At this point I am more concerned with cap tube sizing than pressures.
I would like to see what it does with the present charge.
Printable View
At this point I am more concerned with cap tube sizing than pressures.
I would like to see what it does with the present charge.
I have no way of getting the charge the same as last night, only the processor wattage. I've cranked it down and this is what I have.
R23
Evap Exit: -51ºC
SST: -63ºC (1.7barg)
SCT: -10ºC (18barg)
R290
Evap Exit: 19ºC
SST: -17ºC (1.7barg)
SCT: 55ºC (18barg)
Air in: 24.0ºC
Air out: 28.9ºC
CPU: -15ºC
I removed a good bit of refrigerant as my suction side was 2.9barg after I put in alot of propane to see if it would have any affect on propane evap exit temps. They just rose.
What were the temps after you reduced the load, and before you changed the refrigerant charge?
I reduced the charge then dropped the load so I dont know. I must drive you crazy Gary. From now on, I will follow you to the T and wont touch anything, promise :)
So whats next?
Everyone drives me crazy. Or would that be... drives me crazier?
Let's chop another foot off the R290 cap tube.
Good one Gary ... I knew a wood cutter once.....................Quote:
Originally posted by Gary Lloyd
Everyone drives me crazy. Or would that be... drives me crazier?
Let's chop another foot off the R290 cap tube.
The pressures both evaporators will always be the same Unless we do something to control this. If we have a way to dial in the low stage evap controlling both the suction and metering of the refrigerant to keep it at a proper condensing temp/pressure for r23, this would solve the instability of the autocascade. So if you have an EPR to control the R290 evap. And a hand expansion device and an epr on the R23 side you could dial it in
So if we say keep the suction on the r 290 at 5 psi and the R23 suction at say 20 psi you could use the hand valve to keep the load at a minimum from the r23 and optimize the performance of the R290 side.. Then increase the load by opening the hand valve... Of course you will need to monitor the suction at the compressor it should be no more than 5 psi or the R290 evap will warm up. This is just speculation, as I have only tried this on R409 and R410.. it worked but it was hard to tell the benefit from these but I did notice a lower head pressure with more charge. It is going to be a matter of trial and error til we come up with a formula for building these.
Maybe I'm looking at this all wrong, but as I see it, both evaporators need to be fully active with an outlet superheat of 3-5K/5.4-9F. At this point, the suction pressure will be determined by the heat load versus the compressor's pumping capacity.
Of course that pumping capacity is effected by the high side pressure. While the overall internal volume of the system is sufficient to hold down the static pressure, the R23 portion of the high side seems to be insufficient to keep the high side pressure down.
In looking at the pics, it looks as if the R23 is in the tube of the interstage, while the R290 is in the shell (outer tube). This would be the opposite of how it should be, from a heat exchange standpoint, a velocity standpoint, and an internal volume standpoint.
The shell would/should have the larger internal volume, which is what we would want for the R23 vapor.
The tube would have the smaller internal volume which would reduce the R290 charge needed (less liquid to fill the tube), and the velocity would be higher, which helps to carry the oil. This would be doubly important as the R290 evaporator feed is bottom up.
The EPR's would not regulate the heat load, but simply the evaporator temps.
Am I missing anything here?
Russell_hq, am I looking at the pics wrong, or do we need to swap the tubes on both ends of the interstage, so that the R23 condenses in the outer tube?
What if we charge the system with R290 then balance the system so both evaps are the same temperature, then slowly add r23 monitoring the temperature of both evaps inlet and outlets and the system pressures...
Hmmmm... an interesting idea. In effect, the interstage would become a subcooler. This might help us to balance the cap tubes. Hmmmm...
It would be helpful to have shut-off valves in both liquid lines. With the R23 valve closed we could rough charge the R290 circuit and then open the valve and rough charge the R23. With both shut off we could pump down the system.
You are right with the pictures Gary, so you think switching the tubes will help? Evaporate the R290 in the 1/4" and condense the R23 in the 1/2" . Thats going to be fun to sort out with all the expanded foam just inches away :D
When I was charging i would add some R23 for something to run around the second stage, then added the R290 until I started getting liquid in the sight glass. I would monitor the R23 evap exit and add a little R290, if the R23 superheat went down I would add a little more R290 as I figured I was condensing more R23 with the extra propane. I would do this until the R23 evap temps started to rise and superheat stopped dropping, then i would add a little R23 (evap temp would go down a bit) until superheat was too high and I would start again.
One observation I have made when appling these systems to cpus is I get the absolute lowest temperature when the evap superheat is between 10-20K. When its between 5-10K my temperatures are not at their lowest and my system always has more charge. Not sure why that is. :confused:
I think it will make a very big difference. It needs to be done. Put wet rags between the part you are brazing and the foam.Quote:
You are right with the pictures Gary, so you think switching the tubes will help? Evaporate the R290 in the 1/4" and condense the R23 in the 1/2" . Thats going to be fun to sort out with all the expanded foam just inches away
This sounds reasonable.Quote:
When I was charging i would add some R23 for something to run around the second stage, then added the R290 until I started getting liquid in the sight glass. I would monitor the R23 evap exit and add a little R290, if the R23 superheat went down I would add a little more R290 as I figured I was condensing more R23 with the extra propane. I would do this until the R23 evap temps started to rise and superheat stopped dropping, then i would add a little R23 (evap temp would go down a bit) until superheat was too high and I would start again.
Is that absolute lowest evap temps or absolute lowest CPU temps? In the end, it is the CPU temps under heaviest load that are most important. They are the goal.Quote:
One observation I have made when appling these systems to cpus is I get the absolute lowest temperature when the evap superheat is between 10-20K. When its between 5-10K my temperatures are not at their lowest and my system always has more charge. Not sure why that is.
Theoretically, the ideal is 3-5K superheat. The lower the superheat, the heavier the heat load the evap can handle, but evaporator designs vary. Theory gives us a starting point, but in the real world your mileage may vary.
That would be cpu temps. Wet rags will definitly be the order of the day when I get round to rebrazing this thing, this is what I get when I listen to aenigma and not you. It was easier to braze it up the way I did, now i will have to do alot of choping and joining, gah. I can see my buying more O2 for this thing.
Aenigma generally knows his stuff, but none of us knows everything.
On the evaporator outlet superheat, I find I am having to re-think some of my fine tuning principles. That's probably why I find the OCing sites so interesting. These systems present a new challenge for me :D
I have already changed my thinking on cap tube sizing, as the goals in reaching minimum temperature are different from those of going for fastest pulldown, thus the cap tubing sizing goals are different.
Now I find myself looking very hard at the principles of evaporator feeding. On conventional systems, the heat is absorbed evenly thoughout the evaporator, so minimum superheat at the evaporator outlet ensures even distribution of the refrigerant throughout the evaporator. On these direct die blocks, the heat load is concentrated in the center of the block face, and that changes everything. Low outlet superheat isn't necessarily going to get us maximum performance.
These systems are a learning opportunity for me as well as the forum members. I love it. :D
Yes the phase seperator would act only as a distributor in this case. Once you get the evaporator to operate at the temperature you want to condense the r23 at then you add the r23, there would still be issues as the pressure increased the R290 feeding into the interstage heatexchanger would increase also but the heat load would also increase and if we get the sizing right it should all be proportional. The only thing I haven't worked out is the enthalpy and entropy of r23 related to r290 if we use these values as a basis for the proportionality of the heatabsorbing surface (block) to the interstage heatexchanger we should be able to get it right on the money... I saw some formula on this in some research paper, I will try to see if I can find it again...Quote:
Originally posted by Gary Lloyd
Hmmmm... an interesting idea. In effect, the interstage would become a subcooler. This might help us to balance the cap tubes. Hmmmm...
Valves would help in the setup, but pev would alloow for correction of the balance during operation so we could compensate for the varying load, mabe later we can use the cpu to run these with stepper motors if we can logically describe the operation of an autocascade.Quote:
It would be helpful to have shut-off valves in both liquid lines. With the R23 valve closed we could rough charge the R290 circuit and then open the valve and rough charge the R23. With both shut off we could pump down the system.
Gee thanks for that.But I dont remember giving you much advice, if any, I just kind of agreed with what you wanted to do and generally chatted with you.The main thing I recomended was a txv, it makes life much easier.Everything else was your own doing.Dont try to blame anything on me, I wasnt holding your hand through it.Quote:
Originally posted by Russell_hq
this is what I get when I listen to aenigma and not you.
I am assuming youre talking about the interstage heat exchanger, then yes the way I used mine was the 1/4" is for condensing and the 1/2" shell for evaporation because this gives the most liquid refrigerant contact with the 2nd stage condenser(flooded of course).And this works just fine for me.
I guess this is what I get for trying to chat with you. :stick:
How about taking a full set of measurements for us, then switch the lines and take another set of measurements for comparison?Quote:
I am assuming youre talking about the interstage heat exchanger, then yes the way I used mine was the 1/4" is for condensing and the 1/2" shell for evaporation because this gives the most liquid refrigerant contact with the 2nd stage condenser(flooded of course).And this works just fine for me.
Then you can show us how much finer it works. :D
Also take static measurements for both configurations.
not niceQuote:
Originally posted by Russell_hq
That would be cpu temps. Wet rags will definitly be the order of the day when I get round to rebrazing this thing, this is what I get when I listen to aenigma and not you. It was easier to braze it up the way I did, now i will have to do alot of choping and joining, gah. I can see my buying more O2 for this thing.
Hey captn... whats up....Quote:
Originally posted by captaincascade
not nice
ok so aparently this little thing is over heat exchangers?
tube in tube?
evap on outside or inside? let me ask you this. on a water cooled condenser, where is the water? the outside or the inside? the water is on the outside because that covers the most surface area for the inside to condense refrigerant to liquid. Another reason why we condense on the inner tube for a cascade system, is the amount of heat you would pick up from your second stage being on the outside. We do not want to insulate the first stage evap. we want to insulate the second stage condenser. what i mean by that is we are wrapping it up with the best cold insulater in the world....more cold.. is it me or do i hardly make any sense?
I dont use that heat exchanger anymore, note I said "used" not "use".
The inner pipe being evaporator and outer pipe being condenser seems backwards to me.
The interstage I now use is a tube in shell I made with 2 1/8" pipe and 2 parallel runs of 3/16(easy to bend) and a capped 3/4" pipe in the middle of the 3/16" coils for a liquid displacer.I also use this one the same way, the shell is the evaporator and the tubes are condenser.
I use a 1/2" od corrigated pipe in a 3/4 inch id pipe 12 inches long with the inside of the 1/2" pipe as the evaporator I also coil the cap tube around the outside as well as tie the filter drier next to it. then I insulate with 1 1/2" of closed cell neoprene. This system using 2x 1/7 hp compressors will run my evap at -125 unloaded and -104C loaded at idle and -92C benching. (GPU 9800 XT 2.2Vcore) Orginally I had it the otherway around and the tempos were -87C unloaded and -85 loaded and benching.
The inside. When you pull the end bells to punch the tubes, they are water tubes.Quote:
let me ask you this. on a water cooled condenser, where is the water? the outside or the inside?
As to the insulation argument, we have two fluids below room temp, and we want to insulate both. There is less heat transfer through the insulation if there is less temp difference. If the warmer of the two fluids is on the outside there is less temp difference. The vapor is being cooled by the liquid, therefore the liquid is the colder fluid and the vapor is the warmer fluid.
Yep.Quote:
I use a 1/2" od corrigated pipe in a 3/4 inch id pipe 12 inches long with the inside of the 1/2" pipe as the evaporator
I would call that an improvement.Quote:
This system using 2x 1/7 hp compressors will run my evap at -125 unloaded and -104C loaded at idle and -92C benching. (GPU 9800 XT 2.2Vcore) Orginally I had it the otherway around and the tempos were -87C unloaded and -85 loaded and benching.
I like the corrugated pipe idea. That's excellent. :D