dual pump any good?
 

Hi All,

I'm working on my new setup. One of the things to say is that I'm using AS5, and it's only 80 hrs in place. (according to the web site, it can take 200 hrs to settle)

Here is my issue:
I'm using a dangerdan double heatercore, one dangerdan single heatercore and a BIX (all in series). My waterblock is the Witewater, the DD Z-chip and modded innovatec GPU. I'm also using the DD bay-res.

My pumps are 2 times the MCP600. If I turn one on, my temps are (according to my motherboard temp diode IC7Max3) 37.5 degrees under full load.
This is with one pump.

If I also turn the other pump on, I can see that the water is speeding up a lot. (buble movement in the bayres) (not that there is air in the water, it's just one bubble at the top of the res)

According to all the theory, speeding up my water should give me lower temps.
my temps stay exactly the same! 37.5 degrees.

What could be wrong here?
Could it be that there was some small chuncks in the rad e.g. that are now clugging the WW? (but I did flush the setup prior to filling it) is it possible that could be responcible for not having any temp difference?

Based on Phaestus's pump comparison article, I thought I'd try to estimate how much flow you are gaining from adding the second pump, and then how many °c you might stand to gain.

The system you are using is probably somewhere in between PH's low restriction setup, and the high restriction setup. Lets do an average and call it a "medium restriction setup".
Putting two MCP600's in series pretty much doubles the max pressure of the pump.
The two rings on the graph show the difference between a one-pump and a two-pump setup. In this case you're going from 5.3 to 6.8 lpm (very ish). That’s 1.4 to 1.8GPM (+30%)

Now, looking at PH's test of the DTek White Water, we see that increasing the flow rate from 1.4 to 1.8GPM only yields 0.5°C. (!)
(not even considering added pump heat, although this probably wouldn’t be very much at all for your system)

Considering the limited resolution of on-board temp sensor readings, maybe this would explain why you see no difference in temperature:)

is this in series? (and if so where in the loop?) or in parallel.......

hmm - i'd have thought that 3 rads and 3 blocks in series would be high restriction...
3.9lpm to 5.1(ish) - about 1gpm to 1.3 which would suggest slightly more than 0.5C (since we are higher up the performance curve...) also the rads should be slightly better than before - but also the zcip and gpu blocks should perform better, so iiuc a bit more heat would enter the loop from there...

fhorst - are the pumps in series?

also have you tried putting 1 pump before and 1 pump after the CPU block, one pushing and one pulling... ?

the pumps are is a "push/pull"config.
setup:
bayres - MCP600 - heatercore - BIX - MCP600 - double heatercore - WiteWater - (splitting loop) one to the Z-Chip, one to the GPU, back to bayres. So the bayres has 2 in and one out (to MCP)

The speed of the water is good with 2 pumps, I can see it clearly speeding up, so I guess 3 rad's makes it a restrictive setup., giving with a second pump a 2lpm gain.

The little gain could indeed be that it only gives me 0.5 degree extra...
Hmmm... bommer

As I'm still not even close to the 11 lmp that the pump is capable of in free run, there is no use to place them in paralell. I guess placing the BIX and the heatercore in paralell could do some good, making the setup less restrictive.

One thing not so clear to me: I was guessing that my tygon 1/2"would feel solid with all this pressure, but is't not. I still can squeeze it to flat without problems...

well I guess I need to redo some of the plumming work :(

Thanks for the input!

"setup:
bayres - MCP600 - heatercore - BIX - MCP600 - double heatercore - WiteWater - (splitting loop) one to the Z-Chip, one to the GPU, back to bayres. So the bayres has 2 in and one out (to MCP)"

a heatercore, an Xtreme, and a double heatercore... 3 radiators in series, eh?
OVerkill...

I'd say put the 2 cores in parallel for the CPU with only 1 pump....
and create a separate loop with the BIX and the 2nd pump for the GPU and the Nbridge.

or if the main reason of having 2 pumps in series, is redundancy, then i recommend you to use only 2 radiators in parallel, and get rid of the 3rd one.

First, after the 1GPM point, changes in flow do not change the temperature by a large amount. If you'll look at pHaestus' graph, you'll notice a slowing in temperature drop. This is more noticeable in BillA's graphs, but whatever. Second, you have increased pump heat. Only 6W more, but more nonetheless. Third, your ambient may have changed, or your temp measuring thing may not be very accurate.

Overkill? no way! I run my fans at 7v, to keep the noise down (PAPST), so the air flow is not that much. only way to get water temps as close to ambient as possible, is to increase airflow and/or cooling surface. As long as the water temp is above ambient, it's never overkill. every degree you get out of the water, will give you lower CPU temps. As simple as that.

I placed the BIX and the heatercore in series, as that is the best setup for getting rid of temerature. I have to see if making them paralell, with the decreased flow will give me the same temps. It might be that the overall speed will increase, so the temps may stay the same.

I have been thinking about 2 seperate loops, and still using the bay res. This would mean that the water still is mixed, so the overall setup temps won't change(?). My tubing will be longer......... (twice) as I need an extra line from the ras to the GPU/NB blocks

And yep you are right, for me one of the goals is redundency. I was also thinking I would get a bit better temp it I turn the second pump on, but this seems not to be the case. :cry:

or to increase the water flow..... higher flow helps the rad(s) as well as the blocks...... (though not as much)

plus it'd be worth trading a 1 degree increase in water temp, for a 1.5 degree improvement in (cpu temp - water temp)

maybe it'd be worth loosing a rad (bix?) altogether - then possibly the other fans could be a bit higher for the same noise level...

i guess you'd really need accurate water temp and cpu temp measures to predict the best move - bios temps aren't accurate enough to predict what part of your setup is likely to be the best target for the next half a degree...

just a thought but what is the airflow like for these rads? i.e. where are they and are they using outside air or case air? shrouds? fans pusing/pulling?

well, I did a big makeover at my plumming.
Got rid of one of the MCP's and GPU block, placed the Heatercore and the BIX in parallel

this is how it is now:
bay-res - pump - whitewater - one outlet to NB after this to the heatercore and back to bayres.
Other outlet to double heartercore, after this the BIX and back to the bayres.

My temps dropped one degree :) (did not have to move the WW from the cpu)

My GPU is now in my alu loop, but it ain't all allu anymore :( I used one innovatec GPU block.
p.s. my alu loop is powerd by a koolance EXOS and one small modded pump, and contains:
- 3x HDD
- 4x GPU memory
- 2x System memory
- South bridge
- Silicon SATA chip
- Mosfets
- backside of mobo
- and now also GPU
it's running at 5mm id tubes :)

for the position of rads question:
double heatercore (2 x PAPST 12cm fan) is in the botem front, where the HDD supposed to be.
Heatercore and BIX ( 1 x PAPST 12cm fan) are stacked making a 10cm thick rad at the top of the case.

i'm certain that loosing the BIX would be a good idea then - i doubt a single pabst can draw much air through 2 rads one after the other... you could use the depth freed up for a shroud...

LOL, there is a shroud........ the nice plexi one :)
And it moves just enough air. If I feel it's moving not enough air, I can place a shroud at the back of the case, with one aditonal PAPST, making it a push - pull setup.

all the air that goes out also must go in.... I also have one additonal rad for my alu loop, sucking air out the case with 2 x 8cm PAPST, and I have the NEXUS PSU, who has a 12cm fan.

Letting the fan in the top sucking out to much air, will make that I'll out of balance.

by adding another pump, you also add the heat from the extra pump into your loop :)
in your case, it appears that the lowered C/W of your system with regard to flow is compromised by the added heat output
in short- the extra pump is a waste of money if you plan to continue with the current application- use it for something else, say a separate loop for the nb and gpu blocks

MY GPU is now in a seperate loop :) that's 40w less heat. the pump is only 4w......

I guess losing the GPU was the biggest bang :rolleyes: