Massive Water Chiller

[Stewie007]

Thanks.

It will be some time before I get to put it on the intended load. However, I'll do some effective load testing with maximum of 1500 watts worth of heat load. Tuning this will be something... The compressor can handle quite a lot of load at the temps I'll be running.

I'm going to set aside a day where I will be fully run testing the unit and observing for hours at a time to ensure proper cycle operation and to adjust for issues. I'll create a cycling load that will simulate the variable nature of the components.

I am open to suggestion as far as charging goes. I am probably going to liquid charge for the first shot, and then vapor charge to the suction to make necessary adjustments. For that process I'll have to rent a scale as I've got a refrigerant charge limit of 3lbs given that the receiver has an actual holding capacity of 2.4lbs of 404a/507, and the condenser coil volume.

At this point, I'm not sure where to start. hehe

[runmc]

I would liquid charge all the way just to be on the safe side since you are using blends. Some may disagree with me but that's how I was taught. I can't imagine you putting anywhere near 3lbs of refrigerant in the unit but maybe

[Stewie007]

I will charge mostly by liquid, as I am more comfortable doing so as well. Though, I chose 507 given its ability to vapor charge as well. I figured that would be easier to do while the compressor is running to vapor charge to the suction side. However, if that is an inaccurate assumption then there is no need to do it that way. Is that an inaccurate assumption?

As far as how much refrigerant.... it beats the heck out of me... My experience is in design, application, and building.... not really up to speed on determining adequate charge hehehe.... I'll rely on you and others for that expertise..

[Daddeo2]

I agree with runmc. An initial charge can be with liquid, maybe a lb. to be safe, then you could trickle in a liquid charge slowly while the unit is running to fine tune it. Just be very careful not to slug the compressor by trying to add to much to fast. I can't see that system going over 3 lbs. either. What's the HP of the compressor?

[Daddeo2]

Nevermind. Looks like 3/4 HP. Google is your friend.

[limpkorn]

From my understanding the blended refrigerant must be liquid charged otherwise you are not charging the right portions of the blended refrigerants.

Just use your guage set to slowly charge it in like Daddeo said you want to be careful not to slug the compressor.

[runmc]

On the outside of the boxes of r507 I've seen it has arrow pointing to the bottom of the bottle saying this side up for use. To me that means liquid charge and that's what I've always done. I would start with 50psi static charge. Start the unit and then slowly add in very small amount. Maybe a 2 second blast. Wait 10 - 15 min to give unit time to equalize out the charge. I watch for frost returning on the suction line keeping it back a couple of inches from compressor. After I get it in the ballpark I add the load I want and tweak from there.

[Daddeo2]

". I watch for frost returning on the suction line keeping it back a couple of inches from compressor. After I get it in the ballpark I add the load I want and tweak from there.[/QUOTE]

Agreed. Frost too close to the compressor could increase your chances of slugging.

[Stewie007]

I've always learned to never charge a liquid to the suction line. However, despite that 507 can be charged as a vapor, I think I'll go with the recommendations and carefully charge as a liquid on the suction line. I was just uncomfortable given that the compressor is a direct suction to chamber delivery (albeit protected by an accumulator).

That said, R507 and R404a are two different animals. 507 acts as a single refrigerant in vapor and liquid form, and if you look at its PT Chart, it will confirm that. There is not the problem with temperature glide as with 404a. Part of the reason I chose it.

All this talk of charging is getting me excited hehe.... I'm a bit bummed out that I can't buy any glycol until the 15th of September.... hehe

[limpkorn]

I don't know how R507A acts but I do know that it is composed of 50% R125 and 50% R143A that would lead me to believe that when you charge as vapor you may change that 50/50 mix I would just not do it, but anyway you are using R404A which is a ternary blended refrigerant so you definitely must charge as a liquid.

Just charge slowly as liquid keeping your hand on the valve feeding it a few seconds at a time I have no idea how small amounts of liquid would affect a rotary compressor, but I doubt it would be a big deal if you did get a tiny amount through on initial charge.

[Stewie007]

I don't know how R507A acts but I do know that it is composed of 50% R125 and 50% R143A that would lead me to believe that when you charge as vapor you may change that 50/50 mix I would just not do it, but anyway you are using R404A which is a ternary blended refrigerant so you definitely must charge as a liquid.

Just charge slowly as liquid keeping your hand on the valve feeding it a few seconds at a time I have no idea how small amounts of liquid would affect a rotary compressor, but I doubt it would be a big deal if you did get a tiny amount through on initial charge.

No, I chose R507. Though, I can see how you that was misleading. I was just differentiating between 404a and 507... although they aren't all that different in terms of make-up.... 507 is an Azeotropic blend and 404a is not quite..... But its all irrelevant now, since I'm going to liquid charge it anyways. (Not exactly two different animals as I said... I know... I exaggerated. )

[Moc]

The normal way to fill a system with an expansion valve and a receiver is to start with a initial guess (i'd say 500g with your system...) and then beginn to apply load and check the sightglass!
With your maximum load there should not be bubbles in your sightglass!
(The more professional approach would be to measure superheat and subcooling but thats nothing I can write quickly^^ )

We german builders use cheap immersion heaters from ebay to loadtest our units, its a lot easier then connecting chiller and PC plus you have a specific load so you can judge the performance of your building !

[Stewie007]

I'm going to utilize some immersion coils, yes. I don't intend on connecting it to the computer without having tested it extensively with a known load.

This will be as close to a professional test stand as it can get hehe. I'm going to charge by superheat, given that the TXV will measure by superheat. That is the best way to guarantee maximum efficiency of the design. Though, I'll probably utilize a mixture of techniques that I've learned to determine the correct amounts of charge and performance.

That said, its going to be interesting; I have never done it before, hehe.

The following will be measured :

EWT and LWT from the evaporator.
EWT and LWT from the condenser
EWT and LWT from condenser to the reservoir.
Superheat and Subcooling

I am aiming for a low condensing temp to save on energy as much as possible, and to increase the load capacity, but not too low to cause the superheat to "crash."

All in all, its going to be a lengthy period of observation and adjustment with a simulated and variable load. I want to know what will be happening at <300 Watts and what will happen at 300 to 1000W, and >1000 Watts.

I'm not putting it in a computer until my design is proven hehehe......

[Stewie007]

Didn't do much today.... I did complete assembly of the junction box.. I'm a bit peeved that the top contactor is crooked though.... :p That's the start capacitor on the outside there... I'll protect it better than that, but it will only get voltage for a short time anyways... Couldn't fit it inside hehe....

[Stewie007]

Did some wiring today... All I gotta say is that its extremely frustrating hehehe.... It looks a bit messy, but that's sorta the way it is. Lots of wiring. Most of it is from the compressor wiring hehe.. Conduits weren't working, so I used anchors. So far I've got the compressor wired to the low pressure switch. I've wired the solenoid and the condenser pump so far to the terminal block. I've got the compressor and and the solenoid common wired to the middle terminal.

Not much more to do but wire a few more components. Almost done.

That said, it was very frustrating. Still git a little bit to do... but very frurstrating..

[Stewie007]

I did some work, but nothing really worth posting. I let the vacuum pump run all day as I did some wiring and streamlining. At this point, I can't really do anything else until I get the base built. I have to build the base so that I can route the rest of the piping. Plus, my plastic didn't arrive... online metals has gotten really slow... Ordered on Tuesday... still listed as "checking stock..." WTF...

At the end of the day I got the vacuum down to 470 Microns.... There will be a pause as I have to save up some money for the rest of the parts...

[Daddeo2]

Still very much looking forward to seeing your completed project in action. Keep up the great work.

[Stewie007]

While there is a lull in activity I figured I'd ask some questions.

I was considering the use of either full cover or "universal" blocks for the GPU(s)..... I figure some decent Ram sinks would be fine, but then I have the insulation to consider. A full cover would be difficult to insulate near the socket, while a universal block like the EK VGA Supreme may run into Ram cooling issues since I'll need at lest 1/2 walls (if not 3/4)...

That said, its also possible that the VGA Ram Heatsinks would help to dehumidify the air next to the insulation warranting a slightly thinner wall. That single lonely RAM chip next to the PCIe connector is of concern, however.... So its still up in the air...

I am considering a specialized connection for full cover blocks, so that isn't a concern right now. I just want to get rid of plastic. Unless you think POM Acetal would be fine at 0-15F. I know Acetal (Delrin) has some pretty decent low temperature resistance, but I'm not convinced.

[Stewie007]

I find myself leaning towards one of these.

I think I can do a decent job insulating beneath the block as well as around it... One Sided RAM chips mean the back side insulation can be full PCB, and thinner walled.

[Crabby]

That full cover block looks to be perfect for chilled water. What do you plan to use for insulation between the pcb and block?

[Stewie007]

That full cover block looks to be perfect for chilled water. What do you plan to use for insulation between the pcb and block?

I have not determined fully. It will be the combination of Polyurethane foam insulation (with a contact cement applied to non contact areas of the block to promote adhesion. On the PCB, I'm researching protective coatings used in marine applications (PCB coatings) for a barrier layer. I'll likely utilize a dielectric grease to fill other gaps. I'll seal with urethane foam in critical areas then encase in a proper insulation. All o-rings that are not Buno-N will be changed to Buno-N. Though, I suspect they already are given that many of these manufacturers are using it.

The back of the PCB will be coated and then insulated as well. I'll have to use grease in the PCIe slot, and encase it as well given the proximity of the card and block to the motherboard.

Any plastic parts will be removed from the block and the inlets will be customized.

[tool_462]

http://www.alliedelec.com/search/pro...x?sku=70207172

[Stewie007]

That'll do it. Just gotta be careful of over spray... But that's the stuff. I'll just have to use some tape to protect the critical parts, but that should add a good bit of moisture resistance.

[Buckeye]

Have you thought of using eraser ?

[Stewie007]

Well, this is more of a coating to protect the PCB and components. As far as the eraser is concerned, I may elect to go another route. I can get insulation sheets as thin as 1/8" and they will not conduct heat as well as the eraser. Depending on how much compression stress it would place on the PCB, I may be able to get away with 1/4" sheets.

In spaces where coverage by insulation is not possible, I'll use dielectric grease to fill the gaps beneath the block. The block will be well insulated on the exterior. On the socket, I'll use a spray foam that I can shape after it dries. Or maybe I'll use create a shape mold....