Helium results!!! Window air con chill box build. - Page 2

I have thought of that, and I know others have also done this.

But one of my main reasons for doing it this way was that once the chill box is completed then it will be easy to upgrade my components or to do a complete pc rebuild just using air cooling components... no need for the complexity of water cooling, or the purchase of expensive cpu and gpu water blocks for each and every upgrade but still having the advantages of sub ambient cooling.

Also the problem with "directly" cooling the component to sub ambient temps while the surrounding air is at normal temp is of course condensation.

By using a sealed chill box and cooling the air within it rather than directly cooling the component I hope to avoid this situation as the components will always be warmer than the chill box air (although obviously colder than the external ambient air). I really don't want to go down the route of having to seal my mobo and components against condensation and have it die a few months later because it wasn't done quite right. I'm hoping that the only place condensation will form will be on the evap rad.......but this is a matter of experiment atm.

The other reason I don't want to use it to chill water is that the low temps achieved, although great, require the use of strong antifreeze alcohol solutions and you need to choose the components very carefully to ensure they aren't attacked by the combination of extreme low temps and chemicals.
This may well be the best method for extreme overclocking but I'm not sure I want all those complications.

In any event I'm still considering directly linking the component coolers to the ac evap rad using a mesh of connecting wires, this "may" provide equally low temps without the aforementioned complications. This "may" cause condensation as the components are now being directly cooled...but likewise the air temp will also be similar as that is also being cooled....so again condensation "may" not occur......thats a lot of "mays" obviously this is quite experimental and something I'm going to be looking at in due course.
Edited by technogiant - 6/3/12 at 6:32pm

Thanks Drew.......you've highlighted some pretty important stuff there that has been taxing my thoughts.

I made the chill box as a prototype really more for proof of concept, if I can iron out the bugs I'll make a more substantial one. when I'm ready to do longer test runs I'll simply be duct taping all the seams, that together with the simple idea I had for an expansion chamber should prevent it sucking in atmospheric air because of pressure changes even if some of the seals aren't 100%. I had the lid slightly a jar when I turned it on the first time as I had cables running under it so obviously not air tight.

I've also been thinking that if I start the air con unit a short while before the fans the evap will frost up and draw all the moisture out of the enclosure so it will be dry when I start the fans.
I think starting them both at the same time is just throwing the air around as the moisture is condensing out and that's why the fans frosted....some testing to be done there.

That's an excellent idea about putting a viewing panel in the lid.....I really do need a means to monitor whats going on.

The problem about the aircon unit cycling on and off is a little more vexing. It may have been better if I'd bought an "inverter" one as they moderate the compressor power rather than just on/off modes.
Of course you would not get this problem running a water filled reservoir as you do as you have quite a large heat capacity.

This is a more difficult one. ...the ideal solution would be being able to run the compressor continually to get the lowest possible temps without getting frost and having no need to cycle on off. You never know I may luck out there when I'm load testing, I may find it will run continuously with out frosting up, could even put a heat source in the box that kicks in below a certain temp to prevent it dropping too low and frosting everything when pc power use is low rather than have the compressor cycle on/off.

Or even better, I could sit the evap in a small reservoir of water that would come only part way up the evap rad, it would add heat capacity to the system and chill the evap during the compressor down times.

The problem behind this is frosting, if I can solve that then the compressor can run all it likes without need to cycle.

As regards hdd and ssd's your right , I was looking up mfr's spec last night, they don't like low temps.
Edited by technogiant - 6/4/12 at 8:41am

Well I've done some more testing today and its good and bad news.

The good news is that the chill box was holding a 2kw heat load at +3 deg C and a 150watt load at off scale below -20 deg C.

The bad news is condensation.

I have to admit my original concept is flawed. I thought that condensation would only form on actively cooled surfaces with a large temp delta between it and the surrounding air.

Sadly that's not the case. It's nothing to do with moisture ingress into the chill box, its just the plain fact that as air cools it can hold less moisture and will dump it on any surface regardless of whether that surface is chilled or of its temp delta with the air.

I tried turning on the air con unit before the fans in the hope it would suck the moisture out.

This does help and results in less frost depositing but the chill box air temp doesn't actually drop much until you turn on the fans, then the temp drops like a stone, literally from +20 deg C to minus figures in under 30 secs, and I guess the air just dumps more moisture.

One idea may be to reduce the temp more slowly giving the moisture more of a chance to condense on the cold evap surface, I haven't got any control on those fans yet, or for that matter on the air con unit, perhaps if I left the evap run a little longer without the fans and then started them slowly, together with a gradual decrease in evap temp?

Just wondering how long it is safe to run the evap without load, don't want to damage my compressor?
Edited by technogiant - 6/5/12 at 12:39am

Drew, I think I've pretty much sealed it all up, okay it's looking very ghetto now covered in duct tape so no pictures.

Also I used my expansion chamber idea. (remember plastic bag on breather pipe...lol)...so I'm equalizing the inside and outside pressure to further reduce leakage.

Obviously you can never completely seal but I think it's good enough to test this concept for now, if it works I will spend out more on a robust build.

I think we have come to the solution simultaneously here, it's the fans.

When I ran the evap initially for a period before starting the fans, giving the evap chance to dry the air, that did reduce the frost/condensation.

But I also noticed (using your viewing panel idea) that while doing this the temp in the chill box didn't alter much till I turned the fans on, then it plummets from +20 to minus figures in a very short time.

I think this rapid drop is causing the moisture to dump anywhere.

I need to give it time to condense on the evap. I have no control on those fans yet. Also getting some control on the evap temp may also help to give a slower start.

I'll also reverse the flow on my fans atm they are pushing through the evap,
could even pack some desiccant in the chamber behind the evap like mott555 suggested.

I'm a little concerned about running the evap on its own first without the fans or any load, I don't want to damage the compressor, any ideas how long I could safely do this, it appears to be a rotary compressor with built in accumulator?

Just done some more testing, this time I again let the ac unit run for a bit on its own, then started pulsing the fans on/off just for brief periods so the temp gradually dropped.

Got down to -20 with a 150 watt heat load with no signs of condensation.

One thing I have noticed that seems to be giving false indications of condensation.

Before I open the box after a run I've been leaving the heat source on to speed up the return of the chill box to room temp.

Whats happening is that there are "dead spots" which don't heat up as fast and so remain colder than the chill box air, these dead spots then form condensation.

I think its probably wise to let the system temp return to normal passively , perhaps even leaving the fans run to even out the internal temps.

So its time to put some control on those fans. Don't want to spend out much on this sort of thing (already spend £70 on a rad I don't now need)....any suggestions.

Also going to rebuild the box...I'm happy this is going to work now.

I want to give it a hard shell and obviously an insulating layer, not really sure of the best materials to use, I guess things like MDF would be porous and not suitable?

I'm thinking perhaps a plastic box for the interior air tight layer surrounded by insulation and then a hard shell, perhaps MDF would then suffice for the outer layer? Also how would plastic stand up to sub zero temps?

Just stumbled across another possible problem.. my intention was to put my psu in the chill box, but looking at the mfr spec operating temp range is only down to 0 deg C .....storage temp down -20 deg C.

If I have to run the all the cables through the box wall that's going to be a problem to seal, even if you adequately seal on the outside of the cable vapor will track between the cables bundled together.
Edited by technogiant - 6/5/12 at 4:08am