New ddc in series with old, knackered one. Worst case?

JonJ678

JonJ678

JonJ678

Soldato
Joined
22 Dec 2008
Posts
10,369
Location
England
Title pretty much has it covered. I've got an 18W laing ddc which is running beautifully. It's about to run quite a restrictive loop, ek supreme and quick disconnects being the main issues.

Now, flow rate probably doesn't matter very much in water cooling. But in case it does, I'm planning on putting an old, probably weaker (certainly noisy) 18W ddc in series with it. The idea being that both pumps running at once may get better cpu temperatures, and if they do, I'll leave both on, if not, I'll take one out. However I believe running dissimilar pumps tends to kill the weaker one and distress the stronger one.

So, is the worst case scenario that the old pump dies? I can't really lose if so. I think the danger is in the stronger pump starving the weaker one of water, if so then I reason reservoir -> strong pump -> weak pump should avoid this. Would (could?) there be a benefit to isolating the pumps from one another, putting something like a radiator inbetween, to avoid nonlaminar flow from one interfering with the next?

Cheers
 
I think I'm right in saying there can be issues running 2 pumps in series near to each other in the loop. Bit vague I know! - probably worth googling/looking on xs.

The EK supreme is very temperamental about flowrate from what I've heard - have you plugged the bits into the flow estimator to get an idea of flowrate with one pump?

The other alternative is you could get a dual ddc top, though not sure what effect the weaker pump might have, and a bit pricey... Have you verified it's a bit iffy - does it pump less effectively or just make more noise? Could just be the ceramic bearing is going on it.
 
What are the factors which would make two pumps of the same brand, make and notional rating behave differently? I'm thinking that the dimensional accuracy of the casing will be virtually identical between the two so that could be discounted, assuming whatever top is attached is also dimensionally sound. That leaves the dimensions of the impeller and the speed of the motor.

The dimensional tolerance of a new impeller is an unknown but I would guess they are all made in batches from a injection moulding process so you could argue they should all start off the same. The second question is do they actually wear in your average pc water cooling system? Again and unknown AFAIK but my gut feeling is it would be miniscule unless you have dirty water, i.e. didn't clean your rads out. May be a problem if the pump is second hand and you don't know how it was treated.

That leaves the motor speed, which to me is by far the biggest variable affecting pump performance. The question is will the two pumps both run at the same speed if run off the same 12V molex? I know more about AC motors than DC ones but I guess there will be similar factors in the electronics in the motor which may vary pump to pump which would give a different final speed to the load. And in my opinion matching the speeds is what you want to bring the pumps closer in performance.

To this end what I would do is build the loop but put only one pump in it and measure its speed using the mobo speed wire. Remove that pump and put the other one in and measure its speed. If they are the same put them both in and make sure they are still running at the same speed. Job will be a good un if they are I would say.

If there is any difference in speed then you could perhaps use a potentiometer to adjust the speed of one to bring them together. The only downside is you are relying on the accuracy of the mobo speed sensor, which I guess you could at least test for repeatability to gauge confidence.

If it was me though, I would use the second pump for a second loop, even if it meant running it as a T line so that a second res is not required. I know you are restricted by your rad setup however.

As for running pumps next to each other, I'm sure I've heard elsewhere that it is better to have them next to each other, just to directly contradict MP and everything :)
 
Thank you both. The old pump throws out a lot of water but I haven't measured it or compared it to the new one. It's probably a good idea to try to measure their flow rate as well as their speed (good thing I haven't cut the rpm wires off). Both good ideas, if flowrate and rotational velocity are about equal I imagine it'll be fine. Head pressure would be a good thing to know too, but I don't think I'll be able to measure that one.

Martin's sheet thought my flow rate was poor even before adding quick disconnects and another block (or two) to the loop so I think it'll be quite rude about the current set up. The ek supreme is rather fond of pressure, though with the impingement plate removed (and possibly lost by now) performance at low pressure can still be good.

Wear in the pump may be significant, it arrived thoroughly full of dye deposits. I'm hoping they were a large part of the noise source, and it's actually fine once cleaned. To that end it spent last night in vinegar and is looking much happier for it, so fingers crossed.

Two loops might actually make sense now I've moved to two 8800gt's, a 120mm radiator dealing with the graphics cards while a 240 and a 120 deal with the processor may well be better. That I think I'll have to test.

Thank you both, much food for thought.
 
Hmmm, I've just purchased another 2nd hand DDC to add into my loop when I rebuild it again, but I'm a little bit worried now after reading this thread..

I will be running both of mine fitted to a Watercool dual DDC top like this one:

d1622101.jpg


The first pump I have was just a 10w one, but I modded it so it runs at 18w, and the 2nd one is an 18w one, but it's fairly old by the look of it, and it has a red impeller whereas the other one is blue.

Now I'm not so sure I should run both of these pumps together, just incase something goes wrong.

Would a single 18w pump deal with a quad and double rad (feser double and XSPC quad), Heatkiller 3, 2 x EK FC-GTX280 blocks, and an EK multioption 250 res??
 
I suspect you'll be fine Wolvers, but hard to be sure.

My plans of two in series are aborted anyway, I have an 18W one which works well and a 10W one which I just failed to turn into an 18W one. I'll find a use for it I'm sure, but not in series.
 
Right let's think about this properly physically wise not theoretically. What in the pump is going to break? The rotor in the motor, how can this be broken. The fins, no they will not snap. The bearings, no more force will be put on them.

Now let's look at what would happen with both and one in! With the stronger one it will not be rotating at max? Yes? And so the same can be said for the weaker of he two because there is restriction in the loop. Now with both in the weaker one will spin faster as the stronger one Is also pushing water and vice versa.

At the end of the day if the weaker pump can push as much water with the stronger one as the stronger one on it's on then the weaker one can't be slowin the faster one down can it? And so no more forces on them and so no risk of damaging both pumps. If the case were the pump wouldn't be able to keep up with the faster one then it'll be slowing it down and creating more restriction than adding flow.

I think that they would be absoulutly fine and what you've heard is a load of old women trying to make problems. On the othe hand I would do two seperate loops as that is the proper way to do it!
 
^There are no bearings in a DDC...and cavitation can damage rotor blades relatively quickly.

Jon have you taken the pump apart to investigate the cause of the noise?
 
There definitely is a bearing, it's a little ceramic ball which the impeller sits on. Or at least this is a bearing surface.

I think most of the noise was dye build up, it's a lot quieter now it's clean though louder than a new one. I should take a closer look at the ceramic really, but having cocked up the 10W to 18W mod I'm unlikely to try them together anyway.
 
Ah sorry you are right there is a ceramic ball bearing, just looked at their site.

Anyway when I had my DDC it began making noise and it was due to dye build up as you said. Once I cleaned it it was fine.

How did you mess up the 10W to 18W mod?
 
Lord knows. I'm very new to soldering. Managed to get solder on the copper tab but not bridge the gap, then when removing this solder to try again it pulled the copper tab up. Had a go at bridging to where it looked like the tab came from, but no deal.

Don't think its recoverable :(
 
It probably is recoverable. A pic would help. I'm no soldering pro and have tons of bad habits since I've taught myself but have been doing it for ages (built a walkman from a kit in 1993 when I was 9 and have been soldering ever since).

Sounds like you may have had the temperature too high as the adhesive between the pads and pcb can't take higher temps usually.
 
Interesting. I'll get a photo up within an hour or so and see what you think.

An hour late I'm afraid.
1z6fb7n.jpg


My best guess is scraping the lacquer off one of the traces and soldering direct to that, but its hard to tell which trace
 
Last edited:
I thought I'd seen somewhere that they'd changed the design of the the ddc pumps nowadays so that this simple solder mod no longer works. Maybe all they've done is make it harder to solder.
 
I'm confident this would have been trivial to solder if I'd been less rubbish. I've also heard this mod doesn't work with the newer models, but this is an old one and all the photos of the mod I found online showed a version 3.3 pcb which mine was. Unfortunate :(
 
It seems to be fine incidentally. 10W and 18W in series, not adjacent. Improvement to temperatures and neither has died yet. Considering a second 18W, to move to 2x18W and 1x10W.

That said, I'm getting tubing collapsing in front of the weaker pump. So it's pumping water out faster than water can get in. Weird that it's the weaker pump with the problem, but then the stronger one has a reservoir on top. Starting to look like series pumps aren't that good an idea, I think I'm going to change to multiple loops.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom