flow rate weirdness

CPU water cooling blocks...
erratic
Water Pistol Grunt
Water Pistol Grunt
Posts: 8
Joined: Tue Oct 11, 2011 1:53 am
Contact:

flow rate weirdness

Postby erratic » Sun Aug 31, 2014 1:28 am

I have a mystery for you.

I was swapping out my GPU so did a complete drain, dissassemble, and clean. Which is when I found out that all my Koolance QDCs have stopped sealing closed - "no-drip" ha!

Anyway, everything looked clean - no slime, or corrosion. All the 370's fins were straight and clear, and the O-rings were ok.

I reassembled the loop without the rads for testing:
pump2/res/pump1/cpu/gpu/flow meter (using the bypass top left)

Flow rate was above 8lpm and everything was fine. Initially.
(As a note, using the bypass piece in the wrong place does affect the bleeding ability of the 452x2.)

So far so good. Commence weirdness!
Over the next 5 minutes the flowrate steadily decreased down to 1.9l/m. It's not an issue with the meter, the flow rate is visibly less.

I have a spare pump, and plumbing that in between the GPU and flow meter did not improve flowrate, which is nuts considering there were now three D5s pushing away at full power.

However, swapping its direction and running it alone pumping the loop in reverse gave me a great flowrate. Can't say exactly since my flow meter doesn't run backwards, but it was excellent.

Turning that reversed pump off and going back to a single internal pump (doesn't matter which) and I'm back over 6 l/m again, but it steadily drops down towards 1.9l/m.
Running the reversed pump even for just 5 seconds is sufficient to "reset" the normal flow rate, but it always decays back down to 1.9l/m. It's repeatable and from historic performance I can expect that to be down to 0.4-6l/m in a month or so.

With no rad connected, and with a new GPU block, it has to be either the cpu-370 or the 452x2.
The question is, WTF? Does anyone have any ideas?

erratic
Water Pistol Grunt
Water Pistol Grunt
Posts: 8
Joined: Tue Oct 11, 2011 1:53 am
Contact:

Re: flow rate weirdness

Postby erratic » Tue Sep 02, 2014 2:57 am

As expected, by the afternoon flow rate was down to 0.9l/m and 0.6l/m in the morning. That's after tapping bubbles out of the flow meter to get it turning...
There's also a bubble trapped in front of the GPU block fins which badly impacted temps (46C->94C badly).

So while people are right to say flow rate makes very little difference to temps, this is too low to bleed. Bubbles are not even moving against gravity.

I guess the next step is a partial drain to hook the CPU block up backwards to isolate the weirdness to either the CPU-370 or the 452x2. Oh for non-leaky QDCs!

erratic
Water Pistol Grunt
Water Pistol Grunt
Posts: 8
Joined: Tue Oct 11, 2011 1:53 am
Contact:

Re: flow rate weirdness

Postby erratic » Tue Sep 02, 2014 11:59 pm

That wasn't too painful and reversing the flow through the CPU block may have sorted things.
Flow rate is up from 0.9 to 6.2l/m. That had dropped to 5.8l/m overnight so we'll see how it goes over time.

CPU temps are negatively affected by a few degrees, but still under 60C with a 30% OC and as this chip has problems at an indicated 76C, it's fine. GPU temps are down a few degrees, and the delta between water in/out to the rads is down from 5 to 1 degree.

If it hasn't dropped any further by the time I get home tonight, I'll call that a win.

Still pretty weird behavior. When I disassembled the block everything looked ok. The fins were clean and perpendicular to the jet plate and the jet o-ring, while a bit stretched and a bugger to get back in the groove, was in the correct place.


Return to “Water Blocks - CPU”

Who is online

Users browsing this forum: No registered users and 1 guest