Overclock.net banner

The 24/7 Sub Zero Liquid Chillbox Club

94K views 849 replies 60 participants last post by  [ShowMe!] 
#1 ·
Hello Fellow Chiller's,.

The first attempt at defining this club of extremists was too broad so we have narrowed the focus down to 24/7 subzero liquid chillboxs.

The 24/7 sub zero liquid chillbox club is a place to discuss and catalog the concept of liquid cooled computers that operate under sub zero conditions inside of a chillbox for the purpose of extreme overclocking for daily use instead of benchmarking...

Whats a Chillbox?

1. The chillbox concept came about as a way to overcome condensation issues that arise when cooling anything below the dew point. The dew point is when a cold object starts to collect moisture on it surface due to the temperature difference between the ambient air and the cold object, something that you do not want to happen on your computer components.

2. The concept of the chillbox is fairly simple. Its an air tight and insulated box, that houses your liquid cooled cpu/gpu/motherboard along with a radiator that chills the inside air to near the temperature of the liquid cooled components.

3. During typical sub zero operation, any moisture that is left inside the sealed chillbox, will deposit it self on the radiator that is chilling the air inside the chillbox in the form of small ice crystals, and you are left with is cold dry air.

4.This allows standard water cooling components to operate under extreme sub zero temperatures.



2012 : GTX 580 SLI rig.
20120722003113.jpg

2012 :
20121024211652.png

2015 : TX SLI and Tripple Pane lid = no condensation in the viewing panes.
Further Considerations when building a chillbox are :

1.) What inlets and outlets you will require for any cabling and liquid , because the chillbox needs to be as air tight as possible these inlets and outlets will need to be sealed and air tight once the cabling and piping has been put through and hence adding extra cabling or piping later can be difficult so its best to be really thorough at the start of your build.

2.) Clearances - Do you have enough space in your chillbox for your motherboard and all devices and most importantly a decent sized radiator and fans to act as the air chiller. Allow a bit extra - I nearly had to shave my viewing pane to allow the size of the classified water blocks I run. Remember you are building this for the future devices you may use also.

3.) Do not include the PSUs into the chillbox as they produce waste heat but you would route the cabling through to the chillbox from the PSUs and take it from me - allow as much length in that cabling as feasable. My build worked out very tight with the power supply cabling. My chillbox was an over under system, over was the chillbox and located under were the Piping / PSUs / HDDs etc.

4.) Materials for the chillbox. Possibly the cheapest chillbox is a converted beer chillbox or fishing chillybin etc. These are already insulated so that is a good start. They are reasonably air tight if they have a good gasket near the lid , this can be improved. You can also relatively easily drill through the walls of these chillybins so it makes putting cabling and piping through rather easy.

The negatives to this approach - you can't see anything (the reason you need too comes below in (6)). You can alleviate this by installing an internal camera or modifying the chillybin to have a viewing pane, believe me you need eyes into that chamber . I regularly check my chamber on startup for leaks/problems and i would recomend if you are using a chillybin without a viewing pane that you install enough cameras so you can see all the angles etc. Once its running you will not be able to open it so its better than having blind faith that everything is ok.

A more expensive and time consuming approach is to make a custom acrylic chillbox like i did. This requires insulation to be added to shield the acrylic from the cold , although this is mainly for condensation purposes. I have had my chillbox down to -35c in winter and the acrylic viewing lid which is not insulated did not crack, acrylic seems well suited to the temperatures that a chillbox can obtain.

5.) The radiator acting as the air chiller needs to be the first component in your liquid loop once inside the chamber. This is so the air from the radiator is as near to the liquid temperature as possible. If you have the radiator last you will add extra delta to the air over the incoming liquid temperature so therefore increasing the risk of condensation. It pays to install a humidity sensor in the chillbox, i watch as mine goes from 70%+ on startup to -- (below 23%) after a few minutes. If there is a liquid leak in there the humidity will alter as a bit of the liquid evaporates thats also a good double check for leaks.

6.) Reasons for the viewing pane or cameras where a vane is not possible.

As the 1st post in the thread mentioned the viscosity of what ever liquid you use will increase as the temperature drops as the liquid gets closer to becoming a frozen solid, this will be harder on you pumps, in turn you will need more pumping power to overcome the viscosity of the liquid and therefor you will put more stress on any liquid connections / liquid blocks in your chamber as the PSI restriction increases through the components. Due to this i recomend wormscrew clamps and barbs. Compression fittings for this sort of application i would only recomend for direct connections where the pipe is not bending at all , compression fittings do not have same clamping power IMHO. Orings seem to be fine - i've never had a leak through an Oring , if you want to be sure you can remove the Orings and metal expoxy the threads to the barbs , i have done this one some devices as initially i did not trust the orings under high PSI flow this turned out to be a non issue.

A simple tip wrap all your fluid connections in paper towel, use a non clear antifreeze or some liquid which is coloured and that simple system will allow you to see a slow leak more easily , if you have a slow drip of a leak and your viewing pane gets frosted over within a couple of minutes or your cameras are not up to scratch this will allow you to see a leak very easily.

What if a fan dies and your airflow reduces ? This has happened to me , after all they are in the freezing cold environment. Without proper air flow condensation will build up on your blocks and pipes. Hence another reason to have eyes on the fans or measure the rpms reported via software in windows etc.

My build of the chillbox in pics.


Acrylic chamber which I designed in sketchUp and sent to the local acrylic people and they worked their magic. Note the Gasket lid - this was much heavier than the rest of the chillbox at 20mm thick which worked out well as Its nice and heavy and compresses the gasket prior to screwing in. The rest of the chamber walls were 10mm thick which allows a little flex (not a bad thing due to negative internal air pressure) .The pipe fittings are round incuts , tripple layered for gluing surface. The cable incuts are rectangular. Unless you want to do a lot of soldering then make the incuts into the chillbox frame large enough to take the biggest cable header. You will have to silicon in or use space invading foam or other filler to fill in the gaps between the cables once through. Ideally if it can hold liquid and not drip once you have put cables through you have got it right.

https://imageshack.com/i/g420120622122321j

Added base layer of insulation. The insulation I used was a variety of closed cell foam , but much harder than normal foam , but still able to cut with a craft knife or razor blade. Put side layers on. I had to take particular care to make sure all the levels around the lid were even. Eg the insulation on the inside fitted nicely against the gasket but not so hard that it pushed the gasket out from sitting flat on the rim. The gasket lid has screw holes around the parameter for screwing to the rest of the base. The gasket material is common door jam type stuff ~ 10 mm wide then pushed into the recess of the gasket.

https://imageshack.com/i/jv20120701205255j

Outside insulation added. So the chillbox has internal and external insulation. Overall the walls and bvase are around 80mm thick. Used ados to adhere the insulation to the plastic , this has pulled away in one or two spots internally with the negative air pressure but overall its still fine and doing its job. Note the base insulation and base fibre board have incuts for cabling and piping matching the chamber. Acrylic gravity cradle is glued in place for the radiator.

https://imageshack.com/i/mz20120703225011j

Radiator in place, initially I used a HW labs GTX360 .. this was overkill , in the end I used an XSPC 3 fan radiator that had the lowest restriction of any radiator I could find at the time. I would recommend you use a LOW restriction radiator as its not really going to cool any great wattage if you CPU and GPUs are directly waterblocked. EG the radiator cools the air and maybe the ram and mobo etc ... 200 watts I would think - if that. Really the only reason for the big rad is to get nice even coverage for the air cooling.

https://imageshack.com/i/md20120704201253j

Motherboard ATX frame. I could not find one of these to buy so hacked it out of an old case .. took a while but it was worth it. Note the incuts into the chiller frame for liquid and piping. Ideally I would have added a few inches here and there if I was to design another.

https://imageshack.com/i/mr20120707230118j

Ambient air testing Pic1.
https://imageshack.com/i/f620120708232728j

Ambient air testing Pic 2.
https://imageshack.com/i/cb20120708235655j

Chilled liquid testing (over a year ago .. with older hardware I might add. Two GTX580s @ 1070 mhz. Core I7 930 @ 4.75 ghz - after a bit of trial and error). Note the frosting on top of the viewing pane. It definitely needed a 2nd viewing layer to trap some air and therefore condensate less on the gasket layer.

https://imageshack.com/i/n520120722003113j

The making of the Viewing lid that sits on top of the gasket lid. This basically reduces condensation on top of the gasket lid.

https://imageshack.com/i/7e20120725214227j

Finished the viewing lid and testing. In this pic I am using the original piping config which had a valve system for allowing prechill of the air by running liquid through the radiator only prior to running liquid through the blocks. In the end I got rid of this as I found I did not need to prechill the air prior to running liquid through the system. This improved my flow rates - running this method however its critical to make the radiator first in the cooling loop.

https://imageshack.com/i/mk20120725224609j

Pic of the sensor inside the chillbox , note the air temperature is -17.6 with -- humidity (below 22%). This was with really bad liquid flow too , now days the air temp is typically in the low -20s. Pumping issues plagued me for nearly 6 months .

https://imageshack.com/i/g020120726220507j

20120902155901.jpg

20120902155854.jpg

The above way is only one way to build a chillbox , there will be many . I hope with this thread we will be able to experience a lot of different builds to come and share the ideas.

What are the Challenges?

1.) Creating a liquid chiller (normally by converting an air conditioning unit to chill liquid) and creating the air tight chillbox to house your components. Liquid piping for cooling and cabling must be routed through the chillbox whilst maintain the chillboxs air tight seal.
Building the chiller - using a common household Air Conditioner (A/C) .

This was done over 3 years ago so hence its retrospective look at how i built my chiller at the time. For the full build with more text read pages 1-7 here.
http://forums.extremeoverclocking.com/showthread.php?t=363597

For those with limited time , this is the non TL;DR version. :)

The best type Of A/C to mod is the "window banger" or window mounted A/C . This is because the evaporator is closely located to the compressor and the most compact chiller can be built with less modifications needed vs other types of A/Cs. Dehumidifiers can be used but are less suitable as they have less cooling power and the refrigerants are often boiling at a higher temp than an A/C.

The size of unit to aim for is really hard to say , my unit was an R22 unit (You really need a refrigerant that boils at -46c or lower such as R404a etc) with 1300gms of charge and rated at 12K BTU of cooling power. I would think 8k BTU is too small but it really depends on heatload and how efficient the unit is. 5k etc is really not going to power a full system and chillbox to any great subzero temps under load IMHO - overshooting here is best, you can always thermostat off your chiller if needed if the cooling exceeds your full load temperature requirements.

Spec Shot:
219423591full.jpg

One of the large factors for suitability of the Mod is how the suction lines (copper pipe lines) carrying the refrigerant are located.

In a perfect world these would be up high so your Reservoir is not compromised by having piping going through it at mid wall height.

Successful builds have been done with low lying suction lines but they are more difficult to seal. If at all possible get an A/C with high suction lines. My unit was not quite ideal due to the width of the incline on the suction lines but it still worked out OK.

If you have to breach the resoiver walls with the piping i recommend a solid interface - eg welded or rigid. Soft interfaces eg silicon types etc i have found to leak after a time due to vibration and sealants bonding to copper is not great.

Side Shot: Suction pipe orientation is ok but not ideal.
20120615222212.jpg

Condenser Shot :
20120615222240.jpg

Initial suitability tests : Tape up / block off air outlets and turn A/C on 100% cooling mode , remove temperature sensors so the deicing routine does not kick in. If the unit does not get to -25c or below like this then there could be an issue about its suitability.

20120617103510.jpg

20120617112111.jpg

20120617112616.jpg

With the tests above past , we can move on to removing the A/C evaporator shrouding and fan. There can be be 3-4+ hrs work in this as the shrouding can be really hard to remove. I literally had to nible away at the plastic and break it off bit by bit. Hacksaw off the evaporator fan end.

20120620210937.jpg

Next up - build the resoiver. I started by using a large fishing lidded chillybin and cut out a notch for the suction lines. Chillybins are usually well insulated so its a quick way to start a res. I had to make sure the evaoporator height could be contained by the chillybin To cover the evaporator the chillybin was rather large but i used the excess space for extra insulation.

20120623180104.jpg

Evaporator size tested inside the chillybox.

20120623183802.jpg

20120623183836.jpg

Once the size testing was done i set to makinga chillybin base. I was building the chiller in my garage so in the end it would need to be carried outside (it took 4 poeple lol), hence the handles.

20120626213943.jpg

A/C mounted to frame on base and first layer of foam insulation added to chillybin.

20120630184047.jpg

Added copper piping through resoiver walls near top and siliconed in place.



Walled in excess space with polystyrene and silicon.. 3x90mm walls effectively and 1x120mm wall . Excessive insulation for sure but the chillybin i used was way to big to fill it all up with liquid as liquid pull down would take a lot longer with 80 litres than 40 litres.

Polystyrene will not desolve with methanol / glycol so its safe to use as a insulating packer. Just watch Terpentine near it though if cleaning anything eg cleaning tools etc after using silicon. Terpentine will desolve polystyrene !!



I later cladded this over with aluminium ...

mXsmvu.jpg

... But it wasn't as good as i thought as it created icing around the lid in the air gap. How it looks today with aluminium cladding and foam top seal to lid. Just after the mod so a floater or two in the pic that i removed with suction filter shortly after. You will need to add inline filters to your chiller liquid lines , otherwise your GPU and cpu blocks are going to become the liquid filters and clog.
redface.gif




Skipped forward 3 years in the above pic so back to time of build
wink.gif
now .

Resoiver filled. Ideally fully immerse the evaporator - this will cancel out any rusting that could occur near the top of the evaporator.



Insulating Suction lines.



Installing 2nd lot of pumps , my first was a Little Giant industrial pump that was rather woefull or underspecced for the job . Below 2 x MCP355x2 in series (4 pumps internally) = a lot of pumping power !!



Current config for comparison : Shown the new config below, as the pumps above eventually died due to condensation and blown circuits as the con-formal coating spray was not enough to protect them from the internal condensation when pumping . The pump station now is 4 x PMP500s undervolted < 10v immersed in mineral oil below. 3 @ 12v would have been similar. Immersion makes the pumps quieter , and extends the life substantially of a 12vdc pump i have found. If not immersed the pumps would have to be submersible or be located inside the chillbox to not have the condensation issue. Industrial pumps i tried (tried several) were either too weak, too expensive, too loud or too powerful or had no flow rate control, in the end it was easier to work with 12v pumps.

QQ96rE.jpg

WjIgIc.jpg

Up and running.



One cold night :

negative43c.png



Initial testing in the garage in 2012 .. i wish i had my skyline still
redface.gif




Piping through window :



Exterior Setup :



20120902155901.jpg

20120902155854.jpg

Well if you've made it to here and are still reading then you must be fully interested in doing a chiller build
wink.gif
. I would suggest reading all the channel members threads and the chillbox thread itself as there are different ways of building these chillers and this is only one of them.

2). Working with sub zero liquid and its viscosity which affect flow ability at sub zero temperatures. Pumps, their flow and head pressure ratings, along with operation under freezing conditions.

Rules of entry (... pretty much the thread title)

Rule 1 should be that components eg cpu and gpus are liquid cooled or submerged at least one component. There needs to be liquid somewhere eg Phase only systems although ultra impressive are not applicable.

Rule 2 the liquid must be at subzero temperatures.

Rule 3 system must be able to run 24/7.

Club Members rev 1.0 and threads :

1.) BlackwoodsNC - 24/7 Chiller - Chiller build 2.0
http://www.overclock.net/t/1523560/chiller-build-2-0

2.) ShowMe - 24/7 Chiller - Window AC + Cooler + Chillbox
http://www.overclock.net/t/1517671/window-ac-cooler-chillbox/90#post_23245532

3.) Orthello - 24/7 Chiller - External A/C Chllled Liquid Project
http://forums.extremeoverclocking.com/showthread.php?t=363597&page=23

4.) GIVEitUP - 24/7 Chiller - New chiller build Inside old PC case with water cooled condenser
http://www.overclock.net/t/1540014/new-chiller-build-inside-old-pc-case-with-water-cooled-condenser

Anyone good with logos msg me if so ?

Signature line for the club members - click quote to copy.

post-flame-small.gif
| 24-7 Sub Zero Liquid Chillbox Club |
post-flame-small.gif


Looking forward to the discussion and any input you guys have and ideas :)

Special thanks

Drew (the one and only godfather of extreme cooling) .
TechnoGiant - Thanks for all the ideas mate .
ShowMe - Great input and help with the Thread.
 
See less See more
51
#4 ·
Quote:
Originally Posted by Orthello View Post

Thanks for the heads up Munney , made the changes.
NP man, just wanted to keep ya legal LOL.

Dude, I need some more in-depth pictures and explanation of your box!
 
#5 ·
We so need this. Lets wake OCN up to what a chill box is and how to accomplish awesome performance with a little time and will. In my opinion these are better than phase change units cause you can do the whole system not just one part. Albeit it won't give 100% phase numbers but close to it with out all the insulation work.

Between you @Orthello and @[ShowMe!] I have to step up my game. So with that being said I am gonna move on to my third chiller build MUHAHAHA. Gonna borrow @[ShowMe!] design with the water chilled condenser for space savings and noise reduction. Not that mine won't work, it's just way to loud with that 220v volt fan and the thing is huge. Still gonna make some small changes to mine and I will keep updating the forum on it.

So, @Orthello I needs some details on that chill box. That thing looks sick. What kind of material is that on the outside? Details man details.
:cheers:
 
#6 ·
Quote:
Originally Posted by BackwoodsNC View Post

We so need this. Lets wake OCN up to what a chill box is and how to accomplish awesome performance with a little time and will. In my opinion these are better than phase change units cause you can do the whole system not just one part. Albeit it won't give 100% phase numbers but close to it with out all the insulation work.

Between you @Orthello
and @[ShowMe!]
I have to step up my game. So with that being said I am gonna move on to my third chiller build MUHAHAHA. Gonna borrow @[ShowMe!]
design with the water chilled condenser for space savings and noise reduction. Not that mine won't work, it's just way to loud with that 220v volt fan and the thing is huge. Still gonna make some small changes to mine and I will keep updating the forum on it.

So, @Orthello
I needs some details on that chill box. That thing looks sick. What kind of material is that on the outside? Details man details.
cheers.gif
Dude I love my phase change rig, but the freaking Scythe Kraze that Ales uses is SO OVERPOWERED lolol. I've always wondered if i could mount a CPU block to the evap head and just cool the liquid LOL. I know its stupid, but its a thought since its sitting here!
 
#7 ·
Hey at @Orthello I just skimmed through your build and I will say it is amazing. Maybe you could give everyone on here the tl;dl version. That chillbox is just awesome with that window on it.

I do have a couple of questions though. How are you measuring flow rate? How and where do you have your temp probes to measure water temp in the lines?
 
#8 ·
Hey guys working on an update now to answer all the questions and will put a short version up here of the chillbox build as yeah my thread is kinda in-depth lol :)
 
#9 ·
Sorry Mammoth post guys ...

Hey Blackwood i love the enthusiasm and yeah I agree with you fully , you will not quite compete with a SS phase unit for cooling a single device, but when you can cool your whole motherboard / ram / cpu and gpus below zero that gets really exciting for overall performance. Particularly from a gaming aspect as the GPUs are more the bottleneck than the cpu and they are more easily cooled with liquid rather than phase - especially in SLI or tri SLI etc.

My particular goal was the fastest 24/7 gaming computer I could build , I've just happened to do some benching along the way - WR in 3d mark 2011 (at the time) on page 13 of my thread and still currently 3rd in the Heaven HD 1080p SLI rankings too - once the systems back up i'll try and re-take first. So you can have some benching fun too with chillers.

Also the lack of having to insulate anything once the chillbox is built ... that's a real bonus as anyone whos taken the time to insulate a cpu or gpu against condensation knows its not easy to get it 100% - and may have to you pay the silicon reaper when the insulation fails.

I'll go into more detail re the chillbox , i'll also put pictures in below to break it up a bit.

Further Considerations when building a chillbox are :

1.) What inlets and outlets you will require for any cabling and liquid , because the chillbox needs to be as air tight as possible these inlets and outlets will need to be sealed and air tight once the cabling and piping has been put through and hence adding extra cabling or piping later can be difficult so its best to be really thorough at the start of your build.

2.) Clearances - Do you have enough space in your chillbox for your motherboard and all devices and most importantly a decent sized radiator and fans to act as the air chiller. Allow a bit extra - I nearly had to shave my viewing pane to allow the size of the classified water blocks I run. Remember you are building this for the future devices you may use also.

3.) Do not include the PSUs into the chillbox as they produce waste heat but you would route the cabling through to the chillbox from the PSUs and take it from me - allow as much length in that cabling as feasable. My build worked out very tight with the power supply cabling. My chillbox was an over under system, over was the chillbox and located under were the Piping / PSUs / HDDs etc.

4.) Materials for the chillbox. Possibly the cheapest chillbox is a converted beer chillbox or fishing chillybin etc. These are already insulated so that is a good start. They are reasonably air tight if they have a good gasket near the lid , this can be improved. You can also relatively easily drill through the walls of these chillybins so it makes putting cabling and piping through rather easy.

The negatives to this approach - you can't see anything (the reason you need too comes below in (6)). You can alleviate this by installing an internal camera or modifying the chillybin to have a viewing pane, believe me you need eyes into that chamber . I regularly check my chamber on startup for leaks/problems and i would recomend if you are using a chillybin without a viewing pane that you install enough cameras so you can see all the angles etc. Once its running you will not be able to open it so its better than having blind faith that everything is ok.

A more expensive and time consuming approach is to make a custom acrylic chillbox like i did. This requires insulation to be added to shield the acrylic from the cold , although this is mainly for condensation purposes. I have had my chillbox down to -35c in winter and the acrylic viewing lid which is not insulated did not crack, acrylic seems well suited to the temperatures that a chillbox can obtain.

5.) The radiator acting as the air chiller needs to be the first component in your liquid loop once inside the chamber. This is so the air from the radiator is as near to the liquid temperature as possible. If you have the radiator last you will add extra delta to the air over the incoming liquid temperature so therefore increasing the risk of condensation. It pays to install a humidity sensor in the chillbox, i watch as mine goes from 70%+ on startup to -- (below 23%) after a few minutes. If there is a liquid leak in there the humidity will alter as a bit of the liquid evaporates thats also a good double check for leaks.

6.) Reasons for the viewing pane or cameras where a vane is not possible.

As the 1st post in the thread mentioned the viscosity of what ever liquid you use will increase as the temperature drops as the liquid gets closer to becoming a frozen solid, this will be harder on you pumps, in turn you will need more pumping power to overcome the viscosity of the liquid and therefor you will put more stress on any liquid connections / liquid blocks in your chamber as the PSI restriction increases through the components. Due to this i recomend wormscrew clamps and barbs. Compression fittings for this sort of application i would only recomend for direct connections where the pipe is not bending at all , compression fittings do not have same clamping power IMHO. Orings seem to be fine - i've never had a leak through an Oring , if you want to be sure you can remove the Orings and metal expoxy the threads to the barbs , i have done this one some devices as initially i did not trust the orings under high PSI flow this turned out to be a non issue.

A simple tip wrap all your fluid connections in paper towel, use a non clear antifreeze or some liquid which is coloured and that simple system will allow you to see a slow leak more easily , if you have a slow drip of a leak and your viewing pane gets frosted over within a couple of minutes or your cameras are not up to scratch this will allow you to see a leak very easily.

What if a fan dies and your airflow reduces ? This has happened to me , after all they are in the freezing cold environment. Without proper air flow condensation will build up on your blocks and pipes. Hence another reason to have eyes on the fans or measure the rpms reported via software in windows etc.

My build of the chillbox in pics.


Acrylic chamber which I designed in sketchUp and sent to the local acrylic people and they worked their magic. Note the Gasket lid - this was much heavier than the rest of the chillbox at 20mm thick which worked out well as Its nice and heavy and compresses the gasket prior to screwing in. The rest of the chamber walls were 10mm thick which allows a little flex (not a bad thing due to negative internal air pressure) .The pipe fittings are round incuts , tripple layered for gluing surface. The cable incuts are rectangular. Unless you want to do a lot of soldering then make the incuts into the chillbox frame large enough to take the biggest cable header. You will have to silicon in or use space invading foam or other filler to fill in the gaps between the cables once through. Ideally if it can hold liquid and not drip once you have put cables through you have got it right.

https://imageshack.com/i/g420120622122321j

Added base layer of insulation. The insulation I used was a variety of closed cell foam , but much harder than normal foam , but still able to cut with a craft knife or razor blade. Put side layers on. I had to take particular care to make sure all the levels around the lid were even. Eg the insulation on the inside fitted nicely against the gasket but not so hard that it pushed the gasket out from sitting flat on the rim. The gasket lid has screw holes around the parameter for screwing to the rest of the base. The gasket material is common door jam type stuff ~ 10 mm wide then pushed into the recess of the gasket.

https://imageshack.com/i/jv20120701205255j

Outside insulation added. So the chillbox has internal and external insulation. Overall the walls and bvase are around 80mm thick. Used ados to adhere the insulation to the plastic , this has pulled away in one or two spots internally with the negative air pressure but overall its still fine and doing its job. Note the base insulation and base fibre board have incuts for cabling and piping matching the chamber. Acrylic gravity cradle is glued in place for the radiator.

https://imageshack.com/i/mz20120703225011j

Radiator in place, initially I used a HW labs GTX360 .. this was overkill , in the end I used an XSPC 3 fan radiator that had the lowest restriction of any radiator I could find at the time. I would recommend you use a LOW restriction radiator as its not really going to cool any great wattage if you CPU and GPUs are directly waterblocked. EG the radiator cools the air and maybe the ram and mobo etc ... 200 watts I would think - if that. Really the only reason for the big rad is to get nice even coverage for the air cooling.

https://imageshack.com/i/md20120704201253j

Motherboard ATX frame. I could not find one of these to buy so hacked it out of an old case .. took a while but it was worth it. Note the incuts into the chiller frame for liquid and piping. Ideally I would have added a few inches here and there if I was to design another.

https://imageshack.com/i/mr20120707230118j

Ambient air testing Pic1.
https://imageshack.com/i/f620120708232728j

Ambient air testing Pic 2.
https://imageshack.com/i/cb20120708235655j

Chilled liquid testing (over a year ago .. with older hardware I might add. Two GTX580s @ 1070 mhz. Core I7 930 @ 4.75 ghz - after a bit of trial and error). Note the frosting on top of the viewing pane. It definitely needed a 2nd viewing layer to trap some air and therefore condensate less on the gasket layer.

https://imageshack.com/i/n520120722003113j

The making of the Viewing lid that sits on top of the gasket lid. This basically reduces condensation on top of the gasket lid.

https://imageshack.com/i/7e20120725214227j

Finished the viewing lid and testing. In this pic I am using the original piping config which had a valve system for allowing prechill of the air by running liquid through the radiator only prior to running liquid through the blocks. In the end I got rid of this as I found I did not need to prechill the air prior to running liquid through the system. This improved my flow rates - running this method however its critical to make the radiator first in the cooling loop.

https://imageshack.com/i/mk20120725224609j

Pic of the sensor inside the chillbox , note the air temperature is -17.6 with -- humidity (below 22%). This was with really bad liquid flow too , now days the air temp is typically in the low -20s. Pumping issues plagued me for nearly 6 months .

https://imageshack.com/i/g020120726220507j

The above way is only one way to build a chillbox , there will be many . I hope with this thread we will be able to experience a lot of different builds to come and share the ideas.
 
#10 ·
Quote:
Originally Posted by BackwoodsNC View Post

Hey at @Orthello
I just skimmed through your build and I will say it is amazing. Maybe you could give everyone on here the tl;dl version. That chillbox is just awesome with that window on it.

I do have a couple of questions though. How are you measuring flow rate? How and where do you have your temp probes to measure water temp in the lines?
Sorry missed the questions , the flow rate is measured using an aquacomputer flow rate sensor - the mechanical one. It has some rather awesome software too so that's one of the main reasons I use it. I use a Phobya inline water sensor to measure the liquid temperature. Normally after radiator as that way you get the temperature just prior to components.

Here's the link to the flow sensor i use.
http://www.frozencpu.com/products/16789/bus-316/Aquacomputer_G14_Flow_Meter_Sensor_Block_USB_Version-Aquaero_SeriesPoweradjustFan-O-Matic_ProAlphacool_Heatmaster53129.html

Hers the link to the liquid temp sensor i use.
http://www.frozencpu.com/products/21219/ex-tub-1991/Phobya_G14_Inline_Temp_Sensor_Coupler_-_2-Pin_-_Black_Nickel_71226.html

Bring on chiller 3.0 man !!
 
#11 ·
Quote:
Originally Posted by Orthello View Post

Sorry Mammoth post guys ...

Hey Blackwood i love the enthusiasm and yeah I agree with you fully , you will not quite compete with a SS phase unit for cooling a single device, but when you can cool your whole motherboard / ram / cpu and gpus below zero that gets really exciting for overall performance. Particularly from a gaming aspect as the GPUs are more the bottleneck than the cpu and they are more easily cooled with liquid rather than phase - especially in SLI or tri SLI etc.

My particular goal was the fastest 24/7 gaming computer I could build , I've just happened to do some benching along the way - WR in 3d mark 2011 (at the time) on page 13 of my thread and still currently 3rd in the Heaven HD 1080p SLI rankings too - once the systems back up i'll try and re-take first. So you can have some benching fun too with chillers.

Also the lack of having to insulate anything once the chillbox is built ... that's a real bonus as anyone whos taken the time to insulate a cpu or gpu against condensation knows its not easy to get it 100% - and may have to you pay the silicon reaper when the insulation fails.

I'll go into more detail re the chillbox , i'll also put pictures in below to break it up a bit.

Further Considerations when building a chillbox are :

1.) What inlets and outlets you will require for any cabling and liquid , because the chillbox needs to be as air tight as possible these inlets and outlets will need to be sealed and air tight once the cabling and piping has been put through and hence adding extra cabling or piping later can be difficult so its best to be really thorough at the start of your build.

2.) Clearances - Do you have enough space in your chillbox for your motherboard and all devices and most importantly a decent sized radiator and fans to act as the air chiller. Allow a bit extra - I nearly had to shave my viewing pane to allow the size of the classified water blocks I run. Remember you are building this for the future devices you may use also.

3.) Do not include the PSUs into the chillbox as they produce waste heat but you would route the cabling through to the chillbox from the PSUs and take it from me - allow as much length in that cabling as feasable. My build worked out very tight with the power supply cabling. My chillbox was an over under system, over was the chillbox and located under were the Piping / PSUs / HDDs etc.

4.) Materials for the chillbox. Possibly the cheapest chillbox is a converted beer chillbox or fishing chillybin etc. These are already insulated so that is a good start. They are reasonably air tight if they have a good gasket near the lid , this can be improved. You can also relatively easily drill through the walls of these chillybins so it makes putting cabling and piping through rather easy.

The negatives to this approach - you can't see anything (the reason you need too comes below in (6)). You can alleviate this by installing an internal camera or modifying the chillybin to have a viewing pane, believe me you need eyes into that chamber . I regularly check my chamber on startup for leaks/problems and i would recomend if you are using a chillybin without a viewing pane that you install enough cameras so you can see all the angles etc. Once its running you will not be able to open it so its better than having blind faith that everything is ok.

A more expensive and time consuming approach is to make a custom acrylic chillbox like i did. This requires insulation to be added to shield the acrylic from the cold , although this is mainly for condensation purposes. I have had my chillbox down to -35c in winter and the acrylic viewing lid which is not insulated did not crack, acrylic seems well suited to the temperatures that a chillbox can obtain.

5.) The radiator acting as the air chiller needs to be the first component in your liquid loop once inside the chamber. This is so the air from the radiator is as near to the liquid temperature as possible. If you have the radiator last you will add extra delta to the air over the incoming liquid temperature so therefore increasing the risk of condensation. It pays to install a humidity sensor in the chillbox, i watch as mine goes from 70%+ on startup to -- (below 23%) after a few minutes. If there is a liquid leak in there the humidity will alter as a bit of the liquid evaporates thats also a good double check for leaks.

6.) Reasons for the viewing pane or cameras where a vane is not possible.

As the 1st post in the thread mentioned the viscosity of what ever liquid you use will increase as the temperature drops as the liquid gets closer to becoming a frozen solid, this will be harder on you pumps, in turn you will need more pumping power to overcome the viscosity of the liquid and therefor you will put more stress on any liquid connections / liquid blocks in your chamber as the PSI restriction increases through the components. Due to this i recomend wormscrew clamps and barbs. Compression fittings for this sort of application i would only recomend for direct connections where the pipe is not bending at all , compression fittings do not have same clamping power IMHO. Orings seem to be fine - i've never had a leak through an Oring , if you want to be sure you can remove the Orings and metal expoxy the threads to the barbs , i have done this one some devices as initially i did not trust the orings under high PSI flow this turned out to be a non issue.

A simple tip wrap all your fluid connections in paper towel, use a non clear antifreeze or some liquid which is coloured and that simple system will allow you to see a slow leak more easily , if you have a slow drip of a leak and your viewing pane gets frosted over within a couple of minutes or your cameras are not up to scratch this will allow you to see a leak very easily.

What if a fan dies and your airflow reduces ? This has happened to me , after all they are in the freezing cold environment. Without proper air flow condensation will build up on your blocks and pipes. Hence another reason to have eyes on the fans or measure the rpms reported via software in windows etc.

My build of the chillbox in pics.


Acrylic chamber which I designed in sketchUp and sent to the local acrylic people and they worked their magic. Note the Gasket lid - this was much heavier than the rest of the chillbox at 20mm thick which worked out well as Its nice and heavy and compresses the gasket prior to screwing in. The rest of the chamber walls were 10mm thick which allows a little flex (not a bad thing due to negative internal air pressure) .The pipe fittings are round incuts , tripple layered for gluing surface. The cable incuts are rectangular. Unless you want to do a lot of soldering then make the incuts into the chillbox frame large enough to take the biggest cable header. You will have to silicon in or use space invading foam or other filler to fill in the gaps between the cables once through. Ideally if it can hold liquid and not drip once you have put cables through you have got it right.

https://imageshack.com/i/g420120622122321j

Added base layer of insulation. The insulation I used was a variety of closed cell foam , but much harder than normal foam , but still able to cut with a craft knife or razor blade. Put side layers on. I had to take particular care to make sure all the levels around the lid were even. Eg the insulation on the inside fitted nicely against the gasket but not so hard that it pushed the gasket out from sitting flat on the rim. The gasket lid has screw holes around the parameter for screwing to the rest of the base. The gasket material is common door jam type stuff ~ 10 mm wide then pushed into the recess of the gasket.

https://imageshack.com/i/jv20120701205255j

Outside insulation added. So the chillbox has internal and external insulation. Overall the walls and bvase are around 80mm thick. Used ados to adhere the insulation to the plastic , this has pulled away in one or two spots internally with the negative air pressure but overall its still fine and doing its job. Note the base insulation and base fibre board have incuts for cabling and piping matching the chamber. Acrylic gravity cradle is glued in place for the radiator.

https://imageshack.com/i/mz20120703225011j

Radiator in place, initially I used a HW labs GTX360 .. this was overkill , in the end I used an XSPC 3 fan radiator that had the lowest restriction of any radiator I could find at the time. I would recommend you use a LOW restriction radiator as its not really going to cool any great wattage if you CPU and GPUs are directly waterblocked. EG the radiator cools the air and maybe the ram and mobo etc ... 200 watts I would think - if that. Really the only reason for the big rad is to get nice even coverage for the air cooling.

https://imageshack.com/i/md20120704201253j

Motherboard ATX frame. I could not find one of these to buy so hacked it out of an old case .. took a while but it was worth it. Note the incuts into the chiller frame for liquid and piping. Ideally I would have added a few inches here and there if I was to design another.

https://imageshack.com/i/mr20120707230118j

Ambient air testing Pic1.
https://imageshack.com/i/f620120708232728j

Ambient air testing Pic 2.
https://imageshack.com/i/cb20120708235655j

Chilled liquid testing (over a year ago .. with older hardware I might add. Two GTX580s @ 1070 mhz. Core I7 930 @ 4.75 ghz - after a bit of trial and error). Note the frosting on top of the viewing pane. It definitely needed a 2nd viewing layer to trap some air and therefore condensate less on the gasket layer.

https://imageshack.com/i/n520120722003113j

The making of the Viewing lid that sits on top of the gasket lid. This basically reduces condensation on top of the gasket lid.

https://imageshack.com/i/7e20120725214227j

Finished the viewing lid and testing. In this pic I am using the original piping config which had a valve system for allowing prechill of the air by running liquid through the radiator only prior to running liquid through the blocks. In the end I got rid of this as I found I did not need to prechill the air prior to running liquid through the system. This improved my flow rates - running this method however its critical to make the radiator first in the cooling loop.

https://imageshack.com/i/mk20120725224609j

Pic of the sensor inside the chillbox , note the air temperature is -17.6 with -- humidity (below 22%). This was with really bad liquid flow too , now days the air temp is typically in the low -20s. Pumping issues plagued me for nearly 6 months .

https://imageshack.com/i/g020120726220507j

The above way is only one way to build a chillbox , there will be many . I hope with this thread we will be able to experience a lot of different builds to come and share the ideas.
Wow man... Thats VERY impressive! VERY VERY IMIPRESSIVE.

If you don't mind me asking you... what'd the box set you back? You ever though about making these FOR people?
 
#14 ·
Looking for the sketchup file now of the chillbox, happy to share it if I can locate it lol .. was nearly 2 years ago I think now.

It wasn't cheap to build here in NZ as it was a one off custom job. I'll have a hunt for the invoice .. From memory it was between 350-400 nzd inc GST. I claimed tax on it through my company so it was a bit cheaper than that in the end . Its possible with a larger run the price would come well down etc.

A decent chillybin is near that hear for comparison - for the top end ones with 40mm walls, although I had to add insulation to the acrylic chillbox.

For comparison my 2nd hand A/C unit cost 100 nzd ! Still I've never regretted it as it saved me a lot of time in the building of the chillbox and I got the exact dimensions I wanted internally.

The chillbox has seen three sets of hardware since then so its paid itself off . 2012 had the i7 930 @ 4.750 with GTX580 SLI @ 1070mhz. 2013 had i7 2700k @ 5.4 ghz and 7970 lightning CFX @ 1400mhz , later swapped the cpu and mobo for a 3770k that I benched at 5.1 ghz.. 2014 in the signature.

So yeah not cheap but then again the results were great.
 
#15 ·
I am gonna build a cheap chillbox tomorrow. I stopped by home depot and picked everything up for about 60 bucks(to lazy to look at the receipt). Specs are gonna be W x D x H so 24"X24"X12" nice and square. I needed a bigger cooler to set everything in there flat so i could see everything. So I decided to just build one out of wood.
 
#16 ·
i was thinking about using wood to make one too. wood is a decent insulator. maybe seal the wood with something and use foam to increase the temp. plexi or glass top....
biggrin.gif


cant wait too see how it turns out.

what size rad are you using? i have an old 280mm i thought would do the trick.
 
#17 ·
Quote:
Originally Posted by 97discosd View Post

i was thinking about using wood to make one too. wood is a decent insulator. maybe seal the wood with something and use foam to increase the temp. plexi or glass top....
biggrin.gif


cant wait too see how it turns out.

what size rad are you using? i have an old 280mm i thought would do the trick.
Gonna use FOAMULAR 1/2 in. x 4 ft. x 8 ft. R-3 pink insulation. Gonna double it up and glue it to the inside of the board. Going to use silicone to seal the joints on the inside. Hopefully it turns out nice. The outside walls are going to be 1''x6''x24'' wood planks.

I have been using a XSPC RX240 and it cools it down nicely. That 280 should do the trick no problem.
 
#18 ·
Nice one Blackswood make sure to get some pics of the chillbox build - Sounds like its going to make a cheap but very effective chill box.

I've also just linked to frozencpu that flowmeter and the inline liquid sensor back in the orginal post Blackswood.

Just powered up the rig after 4 days down for maintainence . In those 4 days i've Inner walled the chiller with Aluminium , redid the suction inlet / outlet. Changed the filter. Rerouted the pumps. Changed the CPU block to last and removed impingement plate.

Just on the drop down now .. been about 30 mins and liquid is at -15c from 26c. About 40 litres in the chiller at present. Once she hits below -20c i'll start a prime test and see what removing the impingement plate has done to my performance on the CPU. GPUs should be 2-3c better off at least as they are first in line after radiator now .

Getting 240 l/hr at -16.5c liquid with 8v pumps .. 10v pumps 302 l/hr ... 12v pumps 367 l/hr.

Getting 165 l/hr at -20.6 liquid with 8v pumps .. 10v pumps 224 l/hr ... 12v pumps 308 l/hr

That list of mods worked out rather well .. getting more flow now at 10v pumps than prior with 12v. Seems to have made very little difference to cpu temps also .. also changed tim to gelid xtreme over liquid pro as my cpu is starting to be a real mess with liquid pro - almost needed sand paper to clean it, it seems to dry out also in the chillbox. Dropped 5c + of GPU temps however .. just played one hour of SOM and maxed and 2c on gpu 1 and 4c on gpu2 with -20.5c liquid at the end of the session. The settings i'm using in SOM now are nearly maxing the gpus so usage is very high. GPU1 and 2 were largely subzero the whole time. Should be able to drop voltage now on the GPUs or increase overclock. More testing to come :)

Updated my thread with some pics of the above too.
 
#19 ·
Aight folks here it is.

The one tool that I would suggest you own or get to build this is a miter saw. With a circular saw too many of my cuts where off are crooked. After about a 1/4 of my cuts I made a trip my local Home Depot to pick up a Miter saw. So, with that being said some of my beginning cuts or off a little bit and the rest where square.
























Ran out of time before the football game to finish the top. Might finish it tomorrow or next weekend. Besides the top just got to drill the holes in the bottom and silicone the inside to make it air tight.
 
#20 ·
Great work Blackswood and great pics, i'm sure its going to seal real nice with a bit of silicone.

Agree with you on the mitre saw , i don't have one but need to get one. Home made is never 100% but just has to be good enough. Been a bit of a perfectionist myself sometimes i just have to tell myself that's as good as its going to get with my current tools lol.
 
#24 ·
Hey 66racer , that's the whole point - 24/7, turn it on and game lol.

Hitting some hot days here in NZ lately (hot for us) . Would have been around 32c yesterday outside and liquid hit a sweltering -12c under full load at midday. I normally game at nights and its atleast 7-8c colder then. Compressor was running about 56c (as hot as i've seen it run) too.

Lucky GPUs were still holding at 3 and 5c max (flow speeds up quite a bit vs -20c so the upside is nearly identical performance at the GPUs). Had to increase vcore on the GPUs too as when the temp rises slight vcore adjustments are needed. So in a sense the heatload problem becomes even worse.

Got a third 780 TI coming so i'm either going to have to do some real low voltage optimization per card or change the season here - bring on winter lol ...
 
#26 ·
how easy is it to get access to the inside of these when its all set up? i might set up one of these for benching, but i switch my gpu's out a lot.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top