External A/C Chilled liquid cooler project

[Orthello77]

My worry with the plate hx is i loose efficiency , the fluid temps will not be the same and hence yes you pick up flow rate etc but loose that gain with the extra hx.

I'm not going to stress too much until the chiller is at ground with reduced pipe length. I'm sure this will give it 50 % better flow or near it. At the momment it has nearly a metre of header height and at least 3.5 may be 4m of pipe length and two 90 degree bends that go with that. When its at near no header height and 2m of pipe length i'm sure thats going to help it. If not then i'll just get another pump. Certainly 8-10c i could pick up i reckon with better flow. I'll know when i get it right as GPU2 wont be 5-6c higher than GPU1.

IMO unless you are compressing movies i doubt theres much use for the 2700k over 2500k in a gaming situation , just check the L1 cache size is not larger, that could be of use.

Drew have you got any settings for 4.5ghz + i tried bumping vcore once and the QPI bumped twice and it boots at 4.5 but won't complete intel burn test.

Also one issue i never expected is the pipe lagging seems to not be sufficient especially near the base of the chamber, i'm getting condensation occurring on the outside of the pipe lagging. Fluid temps are averaging around -28c, last night i got to -29c with intermittent testing. I'm guessing the lagging needs to be thicker or is it because the lagging might be compressed in length , i normally do this to ensure good tension at the ends of the piping where the joins are , i'm thinking that might not be the best thing to do as it makes it more solid in a sense ?

[Drewmeister]

That's awesome man 4.44ghz!! Great to see those settings working for you. 4.5 ehh.. bump vcore to 1.42-1.435v. You can easily go to 1.5v with that cooling.. then trim back as needed... low 1.4's should be plenty for 4.5 with your cooling though. Which memory do you have? Screenies would be very helpful and certainly a nice addition to this thread.

Sweet oc on the 580's, too bad about the performance reduction from the drivers. Have you checked how it impacts any of your games? Lol.. what game would lag on your pc anyways?

Tg.. that's a good point to make on the sealed reservoir. It does actually make a difference on certain systems. It's not going to make any noticable differene here though because the frictional losses will offset any gains that could be had from the increased surface pressure.

RE: 2500K vs 2700K. You'll still achieve a great oc with HT on the 2700k. Gaming performance will be the same between the two. I game some too but personally I'd go with the 2700K for the occasional benching, video editting and just to future proof a little more.

Plate hx's are very efficient and capable of approach temps of 1c. If you're interested LMk and I'll look into it more for you.

Yes.. bad idea to stretch or compress insulation. What is the thickness of the insulation? If there's enough clearance between the insulation and tubing then maybe you can add a couple wraps of foam tape under it? That would add approx 6mm more thickness and eliminate any air gaps. You could probably do that without taking everything apart or slitting the pipe insulation.

[Orthello77]

Hey Drew ! yeah i was pretty wrapped with that link you gave me , entered the settings and bang it just worked straight off. Prior i was running 1.45v at 4.0ghz so to drop vcore / and drop QPI and be sweet at 4.41 was great (sure lose HT but didn't try it on at the new settings so maybe it would be ok not sure , needs more testing). Anyway happy with HT off for the momment , will try for 4.5 tonight.

Hm yeah i was thinking older games are probably going to be really sweet under the old drivers .. its just new games and normally the ones i want to play that would benefit from the newer drivers unfortunately. Really quite annoyed at the situation with the 580s , i think i'll just have to see what pans out , will the drivers get fixed - i doubt it. Theres a nice 7970 lightning edition coming out with an EK block in the next few days , that could be nice to play with as it seems to be getting hard to play with nvidia cards.

Its 13mm thick insulation , so its maybe not thick enougth to start with. Its definately compressed, i made it so to make sure it was snug over fittings so i may have to redo that or put extra insulation over it. I like the foam tape idea , i'll have to see how wide i can get that stuff.

Just got the top chamber lid back from the cabinet / acrylic people, its a MDF lid with recess for the acrylic. I'll back it with a frame of insulation. Hopefully will get rid of the condensation on my gasket lid.

Will put some screenies up tonight on how i get on !

Additional Comment:

This took about 20mins ... didnt test below this voltage .. just went straight to 1.43 in bios and bingo. Chamber was only at -8 (gets to -16c) , liquid was at -22 and still dropping (just started up). Under load in this test cpu temps were up to 40c for short periods. Remember this is with the very poor flow rate at present.

Due to extra 65 mhz overclock on the GPUs and extra an 500mhz overclock on the CPU , GPU1 is no longer subzero. I'm hoping extra flow will sort this out soon.





Additional Comment:

Quite excited. Think i have this poor flow rate sorted. Check these pumps out.

http://martinsliquidlab.org/2012/01/...-mcp35x2-pump/

I've always overlooked them due to their size -but their head pressure is exceptional . I've ordered two of them (thats 4 internal pumps in series) !! from frozenCPU so my flow rates should soon be wayyy better. Ordered an extra filter so i can run parrallel filters also. They can be controlled via PWM and speed fan so thats another nice little benefit , when i'dling away i can reduce flow rates and increase the liquid temp drop in the resoiver .. when busy flow rates can increase again. These things are beating up dual MCP655s and remember that 3 MCP655s was superior to what i have now in flow it should boad well.

lol this always happens , too impulsive as usual. The guy got back to me from Little Giant distributors here, he basically told me what i didnt want to hear. The pump can perform a lot better if the inlet comes from the bottom of the chamber, eg positive head pressure on the inlet side. Also pump is not orientated for optimal performance. So now what i think i will do is make the best of my situation now and do just that just with the new pumps. I'll re route the inlet from the bottom of the chamber and use the new pumps which should really kick flow rate into top gear for me (as much as possible with low temp / viscosity etc). Isn't it interesting if you had to build anything twice you could save a lot of money the second time around .... oh well.

[Drewmeister]

My concerns with those pumps would be how well they can handle the temps with the methanol and the possibility of moisture getting inside to the electronics.

You might want to test with the pump inlet where it is but with pump/s mounted low first. The siphon action will help the suction head pressure a bit.

Nice job with the 4.5 on low volts.

[technogiant]

Hope those pumps do the trick for you Orthello...not sure if you are going to be able to use the PWM control though as on board temp monitoring is fubar below zero....like I said previously...it would be ideal if there was a fan/pump control program that used cpu/gpu load %age rather than temp.....but I don't know of and can't find any.

[Orthello77]

TG yeah good point .. PWM may not be an option if GPU temps are registering zero all the time.

Pumps will take a week or so to arrive so will be a while before i can test them. Valid point drew, i might seal them up somehow. They did mention in the review they can get hot at low restriction so maybe the temp inside them will be ok. I might just really insulate them well. If i put the filters before the pump i could possibly just jb weld or expoxy them up so they are air tight but i'd have to be sure i never need to get inside them. I have another filter coming i will run in parrallel to reduce restriction. I still like the idea of the filters been external so i can clean them easily.

If the new pumps do the job as expected i won't alter the inlet position, although it could be done if neccessary.

Additional Comment:

Picture Gallery of the testing phase - Garage setup. In a couple of weeks the A/C chiller will be outside the house. I can barely move without tripping over something atm lol.

I have just finished the viewing pane. In one picture below you can see moisture building up on top of the chamber gasket. Thats been largely eliminated with the viewing pane.

[technogiant]

Looking really awesome Orthello.......and great temps as well...is that -30c in the res and -17 in the chamber...can't wait to see what you achieve on your cpu/gpu when you've got your flow sorted out.....seems you are going to benefit from going the extra mile for the full water cooling system.

I get a little condensation on outside of my viewing panel as well......but no big deal so long as it's not inside...and I see you have a pouch of desiccant in there.

It will be interesting to see if you need to dry your chamber out regularly between uses...I don't have to bother at all with mine.....don't know if that is because of my expansion chamber or if a good air tight seal, as I'm sure yours has, is good enough to prevent making that necessary.

I've ordered my Zalman part heat pipe part conductive cooler....just to have a play with it on my old hardware before I upgrade.....hoping it will take 10c to 20c off my current loaded temp of around 30c.....which isn't too shabby.

[Orthello77]

Hey TG, yeah the cooling seems quite strong, today i thought i'd load test it to find its holding point so far its kept on dropping , at -24.5c fluid now and i've been rolling heaven for about 4hrs. Thats two GTX580s at 1.213v and 1065 mhz and a 4.5ghz i7 at 1.43 vcore and holding the chamber at -14.4c at the same time. So i'll know where its going to stop under load soon. Once i get the flow sorted i expect it won't hold so low in the chiller but that won't matter too much as i should get massive gains on the chamber side cooling.

Yeah so far there is no signs at all that i need to open the chamber. Been testing for a week now and Zero condensation - even the disicants have not dissolved at all. Just seems uneccessary. Humidity is off the scale according to the sensor , it reads to 25% and then goes -- that happens at about -2c in the chamber. So fingers crossed its never an issue.

Trying to resist the upgrade atm .. im sure i can run any game with any settings at 60fps in my res 2560 x 1440 so until i cant might try to stick with what i have.

Yeah with that new cooler you should get some amazing cpu temps. I've tried 4.6ghz here and she boots but i just cant make it through intel burn test .. maybe not without silly voltage .. i'll wait till the flow is sorted and then try. Yopu're right its great to be pushing old hardware !

[Drewmeister]

Incredible work there orthello, you never cease to amaze us!

That looks really slick with the viewing pane. I wonder if a thin coat of rainX would help prevent condensation on the cover.

Glad to hear your guy's systems are staying dry. Just keep an eye on the humidity after the system warms up to room temp. If it's higher than ambient then obviously some air entered the system and the desiccants have become saturated. That doesn't necessarily mean you have to air it out though. A few drops aren't going to hurt anything.. if you have a small puddle forming then might want to air it out. That could take several months or more on well sealed systems like yours.

The specs look great on those pumps. 44psi eh.. lol.. watch the orings! Looking forward to your new core temps and oc results.

Have you checked the compressor temps? How are they under Heaven load at -24.5c? If you don't have a probe can you keep your hand on it for more than 3 seconds?

I agree.. it's great to push older hardware to see how well it holds up to the newer games and apps.

Hey TG! That new cooler perfect for your app. Let us know how it works out. GL!

Btw.. nice pics orth.. thx!

[technogiant]

Drew, I've been a little concerned about compressor temps......I'm not sure if that was the cause but on a couple of occasions the system has shut down without apparent reason and I guessed it may be a thermal shut off. So I now have an independent desk fan blowing onto the compressor.....no shut off's since but it still does get hot...can just about touch it and keep my hand on it for a bit...although not a comfortable temp.

[Orthello77]

Hey Drew, thanks man i'm actually enjoying running it without much left to do now, thought i'd never get there lol.

Yeah i can control the pumps manually via PWM , so i will set them to something lower than maximum and see at what rpm is the best setting. Two of those pumps in series will be as good as it can get i'm sure and like you say i'll have to check my crimps and hope not to blow an oring lol.

Compressor temps i have never noticed been anything but lukewarm. Will measure it after heavy load and see but yeah it doesn't seem to even get that warm. The pump base gets to 38c and thats way warmer to touch. Possibly i wonder if the large refrigerant loading is helping this ?? over 1300 gms.

Ok just out a probe on the compressor its 28c ... so its pretty trivial , inline with the pump been a lot warmer.

[technogiant]

I haven't temp probed mine.....but its at the limit of temp the hand can tolerate, which I think is 60c to 70c.....hope that doesn't mean there is a problem with mine.

[Drewmeister]

The thermal overload is set pretty high on these rotaries, upwards of 120c. You definitely wouldn't be able to touch it at 120c. The rule of thumb is, if you can keep your hand on it for at least 3 seconds then it's under 60c. Rotary compressors tend to run hot.. 60-70c is just fine. If it's exceeding 80c then might need looking into. The compressor also has it's own over current protection, might want to check the amps. Other than that.. maybe just something to do with the controller or sensor.

Orthello, you're correct about the heavier refrigerant charge. That combined with the buffer of the large evap res keeps the superheat steady at a very low temp. That results in much lower discharge pressure/temp to help cool a rotary compressor. I bet the sound levels are pretty good too.

Glad to hear you're finally able to enjoy the system now.. bet those oc's make a noticeable difference in demanding games. Btw.. which monitor do you have? The car looks sharp, sweet rims.. what do you have there?

Really stoked about your results as I was for TG's. Can't wait to see what you can achieve with the new pumps.

[Orthello77]

Yeah i really struggle to hear the compressor, especially over the fan which makes the most noise in the setup.

The monitor is a dell ultrasharp U2711 2560x1440. Its a great monitor with a fairly high DPI. The space taker in my garage lol is a Nissan Skyline R34GTT , its a lot of fun when i get it out - but thats the problem it has been a bit neglected lately. I'm thinking i need to sell it so i can move in the garage lol.

Nvidia sort of quashed my fun a bit with those drivers but for sure when i upgrade next i will get some really good results then. For now my CPU is miles higher than it was (i'll shoot for 4.6-4.7 once i get the new pumps) but the cards are only mildly so due to the driver limitation. Just played about 2 hrs of Max Payne and the compressor cycled during it .. that means liquid got to -27c while playing where i have it set for cycling - i didnt expect it would get there lol, its not putting out the load like heaven.

Pumps should arrive by thursday this week and if all goes well i'm hoping to drop 15-20c on the cpu under load conditions , hopefully both GPUs go subzero full time also so funtimes ahead !

[Drewmeister]

Sweet ride there Orthello.. you have great taste! The Skyline R33-34 and MKIV Supras are my favorite imports. I bet she's a blast to drive! Not sure I could ever sell a car like that.. would probably extend the garage or something lol.

Hopefully somebody will come up with a driver hack for those cards. I've been searching a bit myself... will search some more this evening.

[Orthello77]

Yeah .. i guess its house upgrade time ! lol (the next house will be tripple bay with workbench ) .. i do love the car , just don't get much time to drive it. Its a bit of fun , not been 4wd you can get yourself into all sorts of trouble with a bit of pedal etc .. Its not my daily driver unfortunately - maybe i need to talk to the boss ;-)

Thanks for the effort Drew re the cards, i've raised a few threads in Guru3D and EVGA forums and lots of users affected by the problem but no fix been reported yet. I'm hoping someone comes up with something as i'm planning on holding on to the cards for a while yet.

[technogiant]

Are you sure that the overclocking issue is driver related.....this review of the asus rog 580 shows them going to 1070 mhz with their tweak software.

http://www.hardwarecanucks.com/forum...review-16.html

Ooops...my bad....they are using an even older driver...nvidia 270.61 whql

Found another article about Nvidia and overclocking warranties......seems they are trying to lock things down from all angles.....perhaps an AMD card may be the way to go in future....8000 series just around the corner.

http://www.fudzilla.com/home/item/28...-void-warranty

[Orthello77]

TG , yeah its driver related. The drivers in that review were 270.61. Overclocking works right up to 296 drivers. After the 300series is where it was limited by Nvidia. I've tested myself with older drivers (296 drivers) i can clock to 1075 mhz no problems.

Yeah TG good spotting , i read that today on fudzilla re overclocking voiding your warranty and hence i guess we are seeing a new anti enthusiast attitude from Nvidia. Yes the AMD 8000 series does sound good in light of all of this.

By start of next year i think the need to upgrade will be more real then, new games - new engines and hopefully new cards by then also. If nvidia keep this up then it will be a no brainer to buy amd cards. Espeically with my investment in cooling i won't be limited by nvidia.

[technogiant]

Any news on them pumps your waiting on Orthello?

[Orthello77]

Well they shipped on the 26th according to USPS, still waiting to see them in NZ .. i just hope customs havn't picked on them .. if that happens then it will take another week and have GST added by the time i get them :-( .. hopefully they are just running a tad late.

Apart from lack of flow rate, the systems doing well, got down to -31c liquid yesterday and a -20.5c chamber. CPU is doing fine at 4.5ghz and i'm hoping to get to 4.6-4.7 with the new pumps. Still no luck on getting past the Nvidia 1000mhz limit on the 300+ drivers unfortunately, no hacks or work arounds found yet.

Hoping the pumps arrive early next week now ..