Results 1 to 50 of 55
Thread: Cpu Overclocker
-
14-05-2003, 01:44 AM #1
Cpu Overclocker
Okay, I found a bed-partner in the cpu overclocker department. .... figuratively speaking, of course ....
It seems that I am the only taker on repairing "Don's" cpu chiller...
http://www.chip-con.com/index.php?pageid=15
A Danfoss NF9FX, R-134a, advertised evaporating temperature.... -45 (no load). D*mn... the thing runs in a vacuum!! (suction)
WE may be experiencing a refrigerant leak, it's only gotten on the bench. But the compressor will run a 13" vacuum. I've preliminarily eliminated compressor failure, but the system appeared to be "flat" at 2psig, idle.
As usual, Mr. overclocker is a refrigeration hacker....
"Don", said that his friends recharge with R-404a. Of course, I told him the implications of changing the application in regard to cap tube size, etc. I haven't checked the spec's on the compressor.
I'm just giving everyone a heads up! Because the "Fish" is going into the overclocker business !!!
-
14-05-2003, 10:47 AM #2
Re: Cpu Overclocker
Originally posted by herefishy
It seems that I am the only taker on repairing "Don's" cpu chiller...
A Danfoss NF9FX, R-134a, advertised evaporating temperature.... -45 (no load). D*mn... the thing runs in a vacuum!! (suction)
As usual, Mr. overclocker is a refrigeration hacker....
"Don", said that his friends recharge with R-404a. Of course, I told him the implications of changing the application in regard to cap tube size, etc. I haven't checked the spec's on the compressor.
But in reality it seems to work fine. Recharged Prommies run 24/7 without flaws, though it is necessary to run the fans at full speed.
The gain of switching from R134a to R404a is almost 10 °C at load. That definitely gives your 'bed partner' a hard one
-
14-05-2003, 06:05 PM #3I'm just giving everyone a heads up! Because the "Fish" is going into the overclocker business !!!
-
14-05-2003, 09:29 PM #4
Funny you should mention that, Gary. My customer indicated that that is where he hangs out.
I'll check it out.
-
15-05-2003, 11:16 AM #5Originally posted by Gary
I've been playing with the overclockers at phase-change.com
Funny you should mention that, Gary. My customer indicated that that is where he hangs out.
I'll check it out.
BTW: I was told that Chip-con charges their new Prometeia Mach-II with a special refrigerant mixture called A404. Herefishy, can you confirm this?
-
16-05-2003, 02:51 AM #6BTW: I was told that Chip-con charges their new Prometeia Mach-II with a special refrigerant mixture called A404. Herefishy, can you confirm this?
Well, I've got the Chip-Con critter down to 1000 microns vacuum, right now as I compose (I guess it'll take another hour to get it to 50 microns ).... And I'll let you know. The customer has relieved me af all liability to charge the unit with R-404A (to which I think you refer - A404?). I will monitor "heat sink" (evaporator) temperature, S.S.T., Superheat, and condensing temperature while charging via digital fluke interface . The system is so small, I decided that I would monitor liquid and discharge by temperature, as opposed to having the volume of my high side hose in the system, and the eminent "phphphphhhhht" when I disconnect, on what I perceive to be a critically charged system (of course IF I had in my possesion the loaned out schraeder depressor, I would not be so concerned ). I like the temperature thingy anyway.
I put a slightly oversized drier in the system (it's only a little spun copper drier) because considering the R-404A retro thingy... I thought it would help sub-cooling (I recall Marc O'Brien enlightening me to the effect of oversized driers in cap tube systems). Honestly, not that I am not capable of doing the research, I haven't done any evaluation of the heat transfer rates of the heat sink, nor the condening coil, or even enthalpy of the refrigerant, characteristics of the compressor.... hell, I'm just doing it!.....but I have taken to account some of the successes of the retrofit (so-to-speak) by those on this forumn, other sources, and the willingness of my customer
Actually I feel like a slut !!!! But I'm having fun
I'll report back.
-
16-05-2003, 05:48 AM #7
condenser ambient - entering (condenser and static evaporator)
R.H. = 65%
temp = 80F
Condenser air Exit temp = 92F
discharge (condenser entering) line temperature = 135F
liquid temperature (condenser outlet) line = 92F
The evaporator temperature, according to the electronic controller is below -50C?... I need to talk to Mr customer... I'm not entirely familiar with the controller. It only goes to -50 (whatever), and beeps at me annoyingly because the low temp is out of range. I'm hunting down my surface "k" thermocouple accordingly.... Oh I found it...... the controller must be in "F", becuase the evap temp with no load bottomed at -52.1F
I am concerned that the suction (compressor inlet) is near 40F... unsure.... NO LOAD.
suction is running about -8psig (-60degF?) the information I observed in hacker research was an improvement of 10C in evaporating temperature... sounds like I'm in line with the marginally successful?
compressor seems rather warm.....
Last edited by herefishy; 16-05-2003 at 05:52 AM.
-
16-05-2003, 07:55 AM #8
We need condenser outlet subcooling and compressor inlet superheat. (no load).
-
16-05-2003, 10:07 AM #9Originally posted by herefishy
(to which I think you refer - A404?).
Maybe a blend between R134a and R404a to reduce head?
I put a slightly oversized drier in the system (it's only a little spun copper drier) because considering the R-404A retro thingy...
These small systems are extremely critical in operation. Don't make the mistake of putting a 'normal' dryer with 1/4" ports in instead. It's inner volume is enough to screw up operation.
I am concerned that the suction (compressor inlet) is near 40F... unsure.... NO LOAD.
compressor seems rather warm.....
I would advise your customer to run his fans at full speed all the time. The European Prometeia uses the Danfoss NL11F, and the datasheet indicates that an airflow of 1.5m/s (euh, 4.5 linear feet/second) is required for R134a operation. A bit extra when running with R404a doesn't harm.
I am very interested in your experiences with this conversion since I am planning to run a chiller with R507, using the very same R134a compressor as used in the Prometeias.
And as Gary stressed many times, I am riding the line using R507 in that compressor. So every information is welcome.
-
16-05-2003, 10:27 AM #10I would advise your customer to run his fans at full speed all the time. The European Prometeia uses the Danfoss NL11F, and the datasheet indicates that an airflow of 1.5m/s (euh, 4.5 linear feet/second) is required for R134a operation. A bit extra when running with R404a doesn't harm.
Last edited by Gary; 16-05-2003 at 10:29 AM.
-
16-05-2003, 01:44 PM #11We need condenser outlet subcooling and compressor inlet superheat. (no load).
I was charging the unit up to a point that my suction started rising form <-50 up to about -43F. I removed a bit, and got the suction back down to -52.1F.
Okay, Gary.... I'll put an access into the high side.Last edited by herefishy; 16-05-2003 at 02:35 PM.
-
16-05-2003, 02:06 PM #12
Here's a less accurate alternative: Cut a 3 inch cardboard disk. Poke a hole through the center of it. push your thermometer through the hole and between the fins in the center of the condenser. Hold the cardboard against the fins to block the airflow. This will give you a rough estimate of condensing temperature.
-
16-05-2003, 02:33 PM #13
Too late....
I've got a pigtail welded in... and I'm drawing a vacuum, right now.
Gary.... I'm going to give you ALL you need to know. You won't know what to do with yourself.
-
16-05-2003, 03:15 PM #14
while awaiting a proper vacuum, I dug up the information on the NF9FX compressor:
http://www.danfoss.com/compressors/p...3_Cd43r722.pdf
It is rated R-134a @ LBP -20 to 0, MBP 0 to 45. It appears that the difference in operating range is related to condenser airflow. 3m/s required for LBP
I am aware of the common application of Copeland semi-hermetic low temp R-12 compressors in ice cream merchandiser cases. The manufacturers of the cases would employ the R-12 compressors with R-502 in order to obtain a -30 to -40F evaporating temperature. Copeland would say that they don't authorize it (and assume no liability), but it works.
Sounds like we're doin' about the same in this application, considering we're employing the higher pressure refrigerant at a -50 to -60F evaporating temperature, as opposed to the -20 R-134A application.
hmmmmmm.
-
16-05-2003, 04:12 PM #15
Okay Gary,
High side pressure = 215psig
discharge line temp (condenser inlet) = 132F
liquid line temp (condenser outlet) = 90F
ambient = 84F @ 60%R.H.
condenser air temp out = 92F
suction pressure = 8"Hg
evaporator temp -54F
suction line coming back to compressor <32F (The dern cap tube must be 3' long, and it is really wrapped around the suction line)
When my head is around 220psig, and my suction around 8"Hg, seems to be a point of optimum performance.
-
16-05-2003, 04:57 PM #16
Continue filling it until the superheat is 15-20F, and let's see what the temperatures say then.
-
16-05-2003, 05:09 PM #17
I went further up the suctionline (about halfway), and checked temperature, there. It was oscillating -30F to -15F to -30F.
I think since the thing is unloaded, we're getting a little floodback. there is a large loop-dee-loop in the suction which I consider to be acting as a bit of an accumulator. (At the compressor, I not getting this fluctuation. there's a lot of subcooling going on in the cap tube wrapped around the suction.
What do you think?
-
16-05-2003, 05:15 PM #18
I think just a little more refrigerant would bring the superheat down at the compressor inlet. If the subcooling is still low, then it needs more cap tube.
-
16-05-2003, 05:19 PM #19
With the system at minimum load, we want the superheat to be near floodback. Then at maximum load the superheat will increase, but the evaporator will still be fully flooded.
-
16-05-2003, 05:21 PM #20
You know, like DaBit was saying, that compressor casing is 155F. I was checking at about one inch form the compressor shell. wherever I peeled the insulation back is about where the frost line stopped.
Gee, I have pumped the thing up, but when I do, the evaporator temp rises, from the optimum obtained temp of this -55 to -56F range.
I hear you about the cap tube. And that makes sense because of the higher PD that we now have with 404. pumping it up to about 240 psi head, and 3"Hg suction, I can't get much more than 2F subcooling.
Actually, the condenser fan has been cycling via the controller. when the evap gets to near temp (unloads), the fan goes to slow speed. I wonder if keeping the fan on high (as someone suggesting) would improve the operation?
-
16-05-2003, 05:24 PM #21
With the subcooling low (4F), there is a liquid vapor mixture going through the cap tube. That's not sucooling going on in that cap tube. It's condensing going on. On second thought, it is flashing, since the pressure is dropping.
Last edited by Gary; 16-05-2003 at 05:30 PM.
-
16-05-2003, 05:34 PM #22
Definitely keep the fan on high.
-
16-05-2003, 05:38 PM #23You know, like DaBit was saying, that compressor casing is 155F. I was checking at about one inch form the compressor shell. wherever I peeled the insulation back is about where the frost line stopped.
And forget everything you ever knew about frost line. It doesn't accurately tell you anything.
-
16-05-2003, 06:18 PM #24Originally posted by Gary
And forget everything you ever knew about frost line. It doesn't accurately tell you anything.
As you said, there's nothing to know about a frost line, because it doesn't mean anything. I just mentioned it in an effort to illustrate my observation of potential effect of the compressor case in order... Actually I thought the observation regarding the "frost line" was so meaningless that I didn't consider anyone would think anything was meant by it.
-
16-05-2003, 06:23 PM #25
Sorry, I didn't mean it as an insult. It was more for the lurkers information. Those of you who already know this, please disregard.
-
16-05-2003, 06:32 PM #26
Well, the cuthtomer picked up his hacked up machine. I had it on the bench running and gauged up, and let him observe the -55F heat sink temperature. I wantted to make sure that he was satisfied with what he observed, and I wanted to pinch-off and weld the accesses up with the machine running so I wouldn't alter the refrigerant charge.
I came to the conclusion that at the 8"Hg suction, -55F heat sink temp that the evaporator was fully fed, becuse if I shorted the charge from there, the temp would rise, and if I added charge the temp would rise.
I just don't see how you can wrap five feet of 90F cap tube around a suction line serving a -54F evaporator, and expect to be near saturate at the compressor inlet.
I propose that if the cap tube were not wrapped around the suction, that all the figures would look much better. But like you said Gary, the gas is condensing in the cap tube and as such an extension of the condener. So for that matter, if you take the temperature at the outlet of the condensing coil, you're not really measuring the condenser outlet temperature.... because the condenser ends halfway down the cap tube.
One inch from the compressor, the suction line was at 45F, three inches from the compressor 32F. At 18" from the compressor I was at minus30F !!!! Go figure. It's about 3 feet of line from the compressor to the heat sink thingy, whidh is about 1-1/2" square contact heat exhange surface area.
-
16-05-2003, 06:54 PM #27
It's extremely difficult trying to refrigerate that little inch and a half CPU square, and DaBit tells me that the heat is actually generated in a quarter inch spot in the center of that square. The metal square is on there to distribute heat outward and thus make the refrigeration target bigger.
Last edited by Gary; 16-05-2003 at 07:05 PM.
-
16-05-2003, 06:58 PM #28I propose that if the cap tube were not wrapped around the suction, that all the figures would look much better. But like you said Gary, the gas is condensing in the cap tube and as such an extension of the condener. So for that matter, if you take the temperature at the outlet of the condensing coil, you're not really measuring the condenser outlet temperature.... because the condenser ends halfway down the cap tube.
-
16-05-2003, 07:12 PM #29
yea, the cap tube abuse is truly the problem. With another foot of cap tube, I suppose that the PD would be greater, resulting in decrease of evaporator pressure, and requiring more refrigerant to be in the system and permitting the condenser to emit liquids. Would you agree?
At -20F, R-134a, that compressor was rated at about 657btu. I wonder what capacity the thing had at the -45 R-134a, much less the -55 R-404a. Do I recall a 200w rating?
When I do it again, I may go the extra mile and splice a foot of cap tube into the existing..................... ...JUST KIDDING!!!!
... but perhaps provide a proper cap tube.
Gary, what do you think of my observation of the heat sink temperature in relation to the varying amount of charge?
-
16-05-2003, 07:26 PM #30yea, the cap tube abuse is truly the problem. With another foot of cap tube, I suppose that the PD would be greater, resulting in decrease of evaporator pressure, and requiring more refrigerant to be in the system and permitting the condenser to emit liquids. Would you agree?
Gary, what do you think of my observation of the heat sink temperature in relation to the varying amount of charge?
At 8"Hg, the SST is -62F, so even at -30F (18 inches from the compressor) the superheat is 32F, which is nowhere near flooding.Last edited by Gary; 16-05-2003 at 07:40 PM.
-
16-05-2003, 09:16 PM #31
Just a thought
We repaired a pipe freezing unit the other week for a local company (R407c) as it had lost it's gas charge. When we had found the leak and repaired, recharged and then operated the unit we were "enlightened" to find that it had a small sol valve feeding the head (or contact surface). The operation of this unit may suit what is required for "overclocking" as it is quite a small surface evaporator with the refrigerant supply controlled via a sol valve rather than the normal tev or cap tube.
Not being an expert in these machines, does anyone know the logic behind the control of liquid supply in such a small solid evaporator i.e. no air flow?
Frank
-
17-05-2003, 12:19 AM #32At 8"Hg, the SST is -62F, so even at -30F (18 inches from the compressor) the superheat is 32F, which is nowhere near flooding.
So now we see the error of the overclocker's ways.
In the scenario that I depicted regarding the r-12 to R-502 ice cream merchandiser application... of course a proper refrigerant control was applid.
As usual the overclocker customer said that he had an old compressor, and some other parts from his dad's A/C company, and wanted me to look at them to see what we could do with it. I told him, that instead of spending the time looking at his cr*p, why don't we look at the application he is interested in first, and then we'll see if anything he (or even I) has is appropriate for the desired application. I told him that starting with the equipment to see what you could do with it is bass-ackwards. I think he understood.
He wants me to spec some refrigeration to a 120w and 172w peltiers that he is utilizing, and mentioned his desire for a -30F cooling temperature.
it had a small sol valve feeding the head (or contact surface).
-
17-05-2003, 07:26 AM #33He wants me to spec some refrigeration to a 120w and 172w peltiers that he is utilizing, and mentioned his desire for a -30F cooling temperature.
Let's imagine that we build heat exchangers for those two pelts.
In his Prommy system, we add these to the liquid line with the 172w pelt at the outlet of the condenser followed by the 120w pelt.
We adjust the charge and the cap tube length until we have 15-20F superheat at the compressor inlet and 10-15F subcooling on the liquid line between the pelt coolers (with minimum load).
The air cooled condenser would de-superheat the vapor. The 172w pelt cooler would condense the refrigerant at a temperature far below ambient. The 120w pelt cooler would subcool the liquid.
This would bring the high side pressure well within the compressor's limits and achieve very low evaporator temperatures.
In effect, it would be a cascade system, with the pelts acting as the high stage. What do you think?Last edited by Gary; 17-05-2003 at 07:37 AM.
-
17-05-2003, 02:51 PM #34
It was a small shut off sol valve that was being controlled by an electronics board - opening and closing to feed small amounts of refrigerant to the head. Seems like the head must have some sort of temp sensor in there but as it was sealed up we couldn't really see.
This is the sort of thing I mean http://www.fluesystems.com/pipe_free...nfo/arctic.htm
Seems suitable for this overclocking use as the refrigerant appears to be metered to suit the load.
Frank
-
17-05-2003, 03:10 PM #35Seems suitable for this overclocking use as the refrigerant appears to be metered to suit the load.
-
19-05-2003, 11:26 PM #36
Hey Gary,
The customer brought his Prommy back, because his second-rate refrigeration guy can't weld
I don't know what that meterial is that the flex line to the heat sink is made of... I was using 45% in order not to overheat. I ended up using a fluxed bronze rod for satisfactory results.
...But since I've got it back, I've charged it the way you wish... at near saturation at the compressor. I've got about <-30F at the compressor.
I didn't bother with high side access (since I pinched-off and welded it before), but the condenser outlet is about 121F (the temp in my shop is 100F+), and checking an end-bend temp in the middle of the coil, I get about 117-118F, so I think we've got about 3-4F subcooling(out of the condenser).
-
19-05-2003, 11:56 PM #37
Let's hope his home is air conditioned. That high side pressure has to be way beyond the compressor's specs with the R404a in there, and that high an ambient temp. I wouldn't run it too much in your shop.
-
20-05-2003, 12:07 AM #38
This seems insane!
We've got it on the bench with the customers computer equipment, and we're actually going to load it up.
But Gary... remember what I said about the ice cream merchandisers... the compressor is rated R-134A @ -20, but we're operating at -40 :-/
-
20-05-2003, 10:34 AM #39
Sure, suction pressure is within manufacturers limits, but discharge is not.
Now I wonder whether the 55C/131F maximal condensing temp with R134a is based on maximum pressure, or maximum temperature? At such high condensing pressures and low evaporation pressures, discharge heat increases. Or the motor might just overheat since more torque is required from it.
I would try to keep condensing pressure as low as possible.
-
27-05-2003, 04:27 PM #40
My Overclocking customer called me this morning. He fired up his R-404A (warranty void) Prommy, attained a -35 internal tc, and made the "top ten" on his benchmark program at:
http://futuremark.com/community/halloffame/
"Donebalp" in user top ten rankings at bottom of page
He has commissioned me for a 150 to 200 watt chiller for his video card
When we had the Prommy on the bench in my 100degF shop, we settled for a -18degC internal temperature, when I was satisfied with a -10 to -20 suction return temperature. I think operating the machine in a temperature controlled climate helped performance.
Last edited by herefishy; 27-05-2003 at 04:31 PM.
-
27-05-2003, 09:19 PM #41
Very kewl.
Are you starting from scratch on the chiller or converting the prommie? What temp are you looking for?
-
27-05-2003, 10:04 PM #42
I've a Copeland JFC1-0025-IAA that has been sitting in it's new box, for years - in my shop. It's rated at around 300btuh @ -40F R-12 (100w+?). I think with some R-409A, it'll do just great. Donebalp was suggesting around 150W heat dissapation. I think with a 4 sq.-inch heat sink, splitting hairs at this capacity is not necessary. We're still shootin' for the -40 to -50 mark (C?, F?, who cares!!!!)
I've a little evaporator coil, which I think'll serve for an (oversized?) condenser coil.
I've got 100' of Supco BC-1 cap tube ordered (I think I'll need 47'). donebalp says he has something for a heat sink, we'll see what he's got, otherwise I'm prepared to fashion one.
P.S.
... Is that a new Grandbaby you're holdin'? Congratulations!Last edited by herefishy; 27-05-2003 at 10:08 PM.
-
27-05-2003, 10:27 PM #43Originally posted by DaBit
Sure, suction pressure is within manufacturers limits, but discharge is not.
Now I wonder whether the 55C/131F maximal condensing temp with R134a is based on maximum pressure, or maximum temperature? At such high condensing pressures and low evaporation pressures, discharge heat increases. Or the motor might just overheat since more torque is required from it.
I would try to keep condensing pressure as low as possible.
Does that make sense?.... PROFESSOR!!!!!!
-
27-05-2003, 10:46 PM #44
150 watt = 512 Btu
300 Btu = 88 watt
Is that a new Grandbaby you're holdin'? Congratulations!Last edited by Gary; 27-05-2003 at 11:03 PM.
-
28-05-2003, 05:08 AM #45I've got 100' of Supco BC-1 cap tube ordered (I think I'll need 47'). donebalp says he has something for a heat sink, we'll see what he's got, otherwise I'm prepared to fashion one.
-
28-05-2003, 09:32 AM #46Originally posted by herefishy
[B]DaBit, what you have to consider, is the density of the refrigerant. I'm am only speaking in terms of the concept, but if you have R-134A at a -20 S.S.T., and compare to R-404A at -50 S.S.T., what is the differences or similarities of the mass of the gasses at those conditions?
But about the volumes:
Specific volume gaseous R134a @ -30C/-22F: 0.224 m<sup>3</sup>/kg.
Specific volume gaseous R404a @ -49C/-56F: 0.217 m<sup>3</sup>/kg.
As you can see, the gases have approximately the same density at the same pressure.
Also, heat capacity of the gas and latent heat of evaporation differ only a bit between these two refrigerants.
In the end, mass flow under the same evaporating and condensing conditions, same superheat and subcool, same heat input is very similar for both R134a and R404a, with R404a having a slightly smaller mass flow according to CoolPack.
(I like to mention it again: CoolPack is definitely the king amongst free refrigeration software.)
-
28-05-2003, 02:12 PM #47A too high a condensing pressure might damage the valves, and a too high discharge temp might cause the oil to decompose.
Typically you would interpret a (malfunction) higher discharge pressure with with too great a suction pressure (high mass flow), or restriction of airflow or refrigeration circuit.
In the higher pressure refrigerant application, the condensing temperature and pressure are acceptable for the refrigerant's application. In my mind the gas density comparison of the two applications represents that the compressor valve porting is suitable in order to prevent both overloading and provide adequate capacity.
I would point out in particular the specifications of Copeland equipment hwereas a compressor is rated at say R-134A medium temperature and R-404A low temperature.
In the case of the ice cream merchandiser applications that I have illustrated, the low temp R-12 compressor is regularly employed in the (very) low temp R-502 (R-404A) application with dependability, even though Copeland doesn't advertise the R-404A application in their compressor data.
for instance (referencing some dated information because that is what is handy) a Copeland model KAG*-0100 is rated R-12 25degF to -5degF, R-502 -5degF to -40degF.
This is not unusual.
By heat sink, I assume you mean the water block that mounts on his video card?
But our "heat sink" (cooling block) is going to be a DX " evaporator. he merely brought to me a block of copper. But I will be able to utilize his material... I have a good idea, or two.
-
28-05-2003, 03:33 PM #48
Ahhhhh... then it is not a chiller after all.
-
29-05-2003, 03:49 PM #49
Donebalp has two buddies who are going to bring their Prommies to my shop to void their warranties .
-
29-05-2003, 04:47 PM #50
I just got an update from Donebalp.
Apparently the -38deg reading he was getting while maxxing the CPU out, wasn't accurate. The software that he was using wasn't capable of going much lower.
He loaded a different program capable of measuring temps much lower, and it indicated his internal CPU temperature was minus 48 degrees ! ... while running.
he has clocked his 3.0 GHz CPU at 4.1 GHz.