Ok, I'm Chilled. I'm Chilled. Relax....56K Warning!

[Naja002]

Well, I've had a lot of fun with my other projects, but I realized a while back that a chiller would probably be my best bet all the way around. So, after doing the car rad(s)--indoor/outdoor, geothermal, bongs and HVC....I finally decided to do the chiller-thing. Of course, the chiller scraps everything else...except the indoor rad--I can still use that for winter heat.

I started surfing around checking out this years A/Cs, rechecking on coolers, etc. I wanted one of the Coleman 150qt. 5-day coolers....but I ended up going with an Igloo Polar 120qt., because of price and availability. Still a good cooler, but the Coleman 5-day's (Extreme/Marine) reviews are outstanding!

Anyway, I ran into a snag with the window that I had always intended to mount the A/C in (yeah, I've been thinking about this for a long time)--most/many of the smaller A/Cs require a minimum 14" window opening (height), but there are a few that don't. In the end though it didn't matter, because I said screw the window, I'll pull the rads and reverse the fans and blow some of the heat out the window.

So, while still trying to talk myself out of this for at least a while longer ( )--I started surfing craiglist. I found a small Frigidaire 5KBTU locally for $50. Said: "Screw it". Picked up the cooler. Picked up the A/C. So, this was done....just a little assembly required...

So, this is what I've come up with:






Plumbing coming out of the chiller headed toward the PCs:





Return to the chiller:




I really wanted electronic thermostat/controls and "Energy Saver", but I would have had to buy a new A/C, so that would have about doubled the cost.



I cut a small opening in the lid for the wires, etc to pass through:



Filled it with weather stripping foam just to help prevent the transfer of air:




Marked my spot for when I close the lid:




Pix of the guts:







I added a small 11w 200gph pump to help circulation inside the cooler. It definitely helps even out temps:



2 temp probes side-by-side:




These are the low end temps that I have it set for--the compressor kicks on:





And then kicks off at:





I set this up in the morning and it ran fine all day. No problems whatsoever. About 20 hrs later--I had alarms going off. I checked the coolant temp=116F/46C. I'm not 100% sure what the cause of it was, but I think and guessed that it was the temp probe. I'm assuming that the part exposed to the outside of the cooler was effected by the low ambient temps--so the compressor didn't kick on. So, I put some tubing over the probe length to help it read the temp inside the cooler only. Seems to have worked:


I haven't had any problems since....


When the compressor is off the unit uses 47-48w for the fan motor. When the compressor is running it uses right at:




I took some on/off cycle times. The off cycle averaged 11min 55sec and the on cycle averaged 12min 12secs. So, for me that's close enough to say that it's going to run ~50% of the time. Problem is: that's ~5 cycles every 2 hours or 60 cycles/day. Seems like a bit much. So, I added 6 more gals of water. Hopefully that will help it cycle just a bit less. I encountered a small leak toward the top of the return bulkhead fitting. So, before I can add anymore (water or anti-freeze) I am going to have to remove a bit of the fluid and tighten that sucker up a bit more. Once it's sealed properly I can add quite a bit more fluid. Physics says that there isn't much I can do about the 50% run time. But it also says that more fluid mass should reduce the number of cycles/day.


Here's a 38sec video showing the circulation and return flow:




No, I'm not interested in sub-zero, etc. I'm sub-ambient now....just above the dewpoint<--worst case. Hoping I stay just above the dewpoint.



Cost:

A/C--$50
Cooler--$57
Anti-freeze--$77
I had everything else.



Other then a few very minor issues....I really didn't have any problems setting this up. My estimate is that it should cost ~$20.29/month to run this, so:

Looks like I'm ready for summer!


ApogeeGT water blocks:










Goes from the chiller/pvc to this manifold and then to the PCs:



Crunchin' Boinc: WCG + GpuGrid:






Index of prior Projects:

Yes, Another Car Radiator Thread....Major 56K Warning!

Bong! Bong! Bong! --56K Warning!

Horizontal Volenti Cooler.....56K Warning!

The Hole…Geothermal Loop…56K Warning!

[antiacid]

this looks fun!
I have a question to you, considering you have a lot of experience with those "alternative" cooling methods: What would you say was your most quiet setup? How would you modify your best performing cooling system to make it more quiet and at what cost?

[Naja002]

this looks fun!
I have a question to you, considering you have a lot of experience with those "alternative" cooling methods: What would you say was your most quiet setup? How would you modify your best performing cooling system to make it more quiet and at what cost?

Honestly, that's a pretty difficult question to answer. For myself--they are all pretty quiet considering that the heart of the cooling system is out in another building. What I would focus on is setting up my rigs differently, because the fans in here are actually pretty loud--it sounds like a server room in here. Some of that is the 2 air conditioners, some of it is the fans blowing under the shelf into the area where the PSUs and HDs are.... I have 14x 80mm fans running full volt, 5x 120mm, 2x clip-on fans, 2x car rad fans @ 5v and 2 air conditioners. That's including the old Koolance case setup and the "indoor rad".

For someone else--it's much more difficult to answer. If all of the resources are available--and quiet is the only goal.....then it would have to be the Bongs and/or HVC. They can be setup to be almost silent. They can go well below ambient. But there are trade offs like disinfection. But again: If All the resources are available--I would just setup a chiller in another building--preferably with the A/C in a window so all of the heat pumps directly outside. Disinfection is a minor issue. Cooling the heat produced is non-existent. You can maintain around ambient, below ambient, sub-zero. $20-30/month is affordable for most people, I would think.

I've known for quite a while that I was going to wind up with this chiller. But the other setups were some pretty cool stuff and it's kind of difficult to just scrap it all in one fell swoop. But that's what I've done. It's all been a lot of fun. A lot of learning. A lot of sharing. I've copied my logs onto other forums, so if any one forum goes down for the count.....the heart of the info should be around for the foreseeable future.

But if quiet is the only rule: It would definitely be the Bongs and/or HVC....




Just a quick update:

I fiddled with the 1 small leak some. Seems that I've managed to get it stopped.....or 99.9% stopped. Not sure yet. But I've dealt with aquariums, etc for many years and know that this tiny leak--can be a royal PITA! But I'll get it fixed one way or the other.
Adding the 6 gal of water seems to have altered the on/off cycling to ~20mins or 3x/2hrs or 36x/day. Once the small leak is fixed with certainty--I plan to add 2 more gals of water and 2 more gals of anti-freeze....bringing the total to ~24 gals. Hopefully that will drop the cycling to every 30mins or 24x/day. Still a lot in my book, but without adding another cooler--that's about as good as it's going to get!

Covering the temp probe with tubing seems to have done the trick--no issues since...

[antiacid]

I fell on a link to a "bong lovers" club from one of your threads... This looks like a fun project. I think I'll take it up right away and see how it works out!

I wonder what kind of volume of water is evaporated in a 24 hours period or in an 8 hours period? I'm not sure I'll be able to use an auto-level mechanism on my test bong but I'm sure I could manage to get a very large reservoir or something...

[Naja002]

The amount of water evaporated is going to be effected by many different things like: water temp v. ambient temp, relative humidity and so on. If you put your pump as low as possible in the res it will offer more headway against evaporation. Hooking up to the plumbing for a sink, shower, toilet, washer, etc is not difficult to do. One small hole through a wall can be all that's really needed to run a line of 1/4" ID ice maker tubing to a float valve. Odds are you have incoming water in the next room....there's the source. Holes through drywall are easily fixed.

There are ways to setup a 2-liter bottle, gallon jug, 5gal bucket for short-term auto-refill.....

[eBoy0]

I had my chiller setup similarly, boy did it rattle like crazy resting on that 2x4.

[railmeat]

nice setup man.

im just curiuos tho at your 920 @ 4.1 @ 68c,that does not sound all that good,no offense.but ppl can get that on regular water i suppose.i dunno??is that screenshot with the cpu loaded(primed,occt in the background?) i missed somewere.or is that good for a 920 at 4.1 on chilled water???.


and why are your volts that hi?they needed that hi?
guy below in the link i pasted is it D0 at 4.0Ghz (1.3v vcore)
http://74.125.47.132/search?q=cache:...&ct=clnk&gl=us


at intel specsheet Thermal Specification on 920

if u clik the link and clik thermal specification it says 67.9 for 920,s????
http://processorfinder.intel.com/det...x?sSpec=SLBCH#
and your at 68 core 1 idle?

"Thermal Specification: The thermal specification shown is the maximum case temperature at the maximum Thermal Design Power (TDP) value for that processor. It is measured at the geometric center on the topside of the processor integrated heat spreader. For processors without integrated heat spreaders such as mobile processors, the thermal specification is referred to as the junction temperature (Tj). The maximum junction temperature is defined by an activation of the processor Intel® Thermal Monitor. The Intel Thermal Monitor’s automatic mode is used to indicate that the maximum TJ has been reached"




read this I7 920 @ 4.0ghz temps

"I have a D0 at 4.0Ghz (1.3v vcore) and I hit just over that running OCCT with a large dataset, high priority, hyperthreading on. I max at about 86C in OCCT, Coretemp reads it at 83C. Idle I'm around 53-55C OCCT, Coretemp reads it at 50-53C." read link below

http://74.125.47.132/search?q=cache:...&ct=clnk&gl=us




your Yes, Another Car Radiator Thread IS AMAZING,HAHAHA!!!!

[Naja002]

I had my chiller setup similarly, boy did it rattle like crazy resting on that 2x4.

It's really not all that loud considering that it's a low end air conditioner. There is a slight rattle when the compressor is on, but it's nothing obnoxious. The fan runs 24/7, and when I go out there I pretty much have to look at the killawatt to tell/verify that the compressor is running....not just the fan.

nice setup man.

im just curiuos tho at your 920 @ 4.1 @ 68c,that does not sound all that good,no offense.but ppl can get that on regular water i suppose.i dunno??is that screenshot with the cpu loaded(primed,occt in the background?) i missed somewere.or is that good for a 920 at 4.1 on chilled water???.

and why are your volts that hi?they needed that hi?
guy below in the link i pasted is it D0 at 4.0Ghz (1.3v vcore)
http://74.125.47.132/search?q=cache:...&ct=clnk&gl=us

Just to point out: I'm not sub-zero, I am sub-ambient. Most folks go sub-zero with a chiller, so others see "chiller" and automatically think "sub-zero". I'm currently running water that is 63-68F/17-20C.

I think the main reasons for the temps and volts are:

1) I'm still running on an ApogeeGT Water Block. It's a bit out-dated, but I picked up a couple of holddown plates a few months back from Duniek. Trust me: A heatkiller Cu would drop that temp a bit. But with 3 rigs on the same WCing system--I have to be aware of block flow/restrictions. Mixing and matching blocks can wreak all kinds of havoc.

2) It's a C0/C1, not a DO. So, it doesn't OC as well at the same volts.

3) It was a fairly new OC and could still use some "tweaking". In the past: I was at 4.0 for a short while (few days). Ran 3.8 for 2-3weeks until I knew summer temps where going to force me to lower the OC for the summer. So, I moved it down to 3.4 for a week or so until I got this chiller up and running. Then I went back to 4.0, and then up to 4.1. Checked out 4.2+ and decided that 4.1 was good enough....at least for now. I'm not an OCer. I'm not a Bencher. I'm a Cruncher (Boinc: WCG + GpuGrid) that stumbles along in the OCing world. That said:

4) The volts are set at 1.38 (?). But go up to the 1.424 shown when under load. Truth is: If I could remember, or even knew, what half of the bios settings did....I would do a bit better. I'm not totally clueless, but I could definitely use some "growth" in that area. From what I have gathered from looking around 1.36-1.39 is not unreasonable for 4.0 C0/C1. Maybe I'm wrong.....

5) This i7 is a 24/7 cruncher and my everyday rig. I prefer (usually) a bit more voltage so things are "snappy" and respond well. I don't like limping along on the lowest voltage that I can get by with. Why it moves itself from 1.38 to 1.424 under load--I don't know. But the 1.424 was the same at 4.2+......

at intel specsheet Thermal Specification on 920

if u clik the link and clik thermal specification it says 67.9 for 920,s????
http://processorfinder.intel.com/det...x?sSpec=SLBCH#
and your at 68 core 1 idle?

"Thermal Specification: The thermal specification shown is the maximum case temperature at the maximum Thermal Design Power (TDP) value for that processor. It is measured at the geometric center on the topside of the processor integrated heat spreader. For processors without integrated heat spreaders such as mobile processors, the thermal specification is referred to as the junction temperature (Tj). The maximum junction temperature is defined by an activation of the processor Intel® Thermal Monitor. The Intel Thermal Monitor’s automatic mode is used to indicate that the maximum TJ has been reached"

Well, that confuses me. I've always thought that the thermal specification would be like the high-end "safe" temp...or limit. But my understanding is that 90C is the new 70C (1366 v 775), and that quote seems to have something to do with: Intel Thermal Monitor’s automatic mode. The whole purpose of this chiller is to maintain what I consider a reasonable OC at a reasonable temp (70C or below) 24/7 throughout the summer. So, maybe somebody would care to breakdown what that actually means for me?

read this I7 920 @ 4.0ghz temps

"I have a D0 at 4.0Ghz (1.3v vcore) and I hit just over that running OCCT with a large dataset, high priority, hyperthreading on. I max at about 86C in OCCT, Coretemp reads it at 83C. Idle I'm around 53-55C OCCT, Coretemp reads it at 50-53C." read link below

http://74.125.47.132/search?q=cache:...&ct=clnk&gl=us

Ok, I read it, and I'm not sure what you are getting at..... , but here's the full quote:

I have a D0 at 4.0Ghz (1.3v vcore) and I hit just over that running OCCT with a large dataset, high priority, hyperthreading on. I max at about 86C in OCCT, Coretemp reads it at 83C. Idle I'm around 53-55C OCCT, Coretemp reads it at 50-53C.

For some reason Coretemp(and realtemp) register my temps 3C lower than OCCT does.

I run perfectly stable, not sure how you can get a 4GHz OC, even with a D0, and not go over 70C full load.

Emphasis added.

I've not checked out OCCT. I OC for 24/7 crunching stability and on this i7--I also use it for everyday use. As far as temp readings--which is right? Realtemp seems to be the standard, with Coretemp = or a close 2nd. Then there's Everest, OCCT and what else? I go by Realtemp with a 100 TjMax as I think most do these days. From what I can gather those guys are referring to air, not water. But the last statement sums things up pretty good: It's not easy running 4.0 or above under 70C.....and again: I'm using an ApogeeGT block--that says a lot.

your Yes, Another Car Radiator Thread IS AMAZING,HAHAHA!!!!

Thanx! It was a lot of fun. I definitely learned a lot about the functionality of it all during the process. Sad to see it go, but I've moved on to this chiller..... But No Fear--I am keeping the car rads for now, because: I have a Plan....



A bit of an update:

Putting the tubing over the temp probe seemed to help, but only some. Ambient temps in the out-building directly effect the mechanical thermostat, so when temps outside (/in the out-building) get below the set temp--the mechanical thermostat shuts the compressor off and leaves it off. I've thought and thought, tried to insulate the thermostat, etc. I have quite a bit of experience from my +30yrs dealing with reptiles, aquariums, etc. I cannot see any reasonable, simple way to overcome this. So, I went ahead and bought 1 of these:



Ranco 111000

I looked at the Johnson Controls. I have a couple of Helix Proportional Controllers, but they are Proportional, so they won't work in this situation. Anyway, the Ranco have been around for a long, long time. They are used by zoos and aquaruims worldwide. Many, many animal keepers (public or private) use them. Convenience stores, restaurants for their coolers/freezers. In all: they are tried and true. Very reliable units.

They offer heating or cooling (not both on single stage), temp range from -30 to 220F, and a Differential range: 1° to 30°F.

They are designed for use with resistive loads (motors) and will resolve this issue 100%. The differential range will allow me to set the temps at which the compressor kicks on and off. With the mechanical thermostat it's 5F. Now I will be able to set it anywhere from 1-30F.

I'm hoping it gets here this coming week. I did not buy the wired shown, so I need to pick up the wiring, get it wired up and installed. Until then: My water temps are 63-68F and humidity stays ~50% in here, so condensation starts forming at about 84-85F/30C in here. I'll have to keep the A/Cs running until I get this setup. I usually turn them up to 82F when I'm not here/working. The ambient right at the PCs goes up to 85-86F which is too high=condensation.

Fingers crossed on no more major issues between now and then....

[Plan.B]

I love a good 'ol a/c convert. and it crunches! nice implementation and craftsmanship, again.

[Naja002]

I love a good 'ol a/c convert. and it crunches! nice implementation and craftsmanship, again.

Me too! I should have done this a ways back, but I've enjoyed the other projects.


Well, the controller arrived a couple of days ago. I had picked up the wiring--went with 14 gauge. Call me crazy, but I would rather burn up the controller then burn down the building! I'm just kinda quirky like that! I got it wired up, no problem, but had 1 minor issue. While waiting for it to arrive, I learned that the probe (sensor) is not supposed to be submerged. Well, that kinda screws things up! So, I had a choice:

1) Purchase a $30 "Bulb Well Assembly", pay another shipping charge and wait a bit longer, Or....

2) Improvise.

I chose #2. First was coming up with at least a temporary fix. I considered siliconing, epoxying, etc the sensor and wiring, but I really didn't want to do that. So, I got out a couple of aquarium heater holders and some pvc:



I knew this was going to be temporary, so I glued the bottom submersed cap on, but not the top cap. Drilled a hole in the top cap that the sensor would squeeze through and then babysat the system for a little while. Actually worked pretty good really. The thermal conductivity of pvc is like 0.19 or something low, so I was curious what my temp range was going to be. I set the differential for 2F--it turned out not to be more then ~4F which is actually pretty darn good. But not good enough.....

I knew I had a +1Ft piece of 1/2" copper pipe laying around, but also knew that I did not have any caps for it. So, a stop by lowes fixed that and while I was there I went ahead and picked up a rubber grommet. Yesterday I was pretty tired and I knew things could hold another day, so I procrastinated until today. Got my butt out there and got this fix fixed!

Sweated on the bottom copper cap:




Got the grommet over the sensor and a hole drilled through the top cap:




Even though I could have sweat the top cap on and inserted the probe after things cooled--I knew that some time in the future I would want to remove that probe--the hole is tight, so I can't necessarily just get it back out. I chose to silicone the cap on and silicone over the grommet and wiring:



I have no intention of submersing the top of this thing, so the silicone should do a good job of keeping the moisture out long term--that's all that's really needed.

So, next I needed to mount it into the cooler. Why not go with a good thing already? I knew that the heater holders would not clip onto the copper pipe because it's just too narrow. So, time to improvise again. Took a small scrap piece of 3/4" pvc, cut it into 4 pieces, clipped 1 piece into each heater clip and slid the copper setup right down in there:



Picked up a couple more gals of anti-freeze at walmart when I hit lowes, so I added that:





So, now there is ~7.75 gals of anti-freeze and 17 gals of water. I need to do another time study to see how long my on/off cycles times are now. I doubt I'll do it today, but hopefully some time this weekend.


Just a larger view:







Oh, yeah, I fixed the return bulkhead leak by loosening the bulkhead and siliconing under and all around it. Not the preferred way to go, but useful in these types of situations. The preferred route is no leaking bulkhead to start with....



So far so good:



Max temp over the last 30+ mins is 70C. The mins kinda worry me though. I may move everything up a couple of degrees after everything settles in. I would rather move my max temp up to say 75C than deal with condensation!

Oh, and I lowered my OC down to ~4.0 for now.....



More Updates at Post #46

[[XC] gomeler]

Great stuff. I'll be following in your footsteps, sort of, very soon. Really looking forward to a large reservoir but I think I'm going to make a dewar of sorts out of two plastic barrels.

[Naja002]

Hey, Gom--Thanx. As mentioned in the other thread--I still have that blue barrel--BUT it's thermal conductivity is toooo much. Yes, it's plastic, but it will absorb and release a ton of energy. I learned that when I was using it with my car rads--and at 1 point when I used Only the blue barrel--no rads, no nothing--just the barrel.. My best advice is truly to go with 1 or 2 large coolers. The Coleman marine 5-day coolers are excellent and ~ $82 plus tax at walmart--but ya gotta order it in. They only keep the 120qt polar like I have in store ($52 + tax). I have considered plumbing another polar to this one and doubling my fluid mass. I have the bulkheads, but I don't think I will be doing it any time soon--if at all. First thing is another time study of my current setup.

Skip the blue barrel--go with a cooler!

[[XC] gomeler]

I haven't got the space for the coolers. I know the barrels will leak a lot, hence why I want to try stuffing a smaller one inside of a larger one and pulling a vacuum on the space in between. Should provide fantastic insulation, same concept that keeps those Coleman coolers insulated

[Naja002]

Sounds Good. I wasn't aware of the 2 barrels, vacuum et al, but, Yes--sounds like a plan worth checking out!



Sheesh. Now that it's done....I should have just picked up some of that Plasti-dip stuff and used that for the probe.

[[XC] gomeler]

same concept thermos bottle uses also add some styrofoam peanuts between tubes/container

Yup. Need the stuffing material especially to keep the inside barrel stationary. Guess I should stop thread-crapping and start up my own thread

[Naja002]

Yup. Need the stuffing material especially to keep the inside barrel stationary. Guess I should stop thread-crapping and start up my own thread

No sweat. I don't expect to be going that route any time soon, but it is good food for thought for the future. I don't own a vacuum pump, so I usually don't even think in that direction, but......I may before long.

[Naja002]

Here's a question: How can I tell if the fins on the evaporator are aluminum or stainless steel? I've assumed they were aluminum, but realize that they could be stainless. I've looked at them, but it could go either way.

[[XC] gomeler]

So long as the stainless steel doesn't have a high nickel content you can use a magnet. If it is steel it'll stick, aluminum won't. You can also tell by weight, and ease of bending.

[Naja002]

Thanx. gom. A magnet, huh? It's a low end AC, so I'm assuming it's aluminum. My only thought was the money spent on anti-freeze....with stainless I don't think that I would have needed it, but it may be good to have it in there anyway.....

[[XC] gomeler]

What is the concern? The aluminum/copper mix in your "loop"? I thought adding antifreeze solved that?

[Naja002]

What is the concern? The aluminum/copper mix in your "loop"? I thought adding antifreeze solved that?

It did. I was just thinking that maybe I wasted $77 on the anti-freeze when I may not have needed it. However, until I get around to checking the evap fins--I am definitely going to assume that they are aluminum. Also, trying to get all of this stuff logged into my head for future thinking..... I knew that some evap fins are made of stainless.....just never crossed my mind until after everything was completed......oldtimer's syndrome, I guess.

[[XC] gomeler]

It did. I was just thinking that maybe I wasted $77 on the anti-freeze when I may not have needed it. However, until I get around to checking the evap fins--I am definitely going to assume that they are aluminum. Also, trying to get all of this stuff logged into my head for future thinking..... I knew that some evap fins are made of stainless.....just never crossed my mind until after everything was completed......oldtimer's syndrome, I guess.

I'd say there is a slim chance that they are steel, most that I have seen are aluminum. The anti-freeze wasn't wasted either, now you can go sub-zero

[Otis11]

Well, thing to be careful of, even if it's aluminum the magnets may stick - the tubing inside of those fins are copper. Don't know how to check that...

And reducing the number of cycles is simple...put 3 more computers in the loop. Only 1 cycle/24 hours- ON.

[rpg711]

Well, thing to be careful of, even if it's aluminum the magnets may stick - the tubing inside of those fins are copper. Don't know how to check that...

And reducing the number of cycles is simple...put 3 more computers in the loop. Only 1 cycle/24 hours- ON.

Considering there's only 3 magnetic elements... I think he's safe with aluminum and copper.

[Naja002]

Well, thing to be careful of, even if it's aluminum the magnets may stick - the tubing inside of those fins are copper. Don't know how to check that...

And reducing the number of cycles is simple...put 3 more computers in the loop. Only 1 cycle/24 hours- ON.

Considering there's only 3 magnetic elements... I think he's safe with aluminum and copper.

Just wanted to say that this is a nice "set it and forget it" project. Since everything has been setup, checked out and set.....it just runs itself. No babysitting required.