Morphing Air Conditioner into Autocascade System

[yngndrw]

That is very impressive. Can I ask how much it cost you to make that unit ? :o

[sjg0]

Uhh...I think it is a feat in itself that he made a machine capable of -140*C in the space of a small window air conditioner..

I have high expectations. What can i say?

[mytekcontrols]

So, uh, what is going on in that picture?
What's the bottom machine?

I'll start with your 2nd question first; The bottom machine is called a Helium Mass Spectrometer leak test station (HMS for short). It consists of: A High Vac diffusion pump, mechanical forepump (basic vacuum pump), A Cold Trap, and what is called a Cee Tube (consists of: Ion Gun, Filament Assembly, and Ion Collector). It also has a Roughing Pump, and a throttle valve between the Low Vac and the High Vac manifolds.

The entire purpose of this machine is to sense very small traces of helium while pumping on an external vessel, or in our case, refrigeration unit under leak test. So the basic idea is to either spray helium around the unit under test, or to "bag it" and fill the bag with helium. Since the unit will be under a vacuum (evacuated by HMS vacuum pumps), helium will travel through any holes that are present to the HMS to be detected.

Note: Helium is not readily available in our atmosphere, thus making it an excellent tracer gas for leak checking.

The HMS is a highly sensitive device, and will detect leaks of extremely small size, thus making it an excellent tool for leak checking refrigeration units, and especially cascade or autocascade systems with a multitude of solder joints.

If you guys are interested, I can post some pics of the HMS inner workings.

As for what is going on in the picture; I have the HMS operating in its High Vac mode, and the AC-2 unit is being used to cool its cold trap. The cold trap's purpose is to greatly enhance the High Vac's efficiency by trapping any water vapor that is present, something that is difficult for the diffusion pump to do on its own.

@mytekcontrols: I will be impressed when that thing can reach -250c

Now wouldn't that be something

That is very impressive. Can I ask how much it cost you to make that unit ?

It cost about $800, including refrigerant that I bought from the people I work for. This price is inclusive of the AC unit and the custom front panel costs. Not bad when you consider what it is doing, and how much something similar in a commercial product would cost.

[wdrzal]

I think a MOD should copy the first 603 posts (it's complete @ that point)to the Extreme hall of Fame.(to limit reading to 25 pages) There is no auto-cascades documented there. Nor anyone who took the time to explain and post like Mytek did about this auto-cascade build & it should be there to recognize his work & time given to documenting & teaching others. This thread can be allowed to continue to grow for further Q&A to benefit the forum.

Walt

[plasmatique]

Wholly agreed

[mytekcontrols]

Over the last week I started to notice a degradation in performance (evap temperature warming up). At first I thought it might be an oil freeze-up problem due to the extended run time, and the fact that I am still using the stock oil charge that came with the compressor. However when I powered the unit down and allowed it to fully warm-up, I noticed that the static balance pressure had dropped significantly (about 10 psi).

The pressure drop equated to approximately 1 psi per day. Yes, a very small, and potentially difficult leak to find .

My biggest worry was that the leak might be somewhere in my heat exchanger stack !?@#! Luckily it wasn't, and I was able to pinpoint it to the stem seal on the ASCO solenoid valve being used for buffering

Not having a sure fire method of fixing the problem, I opted instead to replace the valve with one of the valves we use to regulate cooling on the Polycold units I do service work on. This valve (Sporlan XUJ) is very similar to the Sporlan A3F1 which is hermetically sealed, and comes with 1/4" flare nipples on the In and Out connections. Leak fixed!

Now my next step will be to recharge and retest the unit. Because the leak was probably influencing my previous test results, I will also backtrack to an earlier lower quantity charge blend, and then adjust it up with measured additions to re-establish the best charge for this unit. I'll post the results sometime this coming week.

[Jack]

How long is your testing period usually?
It seems that you have been testing this unit for quite some time now...

And sorry to hear about the leak. It was looking so perfectly!

[plasmatique]

let us wait for suite with impatience

[mytekcontrols]

How long is your testing period usually?
It seems that you have been testing this unit for quite some time now...

Good question Jack

At this point I guess you could say I am in the life testing mode (basically turn it on, let it do its thing, and see how long it will work without failing). Keep in mind that the system it is connected to is fully operational the entire time. So it really isn't like many of the earlier tests, where I was trying to establish proper captube sizing, refrigerants to use, and load capacities for various applications.

As for unit design; everything is completely defined and most likely locked in, and should be unaffected by this last phase of testing. That is, unless a major problem is discovered along the way.

When this is concluded; some one should be able to follow this thread, build a duplicate unit, and obtain exactly the same results. If they opt for a different selection of refrigerants, or wish to scale the system size, then it is also my hope that enough information can be derived from the various tests and discussions, to aid in their success.

And of course like many of the other threads on this forum, this thread is a public form of my lab journal (in this public forum I have been covering the high lights, you would be amazed at all the test sheets and charts I have in my real journal).

[godmod]

I would be amazed if i could see and read those sheets

[HDCHOPPER]

thank you big time this thread has helped me a lot in understanding
and your ability is astounding in this field

alltho ss is where eye will start

[Polizei]

And of course like many of the other threads on this forum, this thread is a public form of my lab journal (in this public forum I have been covering the high lights, you would be amazed at all the test sheets and charts I have in my real journal).

I'm envisioning a couple 6" binders full of graphs and spec sheets.

[berkut]

Incredible project, reading this thread was like reading a good book, i couldn't wait for the next page

I also found out that my theoretical thoughts about auto cascades were proven right, and i learned a lot of new stuff. I feel guilty i stopped working on cascades years ago

Recently my "grand daddy" cascades II stage compressor decided to die, after reading this thread i think of copying your auto cascade using my I stage compressor

Maybe its time for a comeback to phase change, it would be interesting to find out how much have i forgotten after all this time and to test my new toys and skills which i usually use while working on cars

[mytekcontrols]

Recently my "grand daddy" cascades II stage compressor decided to die, after reading this thread i think of copying your auto cascade using my I stage compressor

berkut -- I hope you do and I wish you very good luck on this endeavor

Here are the latest specifications after fixing the leak.
Charge:

• 7oz R123
• 3.5oz R22
• 75psi R170
• 75psi R14
• 48psi Argon

Performance (with 21.5C Ambient Air Temperature):

• Suction Pressure = 26psi
• Discharge pressure = 162psi
• Compressor Discharge Temp = +70.3C
• Compressor Suction Temp = +3.2C
• Cold Probe Average Temp = -141.9C

The static balance pressure just after charging was 225psi. This dropped by 20psi after having been run, and then allowed to warm up over 2 days (refrigerant soak-in). Final BP appears to be about 205psi, although I anticipate this will rise and fall a couple of psi depending on the ambient conditions in the room.

The R123 and R22 were first charged into a small tank (modified steel suction accumulator) that was being weighed on a refrigerant charging scale. It was tricky, since I couldn't leave the hose connected to the tank while weighing, for fear of altering the readings. This necessitated adding the liquid in small bursts, closing the tank valve, disconnecting the hose, and then verifying the weight each time.

Once the proper amount of both refrigerant liquids was in the small tank, I then added it to the evacuated AC-2 unit. This was then followed by adding the gases based on pressure. In order to insure that all the liquid from the tank got transferred into the unit; I added the refrigerant gases through the top of the tank, and extracted the liquid from the bottom. Keeping the tank connected inline throughout the entire gas charging process.

The end results were quite good, letting me break the -140C barrier

[yngndrw]

Great job on breaking the -140C barrier with a single compressor. Just wondering about the "refrigerant soak-in" - What happens there, how can the static presure go down if it's a sealed system ?

[mytekcontrols]

"Soak-In" --- quite common in mixed refrigerant charge systems, and even more pronounced when the compressor utilizes some type of oil for lubrication.

Reason:
When initially charged, the warmer boiling refrigerants (liquids) don't fully mix with the oil. Also, the mixture of refrigerants hasn't 100% blended, or gone into solution with one another.

After being allowed to sit for many hours (may even take up to 2 days), everything blends as much as it can for a given room temperature (gases dissolve into other gases, gases dissolve into liquids, liquids dissolve into each other, ect.). As this all occurs the static BP will drop. Sometimes this reduction in pressure can be as much as 10%.

Shaking everything while charging will accelerate the blending process (or running the system for a bit can also do it). It just wasn't feasible in my situation to shake the whole unit. However I did shake the small transfer tank I used for the R123 and R22, and saw the pressure drop a good 40psi.

Also something worth mentioning; when I charge in the gases, I allow about 5 minutes to pass, and if I see the pressure drop, I top it up before moving on to the next gas in my charge list. Once again we are seeing a "soak-in" effect.

I hope that answers your question yngndrw

[yngndrw]

Ah I see, thanks for explaining that.

[[XC] gomeler]

Nice improvement! That heavy Argon charge is paying off well Would be really interesting to scale this up with about 2 to 3 times the capacity and try to use it on a processor. Could you use a larger system with a frequency modulator to adjust capacity? Would the autocascade eventually crash as you adjusted the compressor speed? Just thinking the lower stages might not enjoy the lowered massflow and I can't imagine using a ton of valves to adjust the capacity. Maybe a single solenoid on the evap feed to pulse refrigerant to maintain a temp but those solenoids aren't cheap.

[plasmatique]

-141,9 This merite respect!!!


solenoide? magnetic refregiration

[mytekcontrols]

Nice improvement! That heavy Argon charge is paying off well Would be really interesting to scale this up with about 2 to 3 times the capacity and try to use it on a processor.

Yeah it's a big increase in Argon, but as you noted it seems to be working well, and the running pressures have remained very reasonable. The subcooler is definitely doing what it's supposed to.

Scaling it up for CPU usage would indeed be a great project. Unfortunately it wont be me to do it, since I am looking to scale it down (more of a repackaging effort, then a capacity reduction), and end up creating a production ready prototype. I think there is money to made in this arena, all it takes is finding the right industrial/laboratory/medical marketplace. And as you know, not too many people know how to produce reliable working autocascades, especially very small ones.

Could you use a larger system with a frequency modulator to adjust capacity?

I don't see why not. In fact I think it would not only allow for excellent capacity adjustment, but also extend compressor life, by taking advantage of a ramp-up start-up ability.

Would the autocascade eventually crash as you adjusted the compressor speed? Just thinking the lower stages might not enjoy the lowered massflow...

I wouldn't call it crash, but if run too slow, you would eventually run out of steam. And yes the first signs of this might be a disconnection between stages, as you get out of the optimum range of pressure vs. condensation point for a given refrigerant.

I can't imagine using a ton of valves to adjust the capacity. Maybe a single solenoid on the evap feed to pulse refrigerant to maintain a temp but those solenoids aren't cheap.

Good point, and it could be made to work, but I suspect that the solenoid's life would be greatly reduced due to all the cycling required. Perhaps a better method would be to use an electrically controlled expansion valve. I think Sporlan offers something like this. And if coupled with a pwm output temperature controller, I think this would probably work quite well. It would simply replace the final captube.

-141,9 This merite respect!!!

Thanks plasmatique Quoting Rodney Dangerfield "I get no respect". Well luckily this isn't true for me, although in my business it doesn't come easily either, so it's always appreciatted.

[yngndrw]

Quoting Rodney Dangerfield "I get no respect". Well luckily this isn't true for me, although in my business it doesn't come easily either, so it's always appreciatted.

Well on here there are a lot of people (Both builders and non-builders.) who have a great deal of respect for you and you deserve it.

Good luck if you make a product out of your auto-cascades.

[berkut]

Thank you Michael.

Maybe some scaling up would be a good idea if i were to find some decent parts


A little OT on refrigerant soak in to the oil.

When i began to build phase change systems, when changing oil for the first time in a Danfoss SC10 i let all the oil out from the compressor through the suction line and a rubber pipe to a 2L glass. This was right after i discharged the system, well, when i lifted the glass and put it on my desk the shock from hitting it caused the oil to foam, all the refrigerant in it literally bursted forming a tall fountain. My PC, desk, walls, papers on it, everything within reach was all covered in mineral oil.

Looked very similar to a coca cola light + menthos experiment, or if anyone experimented with overheated water- dropping a single grain of salt in a cup of seriously overheated water.

[plasmatique]

It is sure, it ya a market for this type of product

I want etre inscribed on the waiting list for a central processing unit of treatment of this type...

[Marvin]

I think you deserve to be an xtreme legend :-)

[Xeon th MG Pony]

And remember kids for discharging refrigerant to atmospher you can be fined 25,000 dollars & stripped of your ticket and or jail time! Never mind the fact it is used in: Air dusters (R-134a & R-152a) Fire extinguishers FE-36!!!!

Mytek ever look into R-152a? As any and all computer air duster cans use it!