External A/C Chilled liquid cooler project

[technogiant]

Yeah it's a real case of working this one out as you go along, there will definitely be some optimum level for this degradation to occur...too cold and it will all be solid and nothing much will happen there, too hot and the layer will redissolve.....but somewhere in between is where the problem lies.

Did the LP people come back with any help?

[Orthello77]

Hey TG, yeah i think your spot on.

So far what i'm doing seems to be working , although it needs more testing with lighter cpu loads. What i'm doing is prior to bootup i'm sitting in the bios with pumps off waiting for the cpu to hit around 60c to ensure its getting a good heatup / reflux , and then turning pumps on starting up. Then once in windows, i'm running prime 95 not allowing the system to idle. Although prime temps are very low -3,13,13,18 (5.0ghz 1.5v) it seems to be working ok so far. For two days now i have not seen any reduced performance but i have not gamed or let it idle. I need to test more with games which will have lower load temps than prime95.

Not impressed with the LP people, they never replied. Its fair to say the only types of cooling likely to have this issue with LP are systems like mine where temps cross the MP of LP often, Single phase systems under heavy cpu loads possibly could have this issue with LP also although temps would be around 20c south of mine so maybe not. Ambient cooled or Cascade/LN2 etc won't have the issue i would think as the state of LP would remain constant.

Anyway i have the LP ultra here now , i'll do some game testing with LP and see just how long the issue takes to occur before i try the LP ultra. From what i hear the LP ultra has a lower w/mk compared to LP and therefor does not cool as well. I guess i will find out. First i have to see what set of conditions brings on the degradation in the LP , then i can try the same with LP ultra. I'm figuring long bouts of gaming will cause problems for LP.

I also have the flow meter so hopefully that works out and i can get some flowrate figures finally.

Additional Comment:

Well had a good day, finally got a working flow meter, check this out !

Also redid mount with liquid ultra, temps so far seem on par with liquid pro (LP), fingers crossed it doesn't have the same Seperation/Crystallization issues as LP.

Super impressed with the aquacomputer software , its allows insane amounts of customization also. The sensor connects via USB.

When i started up flow rate was about 250 litres per minute, as i'm typing this the flow rate has dropped to about 170 litres per minute , so the pumps are definately feeling the cold. Resoiver Liquid temp is -28.5c with a chamber temp of -22.7. Interesting to see Aquasuites readings on the internal and external temps. The external temp is through the phobya inline liquid temp sensor which is a bit off (I can't have a chamber temp of -22.7 with a liquid temp of -21.4.). The phobya sensor is plugged into the external temp on the aquacomputer flow sensor. The Aquacomputer internal temp through its own sensor seems pretty close which i didn't realise it had until i powered on.

Whats amazed me is there is a flow rate difference of 40 liters per hour within 4c ! -24c has a flow rate of 210 litres / hr. -28.5c has a flow rate of 170 litres / hr. Thats a staggering 23.5% of flow rate reduced within 4c if that flow rate sensor is correct .. i never would have thought the flow rate loss was that high. Compressor has just cycled off , now liquid is at -23c and flow is at 218 l/hr so the flow rate is very sensitive to liquid temperature.

Testing at 5.0ghz (1.5v bios/1.52v actual) here for comparison against liquid pro (liquid ultra mounted) but cpu is also fine at 5.1ghz with 1.54v bios.

[technogiant]

Loving the pics Orthello...truly impressive build.

Just wondering about that flow rate, do you think that the pumps are being affected by the cold?...or maybe it could just be that the liquids viscosity is increasing as the temps are decreased....although it does seem a significant reduction in flow for for temp reduction of 4c.

Yeah some real nice stuff there.....hope it will give you the tools to help get to the bottom of the LP problems

[Drewmeister]

Great temps there orthello... I wonder about that 4th core though. Not sure if that 16c delta is an indication of anything. I get approx 7-8c max difference on hottest core.. that's with IBT or 30min prime95, cpu @ 5.1HT 1.5vbios/1.488v os -24c liquid. Btw..it's delidded but not naked yet... using ASC on die and ihs. You might want to check flatness of wb and lap if needed. If you're still using springs on the wb then they might be reacting to temps and or stress from tubing.. personally I never use them.

Your coolant is just viscous, the flowrate will drop off suddenly when nearing the slush point. I experienced that too when using -34c rated de-icer.. flowrate is excellent to -26c, then all of a sudden drops off at -28c. I add 5% methanol and back in business. I have 2 gallon res and lose about 2-3%/yr to evaporation. Basically just needs one small bottle of HEET(methanol)/yr. Your core temps are still excellent vs oc though, if the pump motor current is high then maybe increase the methanol.

That's some sweet software you have there. Did you also get the 5.25" multifunction touch screen? I wonder if that would work with a SS unit to around -50c.

I finally got around to insulating my mobo, today was my first attempt with sub 0c liquid on this 3770k. I came pretty close to WR .. I'm sure you have to be close too?

If you have time, could you run another Valley.. 1080p extreme preset?

Great pics!

[Orthello77]

Hey Drew , thanks for the tips. I definately will order more methanol today , i had a feeling it was doing that but couldn't be sure. With all the work i have been doing each drain / remount of cpu i havn't been refilling the resoiver (just capturing the drain and putting that in a bottle for latter), i think because of that the evaporater been building up ice on the top side of it where the liquid level is lower than evaporatier, putting more h20 into my mix.

RE my tim i'm using Liquid pro ultra now , which has a couple of benefits , for one it doesn't seem to get any degradation (been a week now and no problems). Also the temps between cores are more even , core 2 and 4 often equal in max temp. That 16c delta is not there anymore either , its possible because of the lack of degradation its staying much more even and consistent in its bond.

Using springs, i think its ok still, the variation was due to the tim , not the pressure i feel. I never had problems with mx4 , and so far liquid pro ultra seems great too.

Looks like you have some great silicon there. I can do a few things at 5.2 ghz but its not stable. Also due to the bug in my asus mobo HT looks like its impossible to activate atm , people have been talking about it in the forums for months and still no luck. Basically i'd have to change cpus to force it to refresh to enable HT, the momment you disable HT it stays disabled. I'm over this asus board tbh as theres another bug where if the board gets too cold it hangs with an A2 code (IDE) , i'll buy haswell the momment it lands here and get rid of this board and jump on a 4770k. Guys validating 5.0ghz at 0.9v on air (reportedly) !!! thats just insane. I didn't gain any ground with my 3770k over sandy so it won't worry me too much to upgrade so soon. Go hard for that WR Drew, i'm not going to get there with my mobo HT issues and silicon woes. I'll have a good crack at the 4770k when it comes out.

The flowmeter / software should work with that SS unit. You can set manual min / max temperature ranges. You don't need the 5.25 touch screen, just the usb version and download the free software. One word of warning, take the oring out , jbweld it . Mine leaked within 24 hrs and fried its temp sensor. Took it out and dried it and replaced oring with the jbweld . I have another one coming which i will do that mod to on day 1 to get my temp reading back.

I'll run another Valley test for ya ..

[Drewmeister]

Glad to hear you finally worked out the TIM situation. I'm going to give the Liquid pro ultra a try too.. thanks for the info.

Sorry to hear about that HT problem on the asus. Any specific cpu needed to reset the HT ? Well you certainly gave it a good run with the Ivy. Perhaps not the best oc'n chip but I'm sure you had fun and learned a lot from the experience. Thanks again for sharing all this info with us.

My Ivy might be a tiny bit better silicon than your's but it's really nothing special. I'm hoping it scales better once on ss and cascade. My best oc'n amd chips weren't that great with ambient cooling but excellent when on ss and cascade, hopefully the same holds true with this 3770k.

Looking forward to your results with Haswell, that should be here next month right?

[Orthello77]

Hey Drew,

The ivy 3770k has been fun i must admit, more fun than my beloved sandy 2700k. Even the temps were better in the end (lower wattage chip and direct to die cooling) , but unfortunately the nature of the silicon it just hasn't overall outright beat the sandy in performance. Re the HT enabling , the board needs a cpu change to enable the HT, i would have to mount with the sandy to enable HT for this Ivy and then if i disabled HT from there it would be permanently disabled :-( Its lucky i'm a gamer first and a bencher 2nd otherwise this HT problem would have really $#$#^ me off lol. Still i won't be unhappy to see this board go as, i think me and asus have ended our love affair over this board and HT / IDE bugs - and i'll be shopping for a new brand mobo for the 4770k. 4770k boards are been previewed everywhere atm , just no chips to play with yet ...

I think you'll do great on the cold , i think if i had another 5-10c colder temps 5.2 stable would be a reality for this chip, they scale well with the cold. When i started without the delid with chilled liquid my max was 4.9 stable, once delidded with ihs it was 5.0 with no show of 5.1 been stable, and then with direct to die 5.1 was fine (just) . So you don't seem to hit the brick wall like with Sandy it just keeps improving with temp drop.

Anyway i've been thinking about Pumps ... my wife calls me mr pump as thats all i've been fixing for the last few months lol .. so I've looked at the little Giants and the Iwakis again but to be honest there performance per $ is just not there. The performance per footprint isn't there either , not that thats much of a concern , but in general the more industrial pumps are just slower / bigger and more expensive than 12vdc pumps. This is not the states here either iwakis / LGs cost a fortune here. The problem has been longevity of the 12vdc pumps not the performance.

One thing i noticed is the MCP35x2 inside my chamber has been rock solid for ages now. So i've been thinking even with all my pcb spraying and dielectric grease that moisture is still getting into PMP500s pump internals. So anyhow i started to think about how do i just solve the condensation problem with the pumps (if its not the pumps just blowing there fets after a while - and it could still be).

See the pics below ... just a quick setup to prove my new theory on how to solve the pump internals condensation issue ... dunk the whole lot into vegie oil !! ... (i was too cheap for the parrifin oil at $20nzd for 1 litre so veggie will have to do - its RBD soya oil to be precise ) . Early days yet, 2nd day , even the the pump that was going faulty seems to be working better now (its been going solid for 3 hrs after stopping and starting for a day or so). One of the side effects of this is cooling to the pump internals should be better if that ever was an issue. The electrical pcb noise has also been reduced substantially - another benefit , when you get three of these pumps they can have a doppler effect.

I'll have to regular checks for leakage but tbh i don't think thats going to be an issue. My chilled liquid is purple in colour so will be quite easy to spot. I'll give it a fortnight like below , if its still all rosy then i'll set it up a bit more permanent.

[technogiant]

Excellent idea there Orthello.....sure that will work out for you.....but then again I'm biased, I think all things "submersion" are a good thing.... ;-)