My DDC is dead :(

JonJ678

JonJ678

JonJ678

Soldato
Joined
22 Dec 2008
Posts
10,369
Location
England
A while back I posted a thread wondering about the consequences of running two dissimilar pumps in series, and went ahead with it on the basis that it should be fine. I was persuaded that 10W + 18W was a stupid idea when some tubing collapsed, but two slightly different 18W pumps hasn't gone significantly better.

As a follow up to this, one of them is now dead. I don't think this is convincing evidence that running pumps in series is a bad idea, but it has certainly put me off. Parallel loops appear to be the way forward.

Current temperatures on a gentle overclock are 60 degrees idle, 100 degrees load on the processor. 40 degrees idle, 45 load on the graphics card. This room is cold too. It's not stable anymore either, sad times indeed.
 
Sorry to hear that Jon. 60/100 and 40/45 sounds very wrong if it's the same loop.

You are not the first to find dissimilar pumps with kill each other - many watercooling gurus have tried and failed in that.

60 idel and 100 load sound like when I tried to get an early Q6600 to fold on my Reserator 2.
Thats wasn't stable either - but the steam comming out of my res was intresting :D
 
Last edited:
I have seen this regarding DDC pumps

http://www.xtremesystems.org/forums/showthread.php?t=259508

I also remember seeing a thread about running two pumps with the double headed pump top (XSPC & EK make both but cant link)

Did you have the pumps joined together or were they in different parts of the loop?

Also the temps are very high.
I would set everything back to standard if you are overclocking.
 
Cheers SS. I was certainly warned that dead pumps were on the cards, but there's nothing like first hand experience. These were separated by a couple of inches of tubing, wrapped in wire to persuade it not to flatten. Will have to have a think about where to go from here.

Temperatures are a touch unusual. My graphics card was at 40 degrees when I rebooted the computer twenty minutes ago or so, since then I've been running Intel burn test. The processor has made if from 60 back up to 100, but the graphics card is still at 40. So I'm concluding that my radiators are fine (they're a thick 240, two thick 120s, and a thin 120, dealing with about 250W). However the flow rate through the loop is abysmal, and the cpu block is an ek supreme. 60 degrees between water temp and processor is a bit harsh, but it's not impossible.

Things aren't quite back to stock, but it is presently testing at 3ghz. If I load the defaults gigabyte can't cope with 12gb of ram, leaving it unstable until I pull some out even without temperature issues. It's good advice though, very hot + overclocking is a foolish combination.
 
I started off with loop -> 18W -> reservoir -> 10W, but changed to loop -> 18W -> reservoir -> 18W. It's the one just after the reservoir that's kicked the bucket.
 
JonJ678 said:
I started off with loop -> 18W -> reservoir -> 10W, but changed to loop -> 18W -> reservoir -> 18W. It's the one just after the reservoir that's kicked the bucket.
Click to expand...

If there was a lot of turbulence in the res, the second pump could have been cavitating (sucking in air). If this has been happening the pump is very noisy & the centre bearing will burn out.

MikeTimbers said:
Has the second one been removed? I would think that a dead DDC would be a considerable blockage in the loop!
Click to expand...

I agree, when I had two pumps in one loop & turned one off the restriction was incredible.
If the defective pump is still in the loop then you will have to remove it ASAP.
 
EK supreme in a low flow loop - that will be why - that block NEEDS a high flow rate.

I've the HF version of the block arriving in a few days - your welcome to borrow my xspc edge untill you sort out your pump issues.

Four rads (inc a thin one) will also be killing your flow if you do decide to swap out the block lose the thin 120 as well.

As the others have said if you haven't already remove the dead pump.
But based on the 40/45 temps of your graphic card I'm guessing you already have.
 
RIP DDC. This is perhaps the one reason why I will not use 2 pumps in serial. It's all very well having a redundant pump incase one dies but the effect the dead pump can have on the remaining pump can be terminal.
 
sorry to hear

i have only ever ran a pair in series once.

they were both 10w purchased at the same time to ensure that they were identical.

i just use a D5 now as it is a lot easier.
 
The other pump's still present, although it is now unplugged. It probably is doing rather bad things for the flow rate, and worse if parts of the bearing have disintegrated my block and radiators may be full of ceramic chips. I somewhat unwisely left the loop running overnight, it is at least stable at 3ghz, 100 degrees so work can continue, albeit slower than before. I didn't think that one through, it'll be out by the end of the day.

I'll need to check into causes and means of preventing cavitation, but I don't believe this was the problem. The reservoir was full to the brim, as was most of the T line I used to fill it. I also couldn't hear it running. However it was sitting in a corner far away from any form of airflow, and somewhat surrounded by wires. I think the poor thing overheated, in which case it's colleague may be on its way out as well.

Thanks for the offer SS, but I'll decline. There's a good chance my waterblock is stuck to the cpu, as I put a lot of effort into introducing two very flat, clean copper surfaces together with only liquid metal inbetween. I'll be surprised if that's not a solid block by now. Pulling it apart may make things considerably worse.

I'm probably moving to a D5 after this, though while the remaining ddc lasts I'll use that for the cpu. Fair play to everyone who told me multiple loops make more sense than one very long one.

Cheers for the kind words :)
 
Last edited:
np Jon, I'd forgotten you got some gallium to play with - remember rotate to remove

Supreme, DDC 18w +top and thick 120.2 are a perfect combo
your load temp should be well below your current idle with that setup.

Every cloud and all that - worth a good flush first, just in case, but ceramic chips usually wake the dead if you have them in your loop.
 
Last edited:
Did the DDC that died have a top on it? There is some theories that some tops might have an adverse affect on reliability on 18w DDC pumps.
 
Whats the theory behind running 2 pumps in series??

Someone mentioned above as a redundancy incase a pump fails, but surely in that instance you would be better running the pumps in parallel to each other? That way there would be no blockage in the line if a pump failed.

And you would still be increasing the flow rate by having the pumps in parallel.
 
Two pumps in series roughly doubles the head pressure, allowing better flow rates through restrictive loops. Parallel doesn't work out as well.

Anyhow, as an update, I've finally taken the loop apart. The pump isn't dead. The molex connection is somewhat knackered, but the pump itself appears to run as normal if the loose connection is carefully supported. It looks like either soldering iron or crimping a new plug on the end will fix it, which is good news. Thoroughly embarrassing though.

Less good is that the newly simplified loop, which now has a (definitely) working pump, cpu block, three radiators and a chipset block in, still runs the processor hot enough that it shuts down. The system is cold to the touch, including the cpu waterblock and the radiators. However cpuid reports 75ish idle and 100 under load, briefly, before the system turns off. I'm fairly sure it's the processors thermal shutdown.

I think this means the mount is bad, the waterblock is knackered or that it's full of algae. The mount is the same one that achieved 60 degrees a few months back, the waterblock has been unwisely modified, but again it was at 60 degrees before, and the water looks clear. My best guess is that something terrible has happened between the processor and the heatsink involving liquid pro, so the whole damned lot is going to need to come apart.
 
Could the system still have any air in it.
Does it make a lot of noise, like a drain?
or does the level in the res increase when the pump is turned off = air
 
piggin: nah Jon's not likely to miss that.

Jon I reckon you are on the right track, logically if the CPU waterblock is cold but the CPU is overheating then its either the TIM/fitting, faulty chip sensor or you got some contaminant/damage on the pins perhaps.
 
I'm with biffa - my guess would be the supreme jet plate.
The fixing bolts are on the top (just checked) so you should be able to clean it with the cpu still welded to the base plate.

I had the early stepping q6600 on a passive loop and it did the same - flow rate just wasn't up to the job
 
Biffa said:
piggin: nah Jon's not likely to miss that.

Jon I reckon you are on the right track, logically if the CPU waterblock is cold but the CPU is overheating then its either the TIM/fitting, faulty chip sensor or you got some contaminant/damage on the pins perhaps.
Click to expand...

I just thought I would do the obvious first @ that time of the morning. :)

The next thing is go with what has already been said regarding the 'gumming up' of the components.
It is possible with a slower flow to get the additives to 'fall out' of the fluid.
As for horror stories, just Google feser one gunk. (i know that you are not using feser but it could easily apply)
 
You must log in or register to reply here.
Back
Top Bottom