Another Small Cascade (or 2)

[Gray Mole]

While I was working on the quad headed gpu cooler, I chipped away at a couple of small cascades. This one's done now, figured I'd post it. Ended up being pretty good, slightly colder than the other small ones I've been making for some reason.

The specs...

Stage 1

Rotary compressor 10cc/30LRA
404a Refrigerant
20 plate heat exchanger
7' .036" Captube Brazed to return
10k btu condensor/fan

Stage 2

Rotary compressor 7.5cc/25LRA
1150 Refrigerant
Temprite oilsep/handvalve
1m Braided Flex
Duniek (I think) evaporator 40mm
Universal Chilly1 mount (modded to fit) with ALU plates
7.2' of .031" captube brazed to return + SLHX mod

It's installed in a medium size window AC case I modded to fit. Had to make panels and stuff close it up and fit displays as well as the venting to cool the compressors but it worked out pretty good. I couldn't focus on it totally since I have the other project I'm trying to finish, but I spent a few weeks slowly chipping away.

I've made another one even smaller but with basically the same setup, a few pics of this one, since it looks close to the same. The actual cascade doesn't take that long to make anymore after building quite a few I guess, but the finishing stuff like wiring and the case can be real time consumers.

So pics of how it's put together...






It's uninsulated on a few spots, makes it easier to see where the 'cold' goes I guess. The brazed captube is shown on the 1st stage but the second stage cap and flex are insulated. I got a pic of the brazed pipe that makes the very basic SLHX setup. I don't really want more heat exchange on the suction line for the second stage, since it could lead to overheat on the compressor. Just a bit helps precool the gas into the HX as well as helping vaporize the refrigerant on the way back to the compressor.

So the one I'm actually talking about is this one...






Didn't let it cool that long, usually sits around -35 to -37c or so on the display for the HX (-40ish in real temps) and about -110 to -115c showing on the Evap display when it's warmed up (so around -105 to -110c) with the displays being off. I'll get to that.

Finished it in white since it was already a white case, easier to repaint that way.

Handvalve is on top, easy to get at and only needs to be opened every few hours of running. Looks like a flower up there :p

The temps are pretty amazing on it, though when it warms up it runs at 'normal' cascade temps for a smaller one.

The displays are off but the red display (evap) shows 5 degrees too cold, and the blue one (the heat exchanger) shows 5 degrees too warm. Knowing that makes it more or less unimportant though. They're consistently 'off' so if you know how and how much, then working with them is fine.

Seems like every one of these displays (both types) are out by a bit, sometimes too warm and sometimes too cold. I buy quite a few since they're inexpensive compared to the brand name ones from the fridge shops. I'm going to see if I can get the suppliers in Hong Kong to get me a way to calibrate them myself. It shouldn't be a huge issue, but it's all in the price I guess. If it's 50 bucks more for that, then it's just not worth it. I try to make them as inexpensively as I can, since a cascade is already kind of a pricey toy, so whatever you can save you don't have to charge for.

When the compressors are still cold, they actually help it cool the gas, as well as the garage where the workshop is was only about 12c or so, so the unit was running unloaded about as cold as it could ever get. Unloaded temps don't mean a lot, but it's a cool number. The -130c or so temp is actually only -125c. But that 'colder' effect works throughout the range until the compressors start to dump heat into the system.

Actual working results are -110 to -115c unloaded when warm and 20c to 25c ambient temp. -100 to -105 at 200w. -85c when hot and 350w load after about 1hr of load. When totally cold and the first 10min of load it held around -95c at 350w, slowly rising to -85c and holding there. That's all measured on the Fluke, temps showed 5c colder on the display.

for such a small cascade it works really well. The amps are running around 6-7 at full load, 4-5 average. So around 500-600w of power usage. Not bad for the temps. The small Rotaries are pretty good that way.

Measurements of the case are around 18"x15"x15" so it ends up a really small cooler, not a lot bigger than a larger single stage. Using the tweaks and simplifying the slhx setup helps keep the internal size smaller, and it allows for a larger 20 plate HX. The HE condensor really helps too.

This is definitely the best result I've had on the small cascades though. I've been seeing around -100/-80c load on the small ones so far. This one just seemed to work better than the others, but getting the charge and captube just right isn't easy, and it doesn't always happen as well as you'd like.

Since I'm working on another cooler, figured I'd let this one run the tester through an hour or so of load cycling.

I tried to get the timer in there, so you could see the length of time on the increases in load, then the decreases with recovery times.

Basically, I ran it for around 10 minutes to get the compressors just a bit warmed up, so it wasn't showing that silly -130c number. Still not totally warmed up but close where it's starting.






Then, running a few minutes apiece at increasing load in 50w increments.

First pic shows recovery after the short 400w test.






After that, decreasing the load by the same, giving it 5 minutes or so for each setting to recover.

Last, I went straight for 350w for a bit to show the fully hot running temps (if you bench for longer periods then that's what to expect on a serious cpu at full load/high oc and vcore)


back to fully unloaded temp with the compressors heated up.

Basic results can be seen in pics, but you do have to allow for the 5 degree issue. Fully hot it runs -110c unloaded and at 350w it'll do about -90c taking that into account. Shows -115c/-95c unloaded/350w on the display.

Showed a pic of the resistor probe compared to the display. Like I said, around 5 degrees off.

It'll do 400w at a push but the HX does get warmer than I like. Have to add 5 degrees colder for it :p and that means it's getting under the -30c HX temp. I like it under that if possible.

Nice to see it push that high though. In cases with 6 core/12 thread intel stuff you can see some pretty insane wattage. Should handle most things.

So that's the works as far as a basic load test goes. Easier to do than to get pics that make sense and since I'm terrible at taking pics...

But yeah, if you follow along the timer on the power meter, you can see the basic idea of times, recovery etc.

Edit: I found a pot adjustment on the larger display Had to drill a small hole in the plastic housing but it's now calibrated to the fluke. Not going to take all the load test pics again though But it's reading the 5c warmer now, matching up with the Fluke's readings.

This is all copied and pasted from a couple posts I made elsewhere while XS has been down, sorry if it's a bit cluttered

Thanks

Gray

[Sk_rmouche]

Really impressive gray, that result with such tiny compressor, I'm jealous ;o

Did you think about filling with R23 instead of R1150 ?

[Gray Mole]

Hey

Yeah, well don't be too jealous I can't seem to reproduce this result in the other one I'm working on. Even smaller compressors (by just a little) but it's nothing like this one. Just a lucky guess on the captubing, everything just seems ideal. Every once in a while things just work out as good as it gets.

R23 is a gas I'd like to get a bottle of at some point. For a small cascade where temps aren't a serious requirement, something like -90/-70c would be likely with this kind of setup.

But it's way too much here I used it a lot in the UK when I lived there but some guy is selling a 5kg bottle for 1200 canadian and that's on the cheap side. It's more from my supplier. So over double, usually more like triple the price of the 1150.

So not for now anyways. Maybe if the summer goes well (I do auto AC on the side) then I may use the extra money to stock up on gasses. There's a few I'd like to get.

But in your 3 stage if you do go with R14, the 404/507 - R23 - R14 setup is a great setup.

When I finally get back to the 3 stage I have sitting there (just a pile of parts right now) I'll be going with R402a - R1150 - R50 if I can get it to work. R402a I think is the only way can maintain for the Methane in the 3rd stage.

But thank you I'm really happy with how this one worked out.


Gray

[Sk_rmouche]

Yeah, R23 ( and worst : R508b ) is a killer for price, have been lucky to get 6 kg ( ~ 10-12 lb ) for 200 euros.
I wanna try R1150 in third stage, but Air Liquide is really annoying with its prices and relation with individuals, that push me to looking for another brand :o

R14 is still a dream actually, didn't have serious seller with.

[Gray Mole]

Oh yeah 508b is just stupid here. Something like 1k per kilo or more Just too expensive to even consider though it's a great gas.

Ethylene doesn't need to be the high purity stuff for Cascades. If you tell them it's for a pressure fed cutter or something, the standard purity stuff isn't as expensive. Still something like 99.5% and I haven't had any issues, seen people comment on it before too.

That's only around 500 for a 5kg bottle here (plus the rental if it's Air Liquide of course ) so per cascade it's not that bad.

But if you could get that, you could try a Ethylene/Methane 3 stager. Temps have to be perfect on the first 2 stages though, really need an absolute minimum of -100c on the second HX under load. Colder would be better.

But the guy I deal with at AL here is pretty good I think the sales rep makes the biggest difference with them. Big companies do tend to be a bit impersonal.


Gray

[Sk_rmouche]

I didn't deal with AL for ethylene, I called them about R14 ( it's 800 euros per kilo by the way, cheap isn't it ? ) and R23 ( which AL don't sell anyway ).

I'm not a big fan of methane cascade in only three stages, because it has to run @ high pressure, both low side and high side ( to keep compression ratio in decent condition, not 25bars/0.5 bar you know ... ), and if so, static pressure will be I think very high, +30 bars quiet easily.

Here in France, AL is the market truster, but Messer, Linde and others little companies exists, otherwise I find your price a bit excessive for a low purity ethylene, even if I bet that methane doesn't cost less than that ( especially with AL, with them any gas is gold :/ ).

Getting a whole HX @ ~ -100°C with ethylene is I think possible with a GS26CLX and low temperature on first stage without stressing any of the three compressors.

So what do you prefer to fire up : R507/R23/R14 or R507/R1150/R50 ? :p

[Gray Mole]

Well for me it will be the latter using R50.

1. Because of the oil hassles of R14. I'm sure I have a way around, but I haven't bought R14 yet so I figure that's enough of a motivator on it's own.

2. Because it's hard to do Many 3 stagers with R14 have been done. R50 I've only seen once in a 3 stage and it wasn't great. Back then plate HX's weren't much used though.

Makes more sense to go with 'tried and tested' but I like the research. I think the tweaks I've worked out should help a lot though yes, the static can be high and a large suction volume/accumulators, etc. to cope with that are needed.

Air Liquide's pricey for sure, but the guy I work with gives me decent deals, though limited by AL's usual (get as much as you can) policies on pricing

But I just don't see a huge need when R14 comes to it.

I don't mean to slight anyone's work but I see, at best, 15 to 20 degrees max, sometimes less, better than a good Ethylene 2 stage.

I'd love to use Krypton, perfect gas for it. But it was 12 thousand I think? for a bottle here. Per cascade was something like 3-400 per unit so not the end of the world, but having 10k to spend on one bottle just wasn't happening and they only supply that one huge bottle.

But when I work out the price of gas, I try to think in terms of price per cascade since I'm making them somewhat regularly. If I were making one, I'd hate to spend 2k on a whole bottle :o

Gray

[Sk_rmouche]

Well, main reason that I'm not comfortable using methane @ third stage is compression ratio and static pressure. Beside that it's not an outstanding refrigerant. My point is : there's not alot of fluid to fill that -130/150°C range.

Krypton, I've called Linde and Messer, it's something around 3 euros per gram, for a kilo, Messer gave me an invoice at 2200 euros per kilo I think ... Way too much for my student pocket ! And beside that, Krypton has small vaporization heat, don't know if it's the right choice.

Something interesting to try eventually if using methane @ third stage, will be to have a high BP. If you take a look at the P/T chart of methane, you read -108.5°C @ 20 bars for instance. Holding that kind of T° with ethylene, load applied is quiet a challenge, isn't it ? Condensing @ 20 bars isn't that matter, as your low side is as high as 2 or 3 bars, which is already @ low temp, but to get that point, you have to fill alot of gas, which leads to very high static pressure.

You have to see that two stages with ethylene isn't and it will never be "efficiency friendly" as long as you know that piston type compressor should not running > 7 in compression ratio H24

[Gray Mole]

I know what you mean about the Ethylene on 2 stages. Before the captube mods and stuff I had a pretty awkward time with it. I'm getting enough of a reduction in condensing temps from it that it's working well now. As a HX gas on a second stage it's going to be very difficult. I'm definitely going to have to size the base and second compressors a lot larger than the third.

I still don't know if I'll get pure Methane in stage 3 or if I'll have to dilute it with ethylene.

You're right though, there really aren't that many gasses suitable for a -150c 3 stage. Krypton really looks like it could be but the price means...well NO really though if I win a lottery or something I'll try it out. It's as close to ideal as I've been able to find.

Ethylene and Methane both like a higher BP being as light as they are, but they're also light enough to reduce the compressor strain and they tend to play well with the captube mods I'm doing too (or at least I hope Methane does ) so we'll see.

Like I said though, a few things to get through before I'm back on it.

But I have a 16cc horizontal rotary, a 13cc and a 12cc upright rotary and I may downsize the 3rd stage compressor, or if I can find one, upsize stage 2. Got a double length 14 plate for stage 1 and a 20 plate for stage 2.

hoping that's all enough to make it fly. I know I'll get idle and up to 200w shouldn't be any real issue but over that is where I'm concerned.

But I'm a month at least away I think. Hoping to start making it sooner but that's the likely time to get to work on it. Going to have to order more evaps and stuff after I sell this thing though. Had 10 or 15 around when I took a break from building before but only a couple left

Gray

[Sk_rmouche]

hey Gray

About gasses, you can also use NF3 ( nitrogen trifluoride ), but price is quiet higher than R14, so ...

Nothing reliable about mixing R1150 and R50 too.

I'm totally supporting you, I'm just telling you what I see. By the way, if you're going to get some methane, why not trying a 4 stages ? I've in mind to try something new maybe ... since I don't have many space in my case, maybe make an auto-cascade with last stage, ethylene and methane ( or better, R14 with methane ), but it needs a huge compressor, space problem still

Can't wait to see your results with such tiny compressor ( no offense, I've two like yours, rotary and piston type ), mine is 36 and 21cc, not tested with true load, I've to finish last stage.

About ethylene, maybe we could make group order ?

[Gray Mole]

I think the shipping from France to Canada would kill the savings

So the main reason for me to try the 3 stage with Methane is really just to experiment. I'm not making it for anyone in particular so I'll have plenty of time to try different things. Just that I'd like to see it really work. I've seen it tried and never did any better than a R14 3 stage in actual load testing.

But I've figured out how to make my small 2 stage Ethylene stuff work well, so I figure I'll try it out with Methane. Worst case? It doesn't work and I swap to something else.

If I do go with a 4 stage to use Methane then I'll probably use Ethane/Ethylene/Methane if I can't find cheap R14 here. I haven't asked about price yet but it's not going to be cheap.

Not sure but I do think that I could use a large 3rd stage with the Ethylene and maintain -110c or better under load with a HX temp around -70 on that stage.

A really small 3 stage with Ethylene and Ethane would work close to as well as R14 I think without the oil issues. I'm getting Ethylene to work (just barely to be honest) well with small compressors and really I do like the compact stuff but it's a large jump for it.

Just thinking out loud though. Wish I could start sooner doing cascades that are the same over and over gets kind of boring. I'd like something new to do.

Gray

[Sk_rmouche]

I understand. R14 if you can get some is an outstanding refrigerant, definitely... Even with oil issue. For me it'll beat any ethylene cascade, only because compression ratio is so much better than R404a/R1150, for instance ( and similar displacement obviously ).

Anyway, seeing your results is clearly a good point, related to my displacement's. Otherwise I can't find good enthalpy chart ( heat vaporization and others ) for R14 R23 and R1150, did you get them ?

In Canada, Airgas sells ethylene isn't it ?

[Sk_rmouche]

I understand. R14 if you can get some is an outstanding refrigerant, definitely... Even with oil issue. For me it'll beat any ethylene cascade, only because compression ratio is so much better than R404a/R1150, for instance ( and similar displacement obviously ).

Anyway, seeing your results is clearly a good point, related to my displacement's. Otherwise I can't find good enthalpy chart ( heat vaporization and others ) for R14 R23 and R1150, did you get them ?

In Canada, Airgas sells ethylene isn't it ?

[Gray Mole]

I have an account with Air Liquide so I've just been using them. Didn't realise until later that AL is one of the highest priced companies so I may shop around one day.

There's a few other companies I can go to I suppose. I do AC in cars on the side (think I mentioned that before) so after the summer I'll probably start looking at upgrading my 'stock' of gasses again. When spring/summer start to hit me I'm usually too busy to do a whole lot with the PC cooling. That and if I make enough I can afford more gasses for the 'hobby' stuff

R14 is a gas I'd like to have though. I know can get through the oil issues but it's always been that 'almost there' feeling with it.

I mean, the third stage adds a huge amount of cost and hassle to making a cascade, and if you do it right you can get -90 to -100 and sometimes -110c load temps from a good 2 stage with ethylene. Larger compressors than this one of course, but a few guys out there have built 300w cascades that'll hold in the -100 to -110c range at 300w+ with bit compressors.

I see R14 3 stagers only getting -120 to -130 at best with load, mostly closer to -120c.

I just don't see the value. Say it's 1000 bucks for a 2 stage (or at least I try to make them that cheap if I can) with Ethylene, and say it's like this one running the -90 to -100c load temps.

Now I make it into 3 stages, but the cost of the case size increase, another compressor/HX/oilsep (or 2 for R14) and it's like 600 to 800 more to make it right? So it's now 1800 bucks (and for 3 stages, still pretty cheap) but the benefit is only another 20 degrees colder.

That's why I'm looking for a good -150c range gas. Even if it were only -130c under higher (300 to 350w) load, it's still a 30-40 degree boost. Seems more cost beneficial that way.

Oh, I think Air Liquide's site has a bunch of charts for the low temp gasses. Not sure if they have the Enthalpy charts there or not.

http://www.ipu.dk/English/IPU-Manufa....aspx?lg=print Coolpack works though


Gray

[Calathea]

Impressive performance, especially considered the power draw. Nice!

[Sk_rmouche]

Thanks Gray !

You're right, colder we want, extra money come with. R14 is just the best gas we can put with a second stage filled with R23. Compression ratios are quiet balanced that way. Also, you know my opinion about ethylene and methane, but if you could make it works, why not

For me, I'll try methane with high BP, for not stressing too much compressors, dead volume will still be oversized, because with that kind of run you need alot of fluid.

By the way, good R14 cascade can make < -130°C with load, just like 2 stager ethylene do, you're talking about -100°C loaded with R1150, it's around 1 bar, and R14 boils @ -128°C to atmospherical pressure, so -130°C is pretty easy to get, add a fourth stage filled with pure methane, and you get quiet a nice cascade to make WR with any CPU

[n00b 0f l337]

With the adoption of r410a compressors, R50 will be more viable. But I'm still leaning to autocascade technologies.

When you add a 3rd or 4th stage, the gas goes from ambient (after desuperheater) and has to cool all the way to -70 or 80C (delta 100C or more!) before you even condense it. The efficiency starts to suck. But with autocascade, each next stage only cools a bit more.

Awesome work as always Gray. I want to make up a package with a Variac, kill a watt, and load cells and mail it to you though, that old ones looking like it took a beating from all the experience its getting!

[Sk_rmouche]

When you add a 3rd or 4th stage, the gas goes from ambient (after desuperheater) and has to cool all the way to -70 or 80C (delta 100C or more!) before you even condense it. The efficiency starts to suck. But with autocascade, each next stage only cools a bit more.

So you recommend making an auto-cascade for the third stage, R14 + R50 for instance ? I always thought that auto-cascade needs huge compressor to get some raw power @ the evaporator, right ? Plus, It seems to be very tricky to tune, once you get 3 or more gasses, isn't it ?

[Gray Mole]

Yeah, the stuff all works fine but it sure is getting ugly

I just finished the new cartridge based tester though. Cost a hundred bucks for a 500w 110v cartridge and they didn't want to make it for me as it would burn out if you actually used it for a heater but it's great

Will have to strip it and update the old thread I made on it with some pics. Very accurate. The built in K probe in the resistor is ideal for tuning too, can fine tune that last bit to get the best resistor internal temp.

My Auto's sucked. Every one of them did, though I haven't made one for 5 years at least now. The huge one was reasonable but you really need a variety of gasses I don't have right now.

Guess all I'd really need is R23 or ethane, and some argon, and I'd have what I needed. Maybe I could now, with a couple more years experience, but they take so long to cool. Always bugged me, so I focused on the cascades instead. Nice about them though, you don't need a huge compressor to make them work well. You can't get the same relationship with back pressure since it's a 'shared load' in the gasses, but a 1/2hp rotary running 10psi with a decent amount of argon is going to get pretty cold, and I hear about them holding 300w (made well by someone who's good with them)

Oh and you always have to account for that efficiency loss in the latter stages but the tweaks help. Any subcooling mods and slhx setups do a lot. Desuperheaters are mandatory. All the little things add up, and I'm even starting to use that mesh on the rotaries, just to get that tiny boost on load. I guess whether I think it'll help or not, and even if it's only a degree or 2, it's still a boost.


Gray

[Gray Mole]

So I know I need to clean the load test gear

This is another load analysis, figured I'll try to do a basic set of results for everything I do, or at least every time it's a different evap or a different kind of unit (Cascade, -50 unit, -20 unit, etc) on my own evap.

This is the smaller of these 2, with the blue braid.

Runs to close to -100c no load if you leave it a while, but coming just off the initial load of 200w, then to 0 just until it balances gives an idea of where the unit stands in real terms.

So this one, -86 or so. 5 - 10 or more colder is fairly normal from the freshly loaded / unloaded temp to the 'sitting for a while to look cold' temp.

so for those that followed the 'Daisy evap' thread, this is the same kind of deal. I do expect that the differential would be worse on a cascade on mine just as this one is worse. Low temp heat travel should be slower than higher temps. I'll be confirming that, soon as I make a cascade with my evap on it. Should come soon enough. I have enough for 2 more cascades after this one, so I'll probably do them, then gather what I need for a better revision of a 3 stager.

The evap is my last of the old Duniek stepper evaps. The base is thick on it. I did a bit of grinding and modding much as I could to improve things, but a thicker than 4mm base (imo) will give worse differential results.

Anyways, the set of pics. Awful, grainy, but it's the numbers that matter, not the pics.



This is the cooler, really small one



0w, -86, just off load. Got a shot as it balance the evap and heat cartridge internal temp.



50w, -84c, 9 degree differential



100w, -83c, 17 degree differential



150w, -82, 24 degree differential



200w, -82 (still, and I'm seeing a trend with that midrange load, and little difference in evap even with cartridge temp change) 30 degree differential



250w, -81ish, 36 degree differential



300w, -80.9c, 37 and a bit differential



325w (max), -79c, 45 degree differential.


So that's it for this one. Not too bad really, considering the effect of lower temps. I don't have another Duniek to test, but went over the results I'd found so far on the SS stuff at high load, this was closer to 40 differential so I do think that colder means a bigger temp difference.

Hope the pics weren't too ugly. Unit's temps for a cascade aren't the best, but this quick load test is just for my own reference mostly. Wish I'd taken internal pics of all the blocks i'd used now, but the duniek was similar to the smaller of Piotres evap in design. No idea what the actual specs on Piotres' evaps are now for the base and plate thinknesses, mass and locations, so I wouldn't say the same, just a similar format.


Gray

[Kevin Hotton]

It would be nice if you had suction and discharge pressure gauges to cross-check temperatures. I have seen Fluke (and other TC readers) report inaccurate temperatures (reading lower than actual temperature) as the battery gets low on charge. Next time change your battery to a fresh one and see if temperature reported does not go up.

Kevin