nanofluid!

[TheGoat Eater]

Soon to be replaced with a GTX480?

[skinnee]

Soon to be replaced with a GTX480?

Yup, the Swifty MCR320 will be set aside for the big radiator roundup and the GTX480 will go in as the main rad for the test bench.

[STEvil]

Looking good.

Makes me want to finish getting parts for my loop lol

[skinnee]

I had to rob a pump from my main rig for the test bench. Replacement and some additional goodies are coming soon...funds are a little tighter these days than I would like.

[TheGoat Eater]

I had to rob a pump from my main rig for the test bench. Replacement and some additional goodies are coming soon...funds are a little tighter these days than I would like.

I hear you there as I sell all my stuff pretty much in preparation for my impending move to NYC , rather New Hyde Park ( Long Island) for a new job (should know Monday)

[HawaiianSupermn]

Hey bud, does this nano-fluid have a listed freeze point? Any chance of using it in sub-zero environment like a chiller?

[littleowl]

Hey bud, does this nano-fluid have a listed freeze point? Any chance of using it in sub-zero environment like a chiller?

my understanding is it freezes just at 32F just like distilled water.

[relttem]

thanks for the loop pic. Are you using standard thermocouples to read the temps of the radiator? That is what they look like. IF you are brave enough, this is something I would consider doing. You really don't need all of those inlet and outlet air-temps of all three radiators. Unless you have three separate radiators, but it looks like one. If it is three different radiators and they are the same type they should each be removing the same amount of heat anyway. Also, the intake air temp that the fans are going to be blowing is going to be the same - which yours are (within error of thermocouples). So, what I would do, if you have the flexible thermocouples, would be to disconnect the tubing at the inlet and outlet of the radiator, feed a thermocouple into the tubing, and then put it back on the radiator. That way you are reading the liquid temp coming into and leaving the radiator. I would do the same thing at the CPU block too. You might have to get some hose clamps or RTV sealant. You can still take apart the system with either.

[littleowl]

relttem it is one single radiator. I use the same one he is running on my main rig.

[skinnee]

Correct, one Rad. The sensors on Rad intake and exhaust are Maxim DS18B20 digital temp sensors (thermocouples from what I can tell). However, I want to have a sensor for each and use averaging across the sensors to ensure that my DeltaT is accurate. The more sensors you have the better IMO. This is the same thing I do for ambient temps, 2 or more sensors averaged.

The water sensors will be added along with a rotameter this next week, I have a mini CPU block round up starting, then immediately after that a D5 pump top shootout.

[relttem]

there is a link to a good article at the end of the thread

http://www.xtremesystems.org/forums/...=202444&page=2

[123bob]

OK, finally got to testing last night. Took my main rig down for maintenance. It is a QX6700, stepping B3, running at 3.15 gig. It is on an eVGA 680i mobo with an 8800GTX vid card. Testing took place on my garage bench.

Loop is;

MCR320 Rad>MicroRes>DDC2 pump w/ Petra top>Fuzion V1 CPU block>MCW60 GPU block>DD Maze4 SPP (aka NB) block>Rad.

Flow rate, measured when the loop was first put together, is 2.0 GPM.

Machine front view, after cleaning out dust bunnies, with the original water in it. This loop has about 2 ounces of petosin in it and about 6 drops of PTNuke. The rest is distilled water.



Antec 900 case, radiator/pump/res rigged up on a stand external to the machine. (Better to get the heat out of the case in my view. Also the place this goes wouldn't allow a full tower anyway.)


Machine back view with original water and the bottle of nanofluid on the bench.



Note the Fluke meter. I'm using the K type probe it has stuck into the middle rad fan to measure ambient into the rad. There is also a small thermometer on the bench top, where the tubes loop out of the machine.


Here's the after shot. This was after completely draining and filling with nanofluid. I used the nanofluid just as it came, with nothing else added.



Material notes. The stuff smells like a cross between latex paint and some sort of perfumey stuff to me. Taste test.........whoa, wait a minute, I don't know what's in here......taste test not performed....

I did conduct a drying test to see what happens if it is allowed to dry on the bottom dishout of an aluminum beer can. It dried like a watered down coat of latex paint....I thought this might not be good, but found that it wiped right off with a damp rag.

After sitting for about a month to perform this test, (sorry it took so long...) I noted that the fluid was more opaque on the bottom than the top of the bottle, implying some settling. There was no layer of sediment on the bottom, so it stayed in suspension. This is good. I thoroughly shook it up before putting it in the loop.

OK, to the nuts and bolts. Each iteration was allowed to run undisturbed for at least 30 minutes or longer to achieve some equilibrium.

Original water results.

Water Idle Test
Ambient on Fluke; 24C
Ambient on desktop; 25C
Coretemp @ idle; 44,42,44,38C

Water Load Test
Ambient on Fluke; 25C
Ambient on desktop; 24C
Coretemp @ load; 60,58,61,57C

Nanofluid Idle Test
Ambient on Fluke; 25C
Ambient on desktop; 25C
Coretemp @ idle; 45,45,43,40C

Nanofluid Load Test
Ambient on Fluke; 25C
Ambient on desktop; 25C
Coretemp @ load; 60,60,58,56C

Overnight Nanofluid Load Test
Ambient on Fluke; 22C
Ambient on desktop; 22C
Coretemp @ load; 57,57,55,52C

Averaging the ambients, then averaging across the cores, then adjusting for ambient differences, It would appear that nanofluid turned in about .75C worse at idle, but 1C BETTER at load.

Anything else you would like me to try while I still have the nanofluid in the loop?

Regards,
Bob

[relttem]

nice..that is over a 3C (almost 4) difference at load on average(comparing the overnite nanofluid to the water load)..not taking into account the ambient temps. I spoke with Bob and he said he would try to get around to some stringent overclocking tests for me. That should be interesting since we think this is where the nanofluid will really come into play.

[123bob]

nice..that is over a 3C (almost 4) difference at load on average(comparing the overnite nanofluid to the water load)..not taking into account the ambient temps. I spoke with Bob and he said he would try to get around to some stringent overclocking tests for me. That should be interesting since we think this is where the nanofluid will really come into play.

The ambients count. With those factored in, it's nearly a wash at the single OC data point I tried. However, I apparently missed the bigger point of the test. That is the possibility that the nanofluid will allow a higher clock than water. If so, then it is certainly of value to test for this. I may get more out of the mass-cooled portion of the farm with this stuff...

I am no where near the max clock this rig can run at. The 3.15 is just my 24/7 super stable main rig number.

I'll be doing some more tests looking for where my max stable clocks occur with the nanofluid vs regular distilled water. Reltem, YGPM on some possible further instrumentation I could do. We might get those loop water temps too.

Stay tuned, this will take me a few days to work this in between the day job tasks.

Bob

[littleowl]

Great info Bob! Thanks for your hard work!!!!!!!!

[skinnee]

Full Dual Core write-up over here -> Linky

[123bob]

Man, I have a good one for you guys. Weird, but good.

In setting up to do the test suite described above, I installed T type thermocouples in my Res and in the T-line I use to drain the machine. This was to give the inlet and outlet temp of my rad, or looking the other way, providing the heat dump of the full loop. The T line thermocouple was sealed into the tube using a hose clamp and a rubber cork. It was not leaking around the wire, so all was good, right?

Later in the evening, I ran a test point at my current clock. I noticed the display on my omega meter was getting flakey. Is was half gone. Oh oh, WTF!!

I took the thing apart and what do I find? Nanofluid inside the meter!! How did that get there? I thought maybe I splashed some on the meter when I was putting the lines in the rig. OK, no problem, cleaned it up and thankfully the meter was OK. I plugged everything back in and continued. Came back in about an hour. SAME THING!! Meter had Nanofluid in it!!! On careful inspection, I found that nanofluid was being forced up the inside covering of the thermocouple wire!! Not a lot, since it's so tiny, but enough to potentially mess up my meter.

So, I'm down for a few in order to epoxy the ends of the tcouple lines to prevent them from pumping out any more nanofluid!!

I would have never caught this if it wasn't for nanofluid's white color. You can see the wire is milky colored compared to the res tcouple. The Res tcouple was not pumping any out since the res is open to air pressure. The lower rad t-line was sure going to town though!!

What a weird one. Something to watch out on for those sticking thermocouples in pressurized systems!!

Bob

[littleowl]

I started working on the B3 today and have ran into several snags
I will post my testing results for what I have so far but as for my antifreeze test to see the temp differance between nano and it well I can stiff give #'s but lots of math will have to happen to figure it out. my radiator sprang a leak and I have no other way to fix it and keep the same mount.
I will be putting regular 20% antifreeze and steam distilled water on it but it will be in a 120.3 and a different water block.

[relttem]

if you have a smaller radiator you should be ok with that. The nanofluid should defeat the necessity of the giant radiator.

[littleowl]

well I am going to make the loop I hope tomorrow but don't think I will have time till friday
That also puts me down a quad crunching

[relttem]

you might just have to start over..I promise you wouldn't be the first person to have to do that..

[skinnee]

Nice work Bob! You're getting nearly the same margin on the results as I am. Good to see our test data is very close.

[littleowl]

you might just have to start over..I promise you wouldn't be the first person to have to do that..

in order to start over I may end up needing you to send me just a little more fluid because of the 120.3 taking a good bit more then the 120.2 as well as a much longer loop.

@ Bob
Thanks for the pics! you helped me salve my problem on how I was going to mount this radiator for my wifes B3. I am going to (so to speak) take your idea.

[littleowl]

I know I don't have the 20% antifreeze/distilled water to compare to but I thought I would share what I did log so far.
I started out with room temp at 19.4c on stock air cooler using ICD7 TIM.
My core temps are all loaded with boinc.
core0: 62
core1: 62
core2: 59
core3: 61

I changed out the stock heat sink for an old maze4 with my home made hold down. my cores were
core0: 42
core1: 45
core2: 38
core3: 42

That was all done at sock speed. on a Q6600 B3

So I oced the chip to 3.12

My room temp went up to 21.1
My core temps were
core0: 54
core1: 55
core2: 49
core3: 50

I didn't make it to the water testing because my rad sprang a leak so now I will have to wait and maybe test again after I get the quad up the way I wanted it in the first place.
The next set of testing will be with a 120.3 rad and a 6002 water block and my home made hold down for it.
I was using coretemp 0.97.1 to get my readings on temps.

[123bob]

in order to start over I may end up needing you to send me just a little more fluid because of the 120.3 taking a good bit more then the 120.2 as well as a much longer loop.

@ Bob
Thanks for the pics! you helped me salve my problem on how I was going to mount this radiator for my wifes B3. I am going to (so to speak) take your idea.

Take away!! I figure it's best to get any heat out of the case I can, and let the fans handle the rest that's left in the case. It really makes no sense to me to try to stuff a rad in a hot case. But that's just me.... This solution worked for me with the short mid-tower case I was limited to, in order to stuff it under the desk.

@ reltemp, I'm proceeding to collect the detailed data we talked about. So far, so good. (It quit peeing nanofluid out from the lower thermocouple wire..) I'm allowing at least a few hours between setting a data point and taking measurements. I figure that should be Plenty of time for the thing to settle into it's final thermal equilibrium.

Bob