Next testing project starting today

[pHaestus]

Good ole Canada; today's a holiday!

I don't have quite as much time to test now as I did last month because we have a brand new baby at the house now. Nonetheless I have several projects to finish up.

The MCWChill testing was cut a little short by the untimely demise of the Swissflow flowmeter. Dunno wtf could have happened to it; it was working fine a week ago but when I went to test with it 2 days ago it was no longer working. Crap crap crap. At any rate I'll just take the data I have collected thus far and write that review up this week. Thanks again to Swiftech and Bill for sending the prototype.

There are 2 projects I need to take care of quickly: A review of Swiftech's complete water cooling kit that they kindly sent to me, and a review of their new 92mm -V heatsink. I'll try to set up the kit today. The heatsink I am waiting on a 92mm Stealth and Tornado fan (ordered last week from www.bigfootcomputers.com).

Once these reviews are completed, I am seriously going to begin waterblock testing. I also am rebuilding Pro/LAN with new cooling gear and starting on a Pro/LAN-related mod project (bought the stuff last night). Should be fun.

A question about waterblock testing has arisen in my mind due to the recent discussion about www.liquidninja.com 's wb roundup. Would the readers of this site be interested in a waterblock comparison doing the best I can right now with current stuff? That stuff includes: Calibrated AMD internal diode reading, 4 YSI liquid probes, 2 YSI air probes, flow rate measurement, and a willingness to repeat tests repeatedly. I think I can get some handle on CPU W by comparison of deltaT across wb and T/W using a block I got from Bill (a cross calibration if you will). If the review would be met with general derision though I will just continue to do other things while I wait for a good price on a few DMMs suitable for measuring V and A on my die simulator.

[BillA]

go for it

but that V*A business will devil you
I just had a hiccup of 0.1W and my C/W curve is sh*t !
grrrr

[pHaestus]

Yea I have to get started on that though. Would be interesting (to me anyway) to run tests on CPU and then repeat on die sim later to see how close I was.

I need I guess 3 DMMS: 2 for die sim and 1 for pressure transmitter. So much $$ to spend yet!

Wish I could get that Omega working for pressure drop. I may try to open it up and see if I can spot any obvious problems.

[jaydee]

I think you should start testing blocks. Your setup is beyond anything anyone else is using and most of us here know how picky you are with your methods. Not to mention your technical abilities to compensate or figure out oddites in results. IMO your way ahead of everyone else and your results will be a blessing of higher quality to the community that it has desperatly needed since Bill stopped testing.

*Pulls lips off ass*

I also know I would like to send you a couple of my own blocks to test and use the results to hone in my own test bench. Not looking for a review really, but more technical data I can use for reference. If your up to that let me know. Certainly no rush. I don't even have the blocks made yet I have in mind, and after I send them you can do it whenever. I can send you a stack of fan grills for compensation.

[pHaestus]

JayDee one of my main interests is in testing blocks from our WB design forum. I think that as is the ideas by the wb makers in that location are ahead of most of the commercial mfgrs. Giving useful testing results to you guys would I think be a great service.

In particular with the DIY blocks we can have regular contests or challenges such as: Best DIY peltier block, best DIY GPU block, best straight water block, etc and then have regular review/competitions. It would also be really interesting to watch revisions of blocks progress over time.

Since I know the makers personally I can also make requests like "I need a hole large enough to put a type T thermocouple in the baseplate" etc etc too.

That's still in the future though. A first task will be to repeat testing of as many of the blocks Bill has tested as possible. Hopefully I can dial in my testbed like that.

[jaydee]

You the man pH! That is exellent. I can now spend less time and money on my test bench and more time and money on block design and fabrication. Hopefully we can get a few more people capable of making the blocks they are comming up with...

Competition? Woot!

[Blackeagle]

I agree with Bill and Jaydee, go for it. Your test rig is so far ahead of the Player0's on the web it's a non comparison.

I'd love to see you test some to the better performing blocks that have not been tested by BillA the Cascade, Swifty 5002 & RBX to see how good they are for sure, when tested by someone knowing what they are doing.

The Cascade's performance is well known to be very high end, just from user results and Cathar's testing. But the 5002 and the RBX are not well known as they are newer.

[pHaestus]

I am sure I can work out something with Cooltechnica and Swiftech to borrow any blocks I want for testing. Cart before the horse though; MUST get cracking on test setup. I have my "high resistance" loop plumbed for comparative pump testing. I should finish that up tonight.

I am trying to get as much testing as I possibly can done this week while my mom is here visiting. After that I am going to be uber busy watching my sons in the evenings. If I can get the test setup all plumbed and running before that then hopefully I can at least keep a block always running and slowly collect data.

[Blackeagle]

I can see where you'd want to do some comparisons with Bill's data.

What are the blocks you'll be testing for this, Slit Edge, WW and some Swifty's?

I'm very glad to see you moving in this direction pH. Some top shelf block tests will be a refreshing change. And seeing tests of the new designs will also be good. I think that testing of new block designs will help stimulate more work on that area here on the boards. There are a number of good block makers here that havn't been heard from of late.

[pHaestus]

meh You guys give me too much credit. Nothing magical about data I produce; I have folders full of it on this hdd that isn't any good.

[Aardil]

pH, I think I would trust your data and pretty graphics over anyone on the web. No slam on Billa since he cant realistically do testing of "other blocks" anymore. At least thats what I understand.
Being as close to a noob as you will probably ever see you have never at least not tried to answer my questions, wheather I understood the answer well thats a different story.

For the developers I see this as a great asset, and with enough testing it should make it alot easier for us noobs to pick a block that truly does what is promised instead of 6 pages of manufacture hype.
I am behind you all the way just dont ask for anything complicated like a scewdriver or a wrench ( always get them 2 confused)
Aardil

[pHaestus]

Aardil:

It's good to be skeptical of test results I post as well; I still struggle with testing quite a bit.

I sent Joe an article on pump flow rates in a couple of real cooling systems over the weekend; hopefully it'll get posted tonight or tomorrow.

I spent an hour or so setting up my testbed for waterblock testing tonight; still a lot to do. I had to break down the chiller and its 3 loops and then reconfigure everything for the waterblock testing. Will start out with a block that came from Bill and has lots of testing on his part already done. Hopefully tonight I can finish getting the loop up and then tomorrow night I can give it a trial run.

[pHaestus]

My orders from bigfoot computers arrived yesterday and so I think I have all the needed parts for testing out the Swiftech review items:



The Swiftech heatsink will be tested with an extremely high CFM and low CFM 92mm and 80mm fan, and I will PWM the high power 92 and 80s as well. Planning on comparison with an Alpha PAL8045 and a Thermalright SLK800.

The Ys are so I can test 3 configurations of the Swiftech water cooling kit: Single BI Micro, 2 BI Micros in series, and 2 BI Micros parallel. Lots of work coming up.

I plumbed and then replumbed the loop for my testbed last night. I think I am going to have to replumb a third time to really be satisfied with it's functionality.

There are some considerations that most users simply neglect when water cooling: Ease of bleeding loop of air and ease of draining the fluid. When you are swapping waterblocks constantly though this becomes one of the primary concerns as air in the rad (yes I am poor and have no chiller) can really screw up an evening.

I believe the solution is to put in valves at the top of the heatercore (it's a big single pass) and at the bottom of the res and then just bleed the system of air by opening the heatercore valve and drain by opening heatercore and res valve. I am also thinking that I will implement this design in my external watercooling box I am making. I seem to recall myv65 posting a picture of his system plumbed like this several years ago (an Antec SX830? I think it was Dave's anyway) but the design hasn't really caught on.

Getting air out of radiators you have bolted into place on a case is actually a pretty big cooling issue that people don't talk about much.

But I digress. In happier news I have three temperature probes plumbed in to measure the delta T across radiator and waterblock, and they are all within 0.02C of one another at 20C. I will have to use several different water temperatures and see how tightly they adjust up to 35C or so but I am hopeful it will be easy to deal with.

[Skulemate]

For my latest built I had a fill tube put in the top of my heater core. So far I have found it works pretty well, as the core holds enough water to function as a reservoir during filling, and I can be certain that there is no air left in the core when I am done. I would recommend this option to anyone who is interested in doing so based on my results so far.

[jaydee]

There are some considerations that most users simply neglect when water cooling: Ease of bleeding loop of air and ease of draining the fluid. When you are swapping waterblocks constantly though this becomes one of the primary concerns as air in the rad (yes I am poor and have no chiller) can really screw up an evening.

Heh, I just about forgot to add a drain system when I resetup my "play station" this last weekend. See that hose goin off the counter? That is connected to another on of those PVC valves in which I use to drain the system. That monster T I have right before the inlet seems to have solved my pump locking up issue aswell. I let the water get up to 33C and it still worked fine. not to mention the air bleeds in 10 minutes. best part is the entire system uses just over 20oz of coolant.


I need to get a usable flow meter and plumb it in somewhere. Still need to plumb in the thermal probes. I am goingto give that joy stick mod a shot. I discovered a joystick port built into my laptop and my ECSK7S5A also has a joy stick port built in! I am going ot use one of those to monitor the temps. Certainly not leet accuracy but good enough for what I want to do.

Tear it up pH!

[jaydee]

Keep an eye on the HydroThurster pH. The O-Ring in the very front of mine decided to leak last night. Woke up the the nice tune of a half filled pump sloshing water around. By some miracle it appears the overheat protection actually worked this time allthough I didn't get a chance to check to see if the CPU was still good. Had to get to work. best news is all my planing to keep anti freeze off the white carpet in case a leak worked! Not a drop made it to the floor. Only half the fluid leaked out though.

I have had nothing but problems with this pump..... Sucks my Little Giant 170GPH seized up the other day aswell. I am down to one POS 200gph submersable.... When I get home I will tear the HydroThruster apart and see what gives....

[leejsmith]

when you are ready i will send you a couple of my diy blocks.

one made from a tiny fin copper base and a diy cascade attemp.
also i have a cascade style gpu cooler.

they are not much to look at but i am intersted to see if they perform like i think they do.

Lee

[jaydee]

Originally posted by jaydee116
Keep an eye on the HydroThurster pH. The O-Ring in the very front of mine decided to leak last night.

Well it isn't the O-Ring's fault.




Bah... Hopefully the ISP Weldon #4 will fix it...

[pHaestus]

PWM provided by Crystalfontz 633 LCD. Stealth fan was just placed on the delta T vs PWM% line based upon its delta T.

This was just a first stab at testing and I think the heatsink mounting wasn't especially good. I had some hassles getting the unit to mount on my 8K3A the first time. It seems like I have figured it out though now. After remounting, I am looking at CPU temperatures at 50% pwm ~5C cooler than in the graph above. So take this as VERY rough data that may not be representative of final testing results.

[BillA]

nice data
had no idea a Vantec could be made to do that

-> now how do we get the 'outside case' apparent sound levels ?

lol, just being difficult (again)

[pHaestus]

OK I am in the clear. The temperatures were 5C lower than my test run last night because when I pulled out the CPU and changed Goop application I left CPU at 11X multiplier. So it isn't running at the same speed. I'll bump the speed back up tonight (don't think there will be a dramatic difference in settling over 12 hours between 12.5x175 and 11x175 at same voltage) and run tests again.

The heatsink goes on and off much better after a couple of remounts; this doesn't bode well for the reproducibility of the early data sets.

The Vantec Tornado even at 5% PWM is louder than the HDDs. The full power Stealth is quieter than the HDD so that's the noise I hear. Outside case noise levels would imply that I have doors on the case. Can't easily do that and still mount the Digitec air probe.

[Les]

Originally posted by pHaestus
.... The temperatures were 5C lower than my test run last night because when I pulled out the CPU and changed Goop application I left CPU at 11X multiplier. So it isn't running at the same speed. I'll bump the speed back up tonight (don't think there will be a dramatic difference in settling over 12 hours between 12.5x175 and 11x175 at same voltage) and run tests again....

With regards to scaling is interesting if a scientific 5c, boring if a poetic.

Have lost track of Diode calibration status - Diode on this cpu still to be calibrated?

[KnightElite]

CPU is running K7Burn, etc....?

If that is the case, then that Heatsink with the Vantec Tornado is pretty amazing. 27ºC is well... awesome. I should have that SLK-947U next week sometime, I will bring it over so you can test that as well.

[pHaestus]

Les:

Calibration is still pending; as is that MCWChill review (for that reason). Not scientific at all; just when I rebooted I turned on K7Burn and noted die temps were down 4-5C. I was planning on repeating the PWM vs delta T for 5 points tonight and then also running as a function of CPU voltage and of MHz. A late night for me.

I am planning on getting 5 heatsink mountings with the same fan, MHz, and voltage to be able to do statistics. This testing then becomes more time intensive than a waterblock if multiple fans are tested. I kinda like being able to place different fans on the Tornado PWM curve but not sure exactly how useful it is for someone making buying decisions. There has to be a clearer way to tabulate noise/performance relationships. One thought I had was to plot a range of "acceptable C/W" vs. "MFGR dBa". This way you could see what noise level is required to achieve a certain performance. Still ruminating on this.

KE Yes to K7Burn (high priority). But actual die temperatures range from upper 40 (100%) to high 50C (5%). Note the performance is DELTA T; not quite as impressive as one would think but still quite good for air cooling.

[KnightElite]

Oops, not paying close enough attention again .

Still pretty good temperatures though, to be sure.