View Full Version : Temperature controller down to -120C needed for cascade
DaBit
10-09-2003, 03:14 AM
The regulars @ phase-change.com or Gathering of Tweakers already know me for a longer time, but I am new here. Hi guys :)
Well, what can I say. I am currently constructing a cascade to chill my processor. First version is targeted at -63C @ load, and is basically a learning system. Then it will be improved so it is capable of -100C.
Here is a single image of the finished mechanical construction:
http://www.icecoldcomputing.com/images/page_images/proj210k_8oct2003_front.jpg
The rest of the story can be found here:Story about this system (http://www.icecoldcomputing.com/text/show_page.php?id=68), or @ phase-change.com.
And now the question: to complete this system, I need a temp controller capable of reaching -120C. All it needs to do is switch on a solenoid at a set temperature, and switch it off at the set temperature plus a differential.
I know Simpson has a controller capable of doing so, but it's very expensive. There must be more. Can anyone recommend suitable controllers?
Marci
10-09-2003, 04:15 AM
Carel.... same guys that make the prommi's controller... IR32 range... with the correct probe, will monitor it fine... then just knock up yer relays etc as required to hook onto it. Just don't get the same one the promi uses (IR32Z) as people are having problems getting below -50 properly and accurately...
DaBit
10-09-2003, 05:17 AM
The Pt100 (IR32V1E) and thermocouple versions (IR32V2E) can reach -99C. If I cannot find a suitable controller which can do -120C, I could live with these.
The next problem will be to find out what they cost, and who sells the Pt100/thermocouple version. I live in The Netherlands, which may be a problem.
:toast: Welcome DaBit :toast:
Great to have you here.
Your simplest answer might be something like labview tbh (software) and measure the temp that way and use a second pc to contol it.
iirc a labview control card and connector box runs to about £200, and i can 'supply' you with the development application. Useful stuff we use it at work.
]JR[
DaBit
10-10-2003, 03:06 AM
Originally posted by Apoc
:toast: Welcome DaBit :toast:
Great to have you here.
Thanks man! :)
Originally posted by ]JR[
Your simplest answer might be something like labview tbh (software) and measure the temp that way and use a second pc to contol it.
I know Labview. But that solution is like shooting a bug with a cannon. All I need is a simple controller like the ones found on the Proms, and the one I am using in the high stage (Eliwell EWPC-901). But then I am better off trying to obtain a controller like the Simson Phoenix II P20 (http://www.simpsonelectric.com/pdf/webpdfd/Phoenix%20II%20P20.pdf), which is already overkill.
Maybe I am better off knotting a Pt100 RTD to a 16-bit ADC, and knotting that to an Atmel AVR microcontroller. Add spice in the form of some relays, buttons, LCD display and a serial link, and I have a nice controller. But that takes such a lot of time to develop and debug :mad:
Labview does make startup procedures easy too though ;)
But yes it is overkill just to control a cascade.
]JR[
DaBit
10-10-2003, 03:54 AM
On the other hand: a 3.5" VIA Epia single board computer with LabView and a small touch screen TFT would be *the* solution. Albeit a bit expensive I suppose.
I suppose there is no version of LabView that can run on a PowerPC Kahlua II processor running embedded linux?
There is pocket pc labview, but i it needs a windoze ce, and it runs compiled apps only but thats no problem. Just installed the vi runtimes.
Lemme look into it...
(of course running from a pocket pc needs a2d's to integrate with the serial port or irda port)
edit: one of these isnt too pricey (http://www.mini-itx.com/store/default.asp?c=2#epia) might use one for my mini cnc mill im building.
]JR[
http://sine.ni.com/apps/we/nioc.vp?cid=12219&lang=US
I assume we have it at work since we got labview 7 complete suite a few weeks back...
]JR[
DaBit
10-10-2003, 07:59 AM
Hmm, it might be worth the effort. I can probably get those VIA boards with 400MHz processor cheap. If I can also get a small touch-screen TFT cheap, it might be a very nice thing to integrate into the phase-change.
But damn, only the VIA board is already more expensive than my entire cascade...
i got some intel ca810e mobos and processors (333mhz celerons) if you want one cheap theyre 25cm x 25cm all on board jobbies
can send you one for £20 + postage (whatever that will cost)
]JR[
DaBit
10-13-2003, 01:39 AM
Hmm, the cost of such a controller rises quickly. The small TFT's are more expensive than I expected, and interfacing the temp sensors would require either custom electronics or expensive add-on cards.
If my search for sub -100C controllers fails, I might just get the Carel IR32V1E, which goes down to -99C using a platinum RTD.
petervandamned
10-13-2003, 02:32 AM
welcome dabit
:)
DaBit
10-13-2003, 02:45 AM
Thanks peter! :)
Gary Lloyd
10-13-2003, 03:14 AM
Maybe I'm missing something, but I just don't see any need to unload the low stage compressor, since you are going for lowest possible temp.
DaBit
10-13-2003, 03:36 AM
Originally posted by Gary Lloyd
Maybe I'm missing something, but I just don't see any need to unload the low stage compressor, since you are going for lowest possible temp.
I have a few reasons for doing so. The main reason is that with R23, R508b or R170 in the low stage the temp may become too low for the processor. I simply don't know where the limit is, so I would like to have the possibility to set a lower limit.
The second reason is that I expect the heat dumped into the interstage HX after startup of the low stage to be more than it can handle while maintaining sufficiently low temperatures. Thus, I just might need to modulate low stage capacity during startup of the low stage compressor.
(The other solution would be to use a larger TEV orifice to cope with the initial load, but I prefer to optimize for normal operating conditions, not for startup conditions)
These CPU chiller systems work differently from a regular low-temp fridge. We have a huge load during pulldown, and a small load during normal operation. A regular fridge always has a huge load since it is turned on when cooling is required.
This makes it necessary to either oversize components or take other measures during startup.
I am not sure yet if I need hot gas bypass. But when I need it, it's there.
BTW: the first test was unsuccesful. I probably have a blocked captube. I only don't know at which end. I might have brazed the captube shut accidentally. Or I might suffer from scale.
I know I should use nitrogen, and I am currently negotiating about a bottle of either nitrogen or argon.
The nitrogen/argon is not expensive, but the bottle is.....
vBulletin® v3.7.0, Copyright ©2000-2009, Jelsoft Enterprises Ltd.