Dual Pumps - Best Configuration

Dirk Diggler

Dirk Diggler

Dirk Diggler

Soldato
Joined
6 Jan 2013
Posts
22,230
Location
Rollergirl
I've got a dual pump configuration:

Res > Pump > 480 Rad > 240 Rad > 560 Rad > Res > Pump > GPU > CPU > 240 Rad

One of the pumps started making a strange noise:


So I tried to adjust speed and then finally rebooted PC and the noise went away. About an hour later, and I do not exaggerate here, there was a sound like a sander being used somewhere in the distance. It turned out to be the pump again:


I'm in the process of replacing the bust pump, but I'm wondering if my loop configuration has caused it to fail? I ran the bust pump at 100% constantly and I ran the other pump at 20%. This is because the 100% pump is in another room (garage) and it means all the noise and heat is in there.

Should I run both pumps at the same (lower) speed?
Are pumps designed to run 100% 24/7?
 
Last edited:
I would have thought maintaining a more comparable % would mean a less stressful life then running one at 100% and the other at 20%. Is the one in the garage also free of air pockets and getting constant supply of water as they are watercooled?

I think I have a broadly similar loop to yourself, a external radiator away from the PC with a pump in between and then the PC itself. For me I found the EK D5 pump to work flawlessly: https://www.overclockers.co.uk/ek-w...al-d5-pwm-serial-incl.-2x-pump-wc-945-ek.html combines high flow rate of a D5, but you get twice the head pressure with both D5's running. For me just set the pumps to 55% and that is enough to go through 3 radiators, 3 GPU's a CPU and a bunch of 90% fittings.
 
The pumps are water lubricated - I would have expected that kind of noise without a second Res.

It's nothing to do with the orientation of the pump is it?
 
Everything I have read states pumps should run at the same speed other wise the lower rpm pump is actually acting as a drag and ruining flow rate.

The faster pump has to work harder to try to compensate.
 
I have two pumps in serial next to each-other in the loop. They seem to do the job quite well, even though they are different pumps. Swiftech MCP35X and Swiftech MCP655.
 
String said:
I've got a dual pump configuration:

Res > Pump > 560 Rad > Res > Pump > GPU > CPU > 240 Rad

One of the pumps started making a strange noise
Click to expand...
Not sure if related but you should remove the highlighted res as it's of negative benefit.

*EDIT*

To clarify, there's two common ways to run dual pumps, series/serial and parallel. In series the pumps are in line one after the other (may have components between) and this causes them to combine their pressure which is great for restrictive loops, this is the most common method today. In parallel the pumps are side by side and the loop splits into a Y to feed them then recombines after, this method combines the flow rats to greatly increase the amount of liquid flowing through the loop (not much of an issue for most loops these days).

What you have, essentially, is two separate loops which are linked so that they fill each others reservoirs. This has no benefit to pressure or flow, the only "benefit" would be that temperatures between the two loops normalize, however the second res is also adding heat to the loop so overall it's a net loss to performance. Removing the second res from the loop would allow the pumps to run in series increasing pressure, flow and and most likely improving temps.
 
Last edited:
ubersonic said:
Not sure if related but you should remove the highlighted res as it's of negative benefit.
Click to expand...

There's a res inside the case and another outside. It's there in case I need to disconnect from the external cooling and still use the PC. That's mainly why there's 2 pumps (one in the case, one outside).
 
String said:
I'm in the process of replacing the bust pump, but I'm wondering if my loop configuration has caused it to fail? I ran the bust pump at 100% constantly and I ran the other pump at 20%. This is because the 100% pump is in another room (garage) and it means all the noise and heat is in there.
Click to expand...

The pump that bust was not only trying to pump the water round the loop but was also trying to turn the other pump faster as well, or in other words the case pump was slowing the flow in the loop and causing the external pump to try to work harder.

I have got an external rad and multiple pumps but each is on its own loop which is combined by using a single large res.
 
K Olley said:
The pump that bust was not only trying to pump the water round the loop but was also trying to turn the other pump faster as well, or in other words the case pump was slowing the flow in the loop and causing the external pump to try to work harder.

I have got an external rad and multiple pumps but each is on its own loop which is combined by using a single large res.
Click to expand...

Do you really think a second pump running slower could stress the faster pump? Surely it would provide resistance just like a water block? If what you say is true, then running two pumps in series would be almost guaranteed to bust a pump as it's difficult to get them both equal speed?

Two separate loops would mean no dual redundancy. If I was going to trust a single pump then I would just do a single loop because I don't see any advantage in 2 loops?
 
String said:
Do you really think a second pump running slower could stress the faster pump? Surely it would provide resistance just like a water block? If what you say is true, then running two pumps in series would be almost guaranteed to bust a pump as it's difficult to get them both equal speed?
Click to expand...

It would increase the resistance of the loop so it would make the faster pump work harder, whether it would make the pump work hard enough to stress it enough to fail I do not know.

String said:
Two separate loops would mean no dual redundancy. If I was going to trust a single pump then I would just do a single loop because I don't see any advantage in 2 loops?
Click to expand...

On my setup I use a dual pump [EK-XTOP Dual D5] for my GPU's, 2 x 360 rads and the external 480 rad, and a single pump for CPU and a 360 rad, the GPU's are setup in parallel. It allows me to control the flow on either loop independently, I used the dual pump in case of a single pump failure on the GPU loop, on the CPU loop I just hope that if the pump fails that the CPU will throttle back before failure.

The external rad was fitted to reduce both loop temps and room temps.

The machine is being used for both Seti and Einstein.
 
I have been using two dual 5.25 koolance rp-452x2 rev 2 reservoirs, both fitted with two D5 vario pumps for both my loops for a few years now.

Never had any issues at all in that time. I set both pumps in each loop to the same speed and the pc runs smoothly and very quiet.
 
ubersonic said:
Not in your system no that's not possible, as your setup is effectively two separate loops with linked reservoirs the pumps aren't technically interacting with each other.
Click to expand...

No mate, it's a single loop with 2 pumps and 2 reservoirs all configured in series.

The configuration is:

Res > Pump > 480 Rad > 240 Rad > 560 Rad > Res > Pump > GPU > CPU > 240 Rad
 
You must log in or register to reply here.
Back
Top Bottom