For the sultans of Celsius...a semi-noob and his first build

[thenrz]

Finally building my chiller....after 4 years. I have had a nice shiny new embraco compressor (honestly can't remember if it's 1/3HP or 1/2HP...been that long), a cube condenser, all the other stuff including vacuum pump, hoses, manifold, tools etc etc ready to go. Also have a still sealed tank of r507.

Here's where my questions start...within the text, will summarize at the end...

I want to build an 'on demand' water chiller. I will build a tubular heat exchanger. I was considering going for a plate, but I want the hands on value at the moment and will go plate later if its really that much better. I will be building a circuit here at work with a barometer, thermometer etc to find dew points and RH...all that. I want the system to be active at keeping things at ambient+1, but have the power to go subzero if needed. I understand these systems would be charged differently (...right?) but I haven't even gotten there yet. My load is my 24/7 system that I still haven't upgraded in a while. E8600 x48 2x 4850. I have been out of the enthusiast game for a while as my life changed there for a while

Ok, questions:

1. What's a good material to use for the outside tube in my hx? I was thinking about getting some wide diameter copper form mcmaster as we get stuff delivered here all the time from them, but then again, I find no reason to spend the premium on copper as it seems it won't do any good. I want the option to get my coolant down to -30C to -40C on just my CPU (seems like a stretch, but I'm patient and have some money) so I'd want tubing that's good down to -50C as I don't trust specs like that anymore.

2. Can I keep enough coolant in the lines and hx to do what I want? I'd like to avoid using a big reservoir as I want my temps to be on demand and completely controllable from my controller board I'm making. Will I need some kind of small reservoir or will be it (hopefully) unnecessary?

3. I've done tons of reading into the finer point of charging and all that, but I haven't seen a lot of more precise info on charging chillers, a bit on here, but maybe I've missed what I'm looking for...should have searched How will it be tuned differently?

4. I'm going to use either alcohol or alcohol+antifreeze. What do you recommend. I get that more water = more capacity for heat, so in my active system, more water would be better. But at say, -30C, is it ok to use a 60/40 glycol/water mixture? Wikipedia tells me I'd be ok, but would be be getting too viscous? I have a nice Iwaki pump, but I traded my other one away for this one and my buddy hasn't ever sent me the new impellar housing...which is obviously needed. Don't remember the specs...will fill in later.

5. Good, cheap waterblock recommendations? CPU+NB first as this is what I have the most experience with. I'm SO out of the loop, all pun intended, with this stuff. I still have 1A-cooling stuff from like 8 years ago. After stage 1, I'll go ahead and get the video cards going...with maybe another chiller


Thanks for reading!!!

[n00b 0f l337]

1. What's a good material to use for the outside tube in my hx? I was thinking about getting some wide diameter copper form mcmaster as we get stuff delivered here all the time from them, but then again, I find no reason to spend the premium on copper as it seems it won't do any good. I want the option to get my coolant down to -30C to -40C on just my CPU (seems like a stretch, but I'm patient and have some money) so I'd want tubing that's good down to -50C as I don't trust specs like that anymore.

Sadly, your first answer. Copper. Besides that you will need to test materials, I doubt nylon or telfon tubing or armaflex will last very long. Neoprene tubing possibly depending on your coolant.

2. Can I keep enough coolant in the lines and hx to do what I want? I'd like to avoid using a big reservoir as I want my temps to be on demand and completely controllable from my controller board I'm making. Will I need some kind of small reservoir or will be it (hopefully) unnecessary?

You would use a plate heat exchanger for this function and have no reservoir, just a filling point. Yes it is possible, but it is a complex system, and will most likely be taxing to build for a first system.

3. I've done tons of reading into the finer point of charging and all that, but I haven't seen a lot of more precise info on charging chillers, a bit on here, but maybe I've missed what I'm looking for...should have searched How will it be tuned differently?

A TXV won't be happy with your conditions. Nor will a CPEV (but a not horrid option, just not able to be linked electrically).
You will want either an electronic expansion valve, or to create a hot gas bypass system to perform a warming component.
Charging will be like most systems, charge till load holding, adjust captube according to superheat.

4. I'm going to use either alcohol or alcohol+antifreeze. What do you recommend. I get that more water = more capacity for heat, so in my active system, more water would be better. But at say, -30C, is it ok to use a 60/40 glycol/water mixture? Wikipedia tells me I'd be ok, but would be be getting too viscous? I have a nice Iwaki pump, but I traded my other one away for this one and my buddy hasn't ever sent me the new impellar housing...which is obviously needed. Don't remember the specs...will fill in later.

"Capacity" for heat won't be a big issue and probably not something you could really calculate anyway. The chiller will hold or it won't.
I would suggest simply using enough glycol to not worry about freezing (which will be alot), and not worry about eating your tubing or gumming up blocks.

5. Good, cheap waterblock recommendations? CPU+NB first as this is what I have the most experience with. I'm SO out of the loop, all pun intended, with this stuff. I still have 1A-cooling stuff from like 8 years ago. After stage 1, I'll go ahead and get the video cards going...with maybe another chiller

As high flow as possible with this kind of system. If you can find some old Swiftech 6002's, you'd be rocking.

[Holst]

For the HX the outer tubing wont have any impact on cooling performance, so use whatever you want.
Remember the you need high flowrate through the heat exchanger or it will ice up, the Iwaki will help here but keeping restriction low is important.

Im not sure what tubing to use for -50. Standard tubing would probably work, although it will go very hard. You can probably buy some -50 rated tube but it might be hard to bend and expensive.

If your controller is clever then you wont need a reservoir. The only advantage of a reservoir will be that you have more thermal mass so your compressor on/off periods will be longer.
Some sort of reservoir will help with bleeding air out of the system, the more compact you can make things the better as you will find it easier to insulate should you want to go very cold

For very low temps then water+glycol may still freeze, however I have never felt very comfortable with people using alcohol in there systems, especially if they are 24/7 systems not just benchers.
Down to -30 I would go with the water+glycol providing your flowrate is high enough through the HX.

I dont think there is much point in buying a cheap waterblock. Either reuse what you have or get something decent! I have a EK supreme HF which would work fine, but I would read skinnee's reviews and pick one with low restriction and with an easy to insulate design. As block design is so good these days I think a block bought now will last another 8 years and still be competitive. As with the tubing it may also be worth considering if the block will be ok with alcohol at -50, which means no perspex/plexi tops (alcohol will damage them) I would probably go with an all copper block.

I cant help on charging the system, ask in the phase change section and they will explain.

[thenrz]

Hey, awesome guys! Thanks for the responses. I remember both of you from back in the day, nice to see some familiar faces.

A TXV won't be happy with your conditions. Nor will a CPEV (but a not horrid option, just not able to be linked electrically).
You will want either an electronic expansion valve, or to create a hot gas bypass system to perform a warming component.
Charging will be like most systems, charge till load holding, adjust captube according to superheat.

Gotcha. Complicated To try and parse some of your text here, hot gas bypass system? Am I overthinking it?

Sadly, your first answer. Copper. Besides that you will need to test materials, I doubt nylon or telfon tubing or armaflex will last very long. Neoprene tubing possibly depending on your coolant.

Seems like it keeps coming back to that. My boss here is really pushing me away from it due to cost, but again, I've waiting long enough on this project I have a little saved up for fun times.

As high flow as possible with this kind of system. If you can find some old Swiftech 6002's, you'd be rocking.

Ahhh. That would be rockin. I think there' an entire part of my brain dedicated to all the stuff I remember about watercooling and components...that lies dormant...

You would use a plate heat exchanger for this function and have no reservoir, just a filling point. Yes it is possible, but it is a complex system, and will most likely be taxing to build for a first system.

Would a plate hx be the only option? I'm in a fabricating mood, so if it can be done with a tube in tube hx, I'd like to do that. If not, I can buy one if that's what works and change my plan. I was unsure if I could have a system designed both for on demand and as cold as possible.

Im not sure what tubing to use for -50. Standard tubing would probably work, although it will go very hard. You can probably buy some -50 rated tube but it might be hard to bend and expensive.

I'm at a loss here. I would really hate to have to run rigid tubing all over the place. I'm wondering if it wouldn't be better to focus more on the reactive properties of my 'dream' system.

I dont think there is much point in buying a cheap waterblock. Either reuse what you have or get something decent! I have a EK supreme HF which would work fine, but I would read skinnee's reviews and pick one with low restriction and with an easy to insulate design. As block design is so good these days I think a block bought now will last another 8 years and still be competitive. As with the tubing it may also be worth considering if the block will be ok with alcohol at -50, which means no perspex/plexi tops (alcohol will damage them) I would probably go with an all copper block.

Machining my own is an option. If my old 1a-cooling blocks weren't so damn resistive, I'd just use them for now. I guess with an Iwaki at these temps, getting every percentage of flow rate seems moot.



Thanks A TON for your input guys. Will keep the questions coming as this thing gets built!

[n00b 0f l337]

Plate HX will probably get you the most transfer surface area, AND solve your HX issues. Restrictions not bad either.
You can get one for $45 shipped on ebay easily if you are okay with a 10 plate. I would suggest a 20 or 30 depending on your load.

Also as a tip: As much as we like being called sultans (the hats are just plain cool), it makes it hard to open a thread to check if you don't know what it's going to say at all

If you really want to fabricate, build a nice box for tight insulation packing around the plate HX and start on a good reservoir/fillport.

[thenrz]

Plate HX will probably get you the most transfer surface area, AND solve your HX issues. Restrictions not bad either.
You can get one for $45 shipped on ebay easily if you are okay with a 10 plate. I would suggest a 20 or 30 depending on your load.

Right on. I'm planning on disassembling a couple old AC units I've been hoarding. I'll get my fabbing need out on another system. I just don't like single stage direct die systems. I've had SO much trouble putting those together. I have a Chilly 1 (i think!? not called that? was that his LN2 pot?) evap with a flex line and all that and when I was doing all this years ago, I just hated the mounts and all the stuff associated with just getting the dang thing to mount unicorn+rainbow style. Also, I like the idea of a liquid acting as a intermediary between the refrigerant and components. I'm not too concerned about every last degree C, more about a cooler system that can run 24/7 on a mild OC + my desire to build things like this.


As for the sultans of Celsuis...I was overcome with my love for word play... No doubt there should be more info I have to get at a mod to change a thread title don't I...

[n00b 0f l337]

When in the subforum, double click next to the link to this thread, should let you change it.

Right on. I'm planning on disassembling a couple old AC units I've been hoarding. I'll get my fabbing need out on another system. I just don't like single stage direct die systems. I've had SO much trouble putting those together. I have a Chilly 1 (i think!? not called that? was that his LN2 pot?) evap with a flex line and all that and when I was doing all this years ago, I just hated the mounts and all the stuff associated with just getting the dang thing to mount unicorn+rainbow style. Also, I like the idea of a liquid acting as a intermediary between the refrigerant and components. I'm not too concerned about every last degree C, more about a cooler system that can run 24/7 on a mild OC + my desire to build things like this.

I agree, it can be a hassle. But you may find insulating every little bit of anything is going to be a bigger hassle

[thenrz]

When in the subforum, double click next to the link to this thread, should let you change it.

Tried it, allows me to edit for .25 seconds then links me to this page.

I agree, it can be a hassle. But you may find insulating every little bit of anything is going to be a bigger hassle

That's about 20% of my reason for wanting to go ambient. I love the refrigeration stuff, but the nightmare of having to maintain it so closely is a problem with my schedule.

You've been a wealth of knowledge.. what can you tell me about refrigeration systems and keeping them at ambient? I imagine that turning the compressor on and off all the time is bad, so obviously will be avoided. There will have to be an algorithm to determine when the compressor is really needed and when it is not. Is it accepted to throttle the fans on the condenser or will that mess with the tune? In that case would it be better to have a larger reservoir and more coolant? i attached a picture with, the way I see it, my two options. I'm sure it's somewhere in between, smaller res etc. The reservoir options would also need two pumps. I actually see a cool idea here. The res could be recycling with just 'out of the hx' coolant and two valves could switch and run through a normal radiator? Would the radiator, at that point, become a heat dump into the system? Sorry for the babble.

The attached photo is a lovely mspaint masterpiece. Have all the fun graphics stuff at work...not on the good ol laptop.

[n00b 0f l337]

If you want ambient, go r134a, plate HX and reservoir, and use a thermostat to cycle the compressor with at least a 5-10 minute delay on the electrical side. Should be easy.

[thenrz]

I know it's only a 10 plate, but will this be a good one to give a go? Here - eBay

I figure go less expensive on my first build and supplement later if I want more plates. I remember these being more expensive! Woohoo


Is there a reservoir size sweet-spot here? Seems like there would be, although insulating it will be the most important. I will probably build two systems concurrently, the other will be a subzero unit built from window AC parts (have three at my disposal now ). From what I've taken from your guys' posts, the systems will function differently enough to warrant keeping them separate. I think this will be good for me, though, as building two different units should give me a good amount of experience in a shorter time.

Would it be better to use older parts to build the ambient chiller and the new parts to build the subzero unit? Seems logical, but I think the window AC units are rotary type and are probably a bit more powerful...

If you want ambient, go r134a, plate HX and reservoir, and use a thermostat to cycle the compressor with at least a 5-10 minute delay on the electrical side. Should be easy.

Under heavy load, would it be healthier to slow down the condenser fans instead of turning off the compressor? Remember, as I stated before I will be building a little controller that I can work on as my projects get more complex. It will have probably 10 thermocouple inputs just to have enough for future workings. Also, I figure it's not a bad thing to have the system monitored in all spots if its an option, as I'd like it to update via FTP and some windows software what the status is... the reason for my paranoia? I built a TEC system years ago for honors physics in high school. I had it running 24/7. Left one evening and forgot to flip all three switches off. The TEC stayed on. The pump was hardwired into that circuit to help a situation such as this. There was cooling, but no running fans on the radiator. The reservoir was a custom build that sat on top of my clear acrylic case. The rad was also acrylic. Apparently I used too little silicone and the coolant got so hot it started softening up the silicone. Eventually all the water ran out and the TEC went into thermal runaway I assume. Bye bye socket A

Also, why r134a? Does it have more capacity at these temps?

[thenrz]

Update:

Got a workspace set up in my not yet finished basement. Was finally able lay everything out and take an inventory. Apparently the dog (yes, dog!!!) decided to eat my oxy/mapp torch lines. I have a big propane torch for general brazing and the oxy/mapp for doing near compressor/condenser...anything that will suck more heat out of the joint. I'm thinking I can cut the lines square and get some metal barbs and epoxy them in...but I'm not so sure this is wise. Any comments?

Another question... For my load (500-650W abs max) is a 10 plate going to cut it? Would a 20 plate be necessary to maintain negative temps?