dual pump question on improving temps

[Xel_NaGa]

I have seen in various threads (http://www.xtremesystems.org/forums/showthread.php?t=265952) people saying that increased flow doesnt seem to improve temps. But potentially could there also be not enough flow? Currently with my pump on 100% power I maybe get .8 LPM. I imagine this is due to length and makeup of my loop.


res> pump> cpu > mobo > gpu > 1x120 rad > 2nd gpu > 3x 120 rad > res
http://i48.photobucket.com/albums/f248/xelnaga15/IMG_0504.jpg

thoughts?

[jinsean]

If you take a look at martin's spreadsheet estimator, the heat dump from a pump actually increases the loop temp a little bit. So any gains in flow could be negated by the heat dump. If you want redundancy in case of pump failure, then go for it, but otherwise, I doubt you'll see much improv in temps by adding another pump.

[Conumdrum]

.8LPM is pretty low. Way under the recommended at least 1 GPM. It might help.

Could you make a sig showing whats in the rig? I see 2X GPU, a CPU, Mobo and possibly two rads. Flow yes, but also raddage might be an issue.

[vhaarr]

I think I'd just try removing the 120 rad from the setup completely and couple the GPUs directly; that radiator with those fittings and all that tubing could be adding more than its worth of restriction. That's just based on looking at your picture though, and just as an exercise; I have no basis to make this claim :P

Could you maybe flip the 360 rad horizontally AND vertically and dremel some holes for the threads? That way you could eliminate a lot of excess tubing there, too.

What are your temps though? You just said you were not pleased with the flow, but if the temps are alright then who cares? ^^

[Xel_NaGa]

.8LPM is pretty low. Way under the recommended at least 1 GPM. It might help.

Could you make a sig showing whats in the rig? I see 2X GPU, a CPU, Mobo and possibly two rads. Flow yes, but also raddage might be an issue.
lowering my pump speed from 10 to 8 makes the cpu raise its temp on the average of 7c under full load.

Equipment.
Silverstone Raven
Intel Core i7-980x Batch# 3035B799 - With EK Supreme HF block
Rampage III Extreme - with matching Danger Den blocks
2x PNY GeForce GTX 580 - with matching Danger Den blocks
6x Corsair Dominators 2GB 2000
Corsair AX1200w
4x Gentle Typhoons 1225C12B5AP15
Swiftech MCR120-QP 1x120mm rad
1/2ID 3/4OD PrimoChill PrimoFlex PRO LRT
Koolance TMS-200 Thermal Interface Controller
Koolance TNK-200-V2 Reservoir, Inline 200mm V2
Koolance PMP-450
Koolance HX-CU1020V 3x120mm rad


I think I'd just try removing the 120 rad from the setup completely and couple the GPUs directly; that radiator with those fittings and all that tubing could be adding more than its worth of restriction. That's just based on looking at your picture though, and just as an exercise; I have no basis to make this claim :P

Could you maybe flip the 360 rad horizontally AND vertically and dremel some holes for the threads? That way you could eliminate a lot of excess tubing there, too.

What are your temps though? You just said you were not pleased with the flow, but if the temps are alright then who cares? ^^

It's a 3x120 on the back and a 1x120 on the top.

water temps stay at 34c.

[Vapor]

Wow, .8LPM is quite low. And if you're seeing a 7c increase in temps when you lower flow just a little, that says you could could definitely use more pumping power.

That said, I don't see anything horribly restrictive in your loop. Supreme HF isn't bad, MCR isn't very restrictive, the Koolance rad shouldn't be too bad, DD GPU blocks are notoriously low restriction, and while the board block is restrictive, it's not a flow killer. To get those kind of flowrate numbers you either need like a dozen pretty restrictive components (or like 2 dozen Supreme HF's) or a bona fide flow killer.

Is your PMP-450 one with a red adjustable trimmer on the back? My hunch is that your pump isn't turned up all the way.

[Xel_NaGa]

Wow, .8LPM is quite low. And if you're seeing a 7c increase in temps when you lower flow just a little, that says you could could definitely use more pumping power.

That said, I don't see anything horribly restrictive in your loop. Supreme HF isn't bad, MCR isn't very restrictive, the Koolance rad shouldn't be too bad, DD GPU blocks are notoriously low restriction, and while the board block is restrictive, it's not a flow killer. To get those kind of flowrate numbers you either need like a dozen pretty restrictive components (or like 2 dozen Supreme HF's) or a bona fide flow killer.

Is your PMP-450 one with a red adjustable trimmer on the back? My hunch is that your pump isn't turned up all the way.

yeah its set to 5 the max setting

its roughly 10 feet of tubing in the entire system

[Martinm210]

For temp improvements, you might consider making some sort of heat deflector for that PSU. While the bottom mounted PSUs are nice for wire management, they are a terrible place for rear externally mounted radiators. Heat from the PSU is likely sucking right into that rear mounted rad.

That is surprisingly low flow rates. Not completely sure why, but two pumps should help also.

The length of tubing has relatively little effect. Just one of your 90 degree elbows is equal to 4' of tubing for example, but that should amount to much either.

[Xel_NaGa]

As a side note before my upgrade I was at 2LPM

the difference is the old system had 3 GPUs (2 now) but I also did not have the 120 rad on the top.


also the water in the loop is 34c
the water after the CPU is 38-42C underload

[Vapor]

If all you did was swap in two GPU blocks and add a low restriction radiator, maybe the DD GTX580 blocks are extremely restrictive :eek:

But let's rule everything else out first...there's also the possibility you have a big air pocket in your 120 rad and that's bringing flow down to a trickle. Can you move your case around in large sweeping movements while its running (tilting it on all its sides)? If any air is trapped in that rad, it should come out by doing that.

[Waterlogged]

Hm, I see a old Koolance D5 top in your build log and it looks like your using FI/FO, if that's the case, we just found where most of your flow went to. ;)

[Xel_NaGa]

so its the koolance pump top is my problem? or my port choice on that pump top?


I just did a 1080 with the case and there was about an inch and a half of air in the loop didnt increase flow rate much.

[Waterlogged]

Port choice is the biggest culprit. D5 tops don't boost performance anywhere near as much as DDC tops do, but if done wrong, they'll certainly kill pump performance in a heartbeat.

[Vapor]

Inch and a half = water level in the reservoir or a 1.5" long bubble in the tubing?

1.5" of water in the res is a big air pocket :eek:

The pump top and inlet choice aren't great, but the inlet choice mainly effects the Max Flow end of the PQ curve, and you're definitely at the Max Pressure end, so that shouldn't matter. Maybe the pump top is that bad though? Is it the metal one you have?

Although, going from 2LPM to .8LPM requires multiplying your restriction by ~5x.....which I don't think happened with the MCR-120 and the GPU block swap.

[Vapor]

Okay, there is a way the GPU blocks could add that much restriction: the fittings (plugs and barbs) are threaded way too long and the flowpath at both inlet/outlets is no more than 1-2mm tall.

Specifically thinking of the plugs here, what G1/4 plugs are you using on the DD blocks?

EDIT: looking at the photo more, I see a temp probe (I think) and non-BP plugs. BP and Koolance plugs have much shorter threading (Koolance especially) than normal and swapping to those would probably impede flow less (if, in fact, that's the issue). I'd remove the temp probe and switch to BP or Koolance plugs where you have the hex-shaped plugs.

By using longish plugs, you basically closed the flowpath. Twice. Maybe a third time with the temp probe too.

[Xel_NaGa]

Pump top used (http://www.koolance.com/product_archive/product_info.php?product_id=734)


EDIT: I'd like to add that I can improve flowrate by kinking a hose for about 1 second and releasing

flow is sustained untill the pump is turned off

increases to 1.2 lpm

[Waterlogged]

Okay, there is a way the GPU blocks could add that much restriction: the fittings (plugs and barbs) are threaded way too long and the flowpath at both inlet/outlets is no more than 1-2mm tall.

Specifically thinking of the plugs here, what G1/4 plugs are you using on the DD blocks?

EDIT: looking at the photo more, I see a temp probe (I think) and non-BP plugs. BP and Koolance plugs have much shorter threading (Koolance especially) than normal and swapping to those would probably impede flow less (if, in fact, that's the issue). I'd remove the temp probe and switch to BP or Koolance plugs where you have the hex-shaped plugs.

By using longish plugs, you basically closed the flowpath. Twice. Maybe a third time with the temp probe too.

Good thinking on the plugs but I think those are DD's plugs. Sensor plug might be causing some of it though.


Pump top used (http://www.koolance.com/product_archive/product_info.php?product_id=734)


EDIT: I'd like to add that I can improve flowrate by kinking a hose for about 1 second and releasing

flow is sustained untill the pump is turned off

That is definitely not normal. I'm starting to wonder if it's the flow meter itself.

How come you didn't use the GPU SLi fittings you bought?

[Vapor]

Flowmeter might not be dead-on....but a 7C increase in temps when setting the pump on 8 (instead of stock 10) says he's already at super low flow :(

That said, the pinch then spike is a weird behavior :wth:

EDIT: at this point, I'd definitely say a 2nd pump will help (there is a lot of restriction in your loop, even once this issue is sorted). Also replacing the EK/DD plugs with BP/Koolance (I'd go Koolance as they're the shortest threaded ones available) is probably worth it as well. And the pinch then spike is making me think there's more air in the loop (air being compressible).

[Xel_NaGa]

Thanks for the advice guys. all the more reason to spring for this new res (https://www.koolance.com/water-cooling/product_info.php?product_id=2025). Plus it will add a layer of redundancy for my equipment

[Xel_NaGa]

well long story short the koolance bay res is worthless due to the inlet of the res being right near the top pump constantly sucked air from turbulance... also was a pain to drain....sent it back... got ek dual pump top....still had :banana::banana::banana::banana:ty flow....started to just eliminate blocks.... turned out to be ek supreme hf... went from 1.5 lpm to 8lpm.... culprit was gray gunk and hair lodged in jet plate... go figure no idea where it came from and showed no signs of growth.. was simply tangled within the flow plate the rest of the block was spotless.

[MagisD]

well long story short the koolance bay res is worthless due to the inlet of the res being right near the top pump constantly sucked air from turbulance... also was a pain to drain....sent it back... got ek dual pump top....still had :banana::banana::banana::banana:ty flow....started to just eliminate blocks.... turned out to be ek supreme hf... went from 1.5 lpm to 8lpm.... culprit was gray gunk and hair lodged in jet plate... go figure no idea where it came from and showed no signs of growth.. was simply tangled within the flow plate the rest of the block was spotless.

Dust dirt plasticzer take your pick, I found if I didn't rinse all my tubing well I got that in my blocks. primochill clear lrt.
Tended to gunk blocks and 90s
Sent from my X10a using Tapatalk

[antiacid]

well long story short the koolance bay res is worthless due to the inlet of the res being right near the top pump constantly sucked air from turbulance... also was a pain to drain....sent it back... got ek dual pump top....still had :banana::banana::banana::banana:ty flow....started to just eliminate blocks.... turned out to be ek supreme hf... went from 1.5 lpm to 8lpm.... culprit was gray gunk and hair lodged in jet plate... go figure no idea where it came from and showed no signs of growth.. was simply tangled within the flow plate the rest of the block was spotless.

well that's a relief, at least now you know it's not a weird problem ;)