the fact that your new pump is operating less efficiently than the previous one (i.e. closer to the maximum head side of the pump's H-Q curve). Course, that's just a guess.

[nikhsub1]

Originally posted by Skulemate
Those are interesting results. Have you confirmed that your flow is actually better? Your results could be due to the fact that your new pump is operating less efficiently than the previous one (i.e. closer to the maximum head side of the pump's H-Q curve). Course, that's just a guess.

Don't know exactly what that means in English, but why would the 1250 run less efficient than the 1048?

[bigben2k]

Easy question, tough answer.

A pump is rated to output a certain amount of power, and it will do so, regardless of the flow restriction.

However...

Where you were using an Eheim 1048, resulting in a flow rate that allows the pump to be say, 70% efficient, you now have an Eheim 1250 which, although will give you a better flow rate, will also run with a flow rate that is much, much lower than its maximum output.

Look at it this way: if you used one of those 1.0 horsepower pool pumps, it would be struggling, right? Struggling to push more flow that is... It would certainly give you a better flow rate, but for the amount of power that this sucker is drawing, it's not very efficient at all.

But it still doesn't explain your results.

First, there's the additional pump heat that this beast will put in your rig, but the rad should be able to handle it.

The problem is that you're pushing your rad out of its sweet spot.

Try this: use the 1048 to loop the coolant through the rad, and the 1250 to loop the CPU waterblock. (yes, a dual res).

[Slipknot]

First off, I'm a true noob, but if the pump is running less effeciently, is the wasted energy being converted into heat that is added to the water? Does the pump sound like is is running hard/harsh?

[Skulemate]

Ben, I doubt that it's the radiator that's responding poorly to the higher flow. Furthermore, I think that the results in a dual reservoir setup will be far worse than you'd initially think due to the mixing of coolant in such a setup. I still think that it's a pump efficiency issue... yes you've lost your "sweetspot" so to speak, but it lies in pump selection rather than your rad.

[myv65]

The pump energy is only part of the answer. The difference in power of those two pumps is not enough to raise the water temperature by 4°C. Not unless your true flow rate is on the order of 10 gallons per hour anyway.

What other temperatures have you measured? How are you defining "under load" for the CPU?

Normally the culprit is crud plugging up the radiator fins, changes in room air temperature (as much as 2-3°C from floor to desk level), air in the systems, etc.

After all that blah, blah, blah, I wouldn't personally expect more than a 1°C change if all that changed was the pump (and correspondingly the flow rate).

[pHaestus]

I would wait for the air to completely bleed. It is not possible to tell from this picture whether you are using a 90 degree elbow from the block's outlet to the pump intake. If so, I would rethink the loop so that isn't needed.

A system will have a sweet spot: too low and the waterblock's performance is compromised, but too high and radiator performance typically suffers. I wouldn't think an Eheim 1250 would add enough heat to make any difference; air in the rad certainly would though.

[Gerwin]

I don't see what so interesting about it. How do people get it into their heads that higher flow equals lower temps. You have to get rid of the heat somewhere, and that somewhere is the radiator. Even if your flow is 10^6 gallons a second, that doesn't make your radiator any more efficient. The water is only the medium that transports the heat. The temperature of your cpu depends on the temp of the water that's in it. As long as the water flows fast enough so that it can transfer the heat from the cooler to the rad, and that it doesn't have the time to heat up inside your cooler, you're ok. Faster flowing water would heat up less in your cooler, but would also cool less in your rad. Plus that for a high flow rate, you need a strong pump to overcome the resistance, and all that produces extra heat.
I run a wc system with about everything watercooled: cpu (2GHz P4), northbridge, two hdd, vidcard. I run a Eheim 1046 with 8mm piping, and my cpu never comes over 45C, and usually runs at 35C.

[bigben2k]

Originally posted by Skulemate
Ben, I doubt that it's the radiator that's responding poorly to the higher flow. Furthermore, I think that the results in a dual reservoir setup will be far worse than you'd initially think due to the mixing of coolant in such a setup. I still think that it's a pump efficiency issue... yes you've lost your "sweetspot" so to speak, but it lies in pump selection rather than your rad.

pump, rad, they have to be selected together. Wether you pick a rad for a pump, or pick a pump for a rad, it's all the same.

I'm going to call you on what you wrote about that dual res setup. How do you figure that it would be worse? How would the coolant mixing make it "far worse"?

The idea here is to isolate the flow to the rad, from the flow to the WB. This way, the rad can run within its sweet spot, and so can the CPU, as both loops share the same res, so the heat can get out.

[bigben2k]

Originally posted by myv65
The pump energy is only part of the answer. The difference in power of those two pumps is not enough to raise the water temperature by 4°C. Not unless your true flow rate is on the order of 10 gallons per hour anyway.

Normally the culprit is crud plugging up the radiator fins, changes in room air temperature (as much as 2-3°C from floor to desk level), air in the systems, etc.

After all that blah, blah, blah, I wouldn't personally expect more than a 1°C change if all that changed was the pump (and correspondingly the flow rate).

Actually, the 1048 is rated for 10 Watts, and the 1250, for 28W. That's an extra 18 Watts. It's nowhere near enough to explain a 4C temp diff, I agree.

We'll have to wait on nikhsub1 for more details.

[the creature]

nikhsub1: Just a little question, what case is that and where did you get that heater core?

[pHaestus]

Looks to be a Liteon FS020.

Ben: Not sure about the whole "try to get high flow through block and low flow through rad" thing. Better procedure to either design block to increase velocity at that point in the loop or get a bigger rad (with low resistance).

Bigger pumps don't always help; this is known. But a higher flow rate will always improve temps provided the radiator's performance doesn't suffer (and can deal with extra heat of pumps).

[bigben2k]

Originally posted by pHaestus
Looks to be a Liteon FS020.

Ben: Not sure about the whole "try to get high flow through block and low flow through rad" thing. Better procedure to either design block to increase velocity at that point in the loop or get a bigger rad (with low resistance).

Bigger pumps don't always help; this is known. But a higher flow rate will always improve temps provided the radiator's performance doesn't suffer (and can deal with extra heat of pumps).

Agreed.

If I was going to use 2 pumps with a double res, I'd probably end up with loop 1 having a small pump, a small rad, and I'd throw in a NB or GPU block, to slow the flow a bit, and loop 2 would be CPU only.

[nikhsub1]

Originally posted by pHaestus
I would wait for the air to completely bleed. It is not possible to tell from this picture whether you are using a 90 degree elbow from the block's outlet to the pump intake. If so, I would rethink the loop so that isn't needed.

A system will have a sweet spot: too low and the waterblock's performance is compromised, but too high and radiator performance typically suffers. I wouldn't think an Eheim 1250 would add enough heat to make any difference; air in the rad certainly would though.

There are no 90 degree bends anywhere in the system. The reason for this "test" is that just about everyone uses the 1250 over the 1048 in a 1/2" system. I wanted to see for myself what the hooplah is all about, and so far, people seem to be better off with the 1048. The H20 Motto has always been, "more flow is better, more gph etc. But, I will wait for the system to bleed (should be today) and let the as3 settle too. But, if temps are even the SAME as with the 1048, why would anyone want the 1250? Food for thought.

[BillA]

quite simple really
the wb is at the 'limit' of its effectiveness
which is to say that even a lot more flow gains very little temp reduction

and yet again


what do you think the initial flow rate was ?
and after the pump change ?
use the pump P-Q curves for a fair estimate,
make a simple manometer across the pump's inlet and outlet for more accuracy

if you can measure that gain . . . .
of course the worse the wb is, the more change one will see

an observation:
we are now at the stage where this stuff is known, and has been demonstrated;
perhaps its time to start applying the info to the system's design ?
better components are not needed, more thought in their selection and compatability is

[bigben2k]

Actually BillA, nikhsub1 is using the Dtek Spir@l, not the TC-4.

[morphling1]

When I had my 1250 I did qute some flow tests, and I came to conclusion that this pump is not to efficient in water cooling setup. I had rig with my block, 1m of 1/2 silicon hosing and a heatercore, I measured flow for this setup and it was ~370 l/h , so way bellow rated 1200. And now I'm using two maxy-jets 1000 l/h pumps in series for double preasure output and my flow is now above 500 l/h and both pumps still put out less heat then one Eheim 1250, plus they are much smaller. The most important data you need to look at is preasure output not so much the flow rate.

[BillA]

and ?

what makes you think the curve for Dtek Spir@l would be any different ?

did not ask 'could', used 'would'

please explain

a HUGE part of 'the problem' is the naive belief that each wb creation is unique and represents a new solution blah blah blah

bigben2k
you have a lot of posts here, you see no pattern (about the wbs, eh) ?
you still thinking that each bit of copper's placement opens a new dimension of Physics ?
suggestion: do some grouping, 'optimize' each group, compare -> generalize for a set of 'known facts'

BTY, the spiral in those results is the Cooltech; but a Spir@l is quite the same as the TC-4 - both are labyrinths of similar size and respond similarly

[nexxo]

Interesting... does the 1250 increase flow compared to the 1048? Depends on the flow restriction, as was said. Else all that power is just converted into useless heat.

But aside from that, who says that increased flow means increased cooling?

Let's find out in a thought experiment. Let's say we bolt on a super-duper pump which makes the coolant flow at infinite speed (beyond speed of light, say, Ludicrous Speed). For sakes of argument we will assume this is not going to cause problems with friction, pressure, heat, time-paradoxes etc. blasting your rig into far orbit, because we slapped a SEP (Somebody Else's Problem) field generator on it.

What you will get is that heat will still transfer from block to coolant (as it should) and will transfer from coolant to rad to air (as it also should). The coolant will achieve a certain even temperature overall.

Now we slow down the flow a lot, down to, say, an Eheim 1048. We find that the coolant in the block will get a bit warmer because it hangs around there longer, but that's OK: it will lose that extra heat in the rad by virtue of the fact that it hangs around there longer too (in fact, the hotter the coolant, the greater the transfer of heat). Provided that it doesn't stand still, the slower the flow, the bigger the coolant temperature difference between block and rad, but the overall cooling capacity is still the same. Q.E.D.

[BillA]

addendum

quite agree with morphling1
the pump is the pivotal component in a WCing system, everything else has to match its capability

an interesting thread involved with wb design and pumps is here

nexxo
suggest you swap your speculative thoughts for the facts shown in the above graph

[nexxo]

Hey, all science starts with hypotheses and speculation. The graph, if I interpret it correctly, says sort of the same thing I did? That beyond a certain point, flow increase does not matter.

[bigben2k]

Originally posted by unregistered
and ?

what makes you think the curve for Dtek Spir@l would be any different ?

did not ask 'could', used 'would'

please explain

a HUGE part of 'the problem' is the naive belief that each wb creation is unique and represents a new solution blah blah blah

bigben2k
you have a lot of posts here, you see no pattern (about the wbs, eh) ?
you still thinking that each bit of copper's placement opens a new dimension of Physics ?
suggestion: do some grouping, 'optimize' each group, compare -> generalize for a set of 'known facts'

BTY, the spiral in those results is the Cooltech; but a Spir@l is quite the same as the TC-4 - both are labyrinths of similar size and respond similarly

Ok, I guess you were addressing nikhsub in the first part...

Do you honestly believe that the performance of the spiral is similar to the TC-4? I mean, ok, it's probably close to the TC-4, but you're talking about a spiral versus a Z-type channel block, then you point out that the spiral (above) is the Cooltech, but it's significantly (?) offset in your graph. What gives?

Honestly, what I see is that all the new block (except for the odd crappy exception) have similar performances, and yield the same results, within 2 to 3 deg C, but I haven't paid too close attention to it all.

I'd love to take the time to categorize the blocks out there. Off hand, there's the maze types (like DDs Mazes and TC-4), the spirals (like Dtek), the plain-and-simple (Swiftech, #rotor) and the odd ones (Innovatech). Seems like the pyramids (a la gone_fishin)are stirring up in the homemades, but haven't made a commercial appearance.

This categorization might appear to be based on looks, but I'm sure that if I put some thinking in it, I could justify that it's by flow type. Maybe I'll write an article about it, are you interested in this?

[nikhsub1]

Look guys, I know alot of you are very technical and have done MASSIVE research. I have not. All I have done was swap pumps, and the results are a little shocking to me at least. I realize my results are totally unscientific and unmeasured but so will "most" people's that water cool. From years of browsing forums and reading and doing mostly unscientific research, the info I always got was higher flow is better. I can now say, at least with MY particular set up that it is not. And BTW, my setup seems pretty popular, a heater core, 1/2" lines etc. I just don't want this thread to turn ugly.

[bigben2k]

It's all good... BillA is in a better mood at the OCAU forums, check it out!

The pump efficiency thing is a moot point. My take on it is still that you are pushing your rad with too high a flow.

To prove it, you could install a bypass to your rad, where you'd connect the inlet and outlets together, through a ball valve: leave the valve closed, and you've got full flow through the rad. Open up the valve completely, and you're almost completely bypassing the core. Somewhere in the middle should be your goal.

I don't know if you're willing to try this. It'll cost you a couple of T's, and a valve ($3-4 ?).

The bonus is that you'll achieve an even higher flow rate, since you're reducing the flow restriction from the rad. That's what I call proving a point...