Thermochill PA120.3 Pressure Drop Test Results

[Xilikon]

mcmaster.com 4860k657

Yes, it's great... It come with a kind of rubber gasket to seal. Mine never leaked till now

[Waterlogged]

Yes, it's great... It come with a kind of rubber gasket to seal. Mine never leaked till now

*sigh* what did you do now? *sigh*

[Xilikon]

*sigh* what did you do now? *sigh*

Sould I say "Never leaked so far" ??

Rereading what I wrote, it didn't make sense

[Waterlogged]

Ah, got ya.

Should have said "sealed first time, every time...still leak free", corny I know, but it works.

[Xilikon]

Hate it when the french bite me in the back...

[Nickel020]

Thanks for taking the time to do this Martin! There are awfully few Watercooling reviews on the net

But I don't understand why you plot flow rate against pressure drop and how you even get the data. Or how to make sense of it (well I know that less pressure drop at a given flow rate is better, but what does that mean for actual performance?). Do use a variable voltage on the pump to obtain the different flow rates and then note down the pressure drop? I guess the pressure drop refers to the pressure of the loop without the rad versus the loop with the rad & fittings, right?

Sorry for all the questions, but I can't make any sense of that graph. The way I would test this is hook up the loop with the different barbs which would give me one flow rate and one pressure drop value for each barb. Then I'd check how that affects CPU temperature. But I must be missing something...

[Xilikon]

No, the varying thing is the flowrate that Martin is adjusting with a valve till the flowmeter indicate the right flow. By then, he mark the pressure drop indicated on the manometer then increase the flow in steps till you get enough data for a pressure drop curve.

[Martinm210]

Yes, I just adjust the flow rate in .5GPM increments.

[Xilikon]

Oh, interesting observation... Too bad my radiator is too busy cooling my loop

[Pedalmonkey]

:p. maybe we'll get marci involved for this question!! the secret is revealed now thermochill!! :P

[Martinm210]

Thanks for taking the time to do this Martin! There are awfully few Watercooling reviews on the net

But I don't understand why you plot flow rate against pressure drop and how you even get the data. Or how to make sense of it (well I know that less pressure drop at a given flow rate is better, but what does that mean for actual performance?). Do use a variable voltage on the pump to obtain the different flow rates and then note down the pressure drop? I guess the pressure drop refers to the pressure of the loop without the rad versus the loop with the rad & fittings, right?

Sorry for all the questions, but I can't make any sense of that graph. The way I would test this is hook up the loop with the different barbs which would give me one flow rate and one pressure drop value for each barb. Then I'd check how that affects CPU temperature. But I must be missing something...

I plot flow rate vs. pressure drop because with that equation you can estimate the flow rate for any system. It's sort of the ultimate tool that you can use. I could simply test a whole bunch of actual radiators on a test loop and give you the flow rate that was produced, but that's only good for that one loop. If you figure out the whole curve, I can then use it to estimate flow rates on any system.

I vary the flow rates in my pressure drop testing with household water on a facet and I control flow rates using a standard sink facet valve. I wouldn't recommend that with a bucket flow rate test because if pressure varies during the filling of the bucket, you're flow rate accuracy isn't very good. But in the case of using a flow rate meter, I get instant readings so I can read flow rates and pressure drops at the same time. The other advantages to using household water is that there is not air in the lines, so bleeding of reciculated loop with a pump is not necessary. Also my water pressure in my house is extremely high, so it provides the ability to test clear out to very high pressures and flow rates without running out of pumping power. My first few trys in pressure drop testing witha D5 made that very apparent when I couldn't get results beyond 2 GPM for alot of blocks....not a problem with the household water.

Regardless, I think it would probably be helpfull to at least estimate some flow rate differences between some systems in addition to the curves. This would at least put better scale to the result differences.

The actual flow rate differences between these radiators will be very small because they are all very very low in restriction.

[Martinm210]

I also just use household water for my source.

[Nickel020]

Thanks for explaining, makes sense now! But I still don't know how I these values translate into real-wold performance.

I plot flow rate vs. pressure drop because with that equation you can estimate the flow rate for any system. It's sort of the ultimate tool that you can use. I could simply test a whole bunch of actual radiators on a test loop and give you the flow rate that was produced, but that's only good for that one loop. If you figure out the whole curve, I can then use it to estimate flow rates on any system.

How do you go about estimating this?

Looks like my sample has a bit of a leak in the dividing plate corners. This would allow some water to bypass the radiator entirely. Can anyone else see the same thing?

This is very common! The German distributor states on it's website that this does not affect performance. Though I don't know if this is true...

[Pedalmonkey]

aww, i wanted some sort of conspiracy theory or something. although i geuss i can just wait and find out if it is.

[Martinm210]

Thanks for explaining, makes sense now! But I still don't know how I these values translate into real-wold performance.



How do you go about estimating this?

You can estimate the real world flow rates with my estimator. What that means in terms of temperature is more difficult, generally there isn't much more than about .5C gain from 1GPM to 1.5GPM on something like the D-Tek Fuzion and then it's about .2C from 1.5 to 2.0 GPM, so we're talking pretty small here. That's why I try to just recommend keeping your flow rate above 1 GPM and don't worry about it much, shoot for 1.5GPM to leave some reserves for more restriction later.

Thermal testing rads is something I havn't done yet, but I'm working on it...hope to give it a try soon.

[Navanod]

It could just be this sample, there's alot of hand work in these things, so it would be easy enough to miss a little solder.

I think when I was making the copper elbows, there wasn't a single one exactly the same....soldering is an art..

Just looked at 2 pieces of PA120.3 with a torchlight and both had light peeping through the corner of the divider plate (the corner further away from the fins)...one even had a tiny bit of light coming from the middle of other side of the divider nearer to the fins (ie. both sides of the divider had a leak)

same goes for the PA120.2 i got...2 leaks in the divider as per the 2nd PA120.3

I cannot see the top of the divider thats closer to the barb threads...i sure hope there are no more shortcuts there!

what gives?! does this means the rad is slightly less restrictive due to this (somewhat similar to how the Dtek fusion was before the additional washer from the nozzle kit was added)?
Does this also means that we would be losing a few watts of thermal dissipation performance?

I would agree with you that this is a flaw instead of a deliberate disign feature since 2 my rads differ in having an extra gap on the side proximal to the fins. Since these gaps are only about half the size of one of the tubes running along the rad, does that mean the leak is negligable (ie, no measurable performance difference)?

[Martinm210]

I wouldn't worry about it....

[Pedalmonkey]

HA, so my conspiracy theory created off of martins hard work an close looking at the actual rad and navanod's confirmation will live to see another day!! woo. wait thill HW labs hears this secret.

[eXa]

Regarding barbs; ive drilled all my ek barbs to 10.3mm and removed all the sharp edges. On the pa tho, i could drill out the barbs to about 12.5mm because they are limited by the neck not the threads. i beet that beats the G1/4 adapter. Tho i want to use 5/8 G3/8 barb if i only can get my hands on 2. And i will shine a torch down my pa when i get home from work...

[Navanod]

HA, so my conspiracy theory created off of martins hard work an close looking at the actual rad and navanod's confirmation will live to see another day!! woo. wait thill HW labs hears this secret.

Maybe we should all start shining a light down whatever rads we happened to find...including HWLab's

[Xilikon]

I think it's not too much of a issue since water will always take the path of least resistance. Logically, it's thru the fins and back but the leaking might account for some lost dissipation power.

However, since water in a loop will reach a equilibrium, even if some water bypass the fins, it will be cooled by the surrounding water or flowing thru the fins on the next pass ( in 10-15 seconds).

[Waterlogged]

I can also confirm that "I have seen the light". It was a little bit worse on my 120.1 than my 120.2 but both had this defect(?). Also checked a P.O.S. GTS and could not see any light, checked my BIP and then realized I was wasting my time with that one.

[Navanod]

I can also confirm that "I have seen the light". It was a little bit worse on my 120.1 than my 120.2 but both had this defect(?). Also checked a P.O.S. GTS and could not see any light, checked my BIP and then realized I was wasting my time with that one.

@ the BIP

hmmm...we may be on to something here? haha

[Navanod]

I just had a strange thought...

Maybe that little leak ain't too bad after all...
if we KNOW EXACTLY where our leaks are between the 2 compartments, we can avoid the pains of trying to make air bubbles do a U-turn along the entire length of the rad, esp if we're not mounting the rad with the bleed screw side up...

just some of my random thoughts...

[Martinm210]

Hehe.

I've never really had a problem bleed a radiator, they always seem to flush themselves out after long enough....at least my old MCR320's seemed to flush well enough.