Welcome to WebmasterWorld Guest from 22.214.171.124
Forum Moderators: phranque
I want to get the lot out off the dining room floor and into the attic.
I know it gets damn hot up there. I went up last night and measured it at 30 degrees C.
It has been really hot in the UK these last couple of days, but it's a regular occurance at least once a year. It's nice and cool in winter. Doesn't get damp.
Am I likely to run into problems with the PC's up there? There is air circulation, but generally hot air doing the circulating!
Anyone here run PC's at extreme temperatures and got away with it?
I don't want to go to the expense of air con, although I might look into watercooling, if I could somehow just plumb it up to the mains water supply rather than using pumps....
Did you try those drive cooling devices?
The only systems that experienced the problem were the heavily-used DB servers ... so disk access was definitely a factor. Still, the heat seemed to be as well.
It's been blocked off with a cap, but I'm sure could be made serviceable again. It's about 15cm in diameter.
My advice is don't put your computers in the attic.
Assuming I don't have a choice, and I need to lower the temperature of the location rather than change the location, are there any other options?
MTBF = Maximum Time Before Failure or something similar?
I saw some AC's on the net for as low as $339.
I already own two of those ;-)
Yes, they're great for the money, and I've thought about getting another one for the attic, but the consumer household ones are really not designed to be run 24/7.
raywood is correct. I use to run ALT, (accelerated life test) on ICís for Motorola. Itís not the microP you have to worry about. Itís the power supply.
I had a fan go out in a computer and the thing over heated. It wasnít the Ďcomputerí that over heated, it was the power supply. I know this because I have a fan card to help in the main cavity of the box stay cool. I replaced the fan and the system started working again, but, if I let the temp go over about 73F, the power supply starts to fail. I can hear the fanís speed changing. I believe it is due to a cap leaking because of the heat stressing. I never had that computer running at a higher than ambient temp of 75F. When the PS starts to fail, it causes the hard drives to create bad blocks. After 32 hours of running scandisk, I was able to correct the bad blocks, but had to find all the DLLís et. al. that were damaged and replace them.
As a matter of fact, I didnít know the power supplyís fan had gone bad. I started getting bad blocks on drive C and research on the net said that a bad power supply could cause such a problem. I then looked to see if the fan was on, and it wasnít.
MTBF = Mean Time Between Failures
>>How about if I have the PC's in the attic and the Power Supply outside the attic (much cooler)?
There is resistance in wire. Not much, but some. The longer the wire the higher the resistance. You get to a point where it is like putting a resistor inline with the B+. How much that resistance is depends on the quality and length of the wire. Remember that the power supply was not designed for a longer wire than they come with. You would need to do some tests to measure the current limit with various lengths of wire. When resistance goes up, voltage goes down and current goes up. Ohms Law.
And without the power supply's fan, the air flow of the box will be changed and that could have an affect on the other components in the box.
[Edit]Longer wire could also result in power supply component damage from running for an extended length of time at higher current levels.[/Edit]
[edited by: jim_w at 2:27 pm (utc) on Aug. 4, 2004]
Eventually discovered the ideal way ..put them in an insulated box ( an old fridge works ok and has the shelves to make a "rack system" ) and force cold air from a reasonable sized air conditioner through the box ...pick one up for around $1,000.oo ( even less if you get one that got dropped or has damaged case ..which wont matter to you ) ...run it from an in box thermostat control ...
Bit Heath Robinson but efficient ..
(the frig thing)
If it works and it is cost effective, then it is a great idea. I would at least put some kind of art work on the old frig ;-)) No magnets though!
A bit off topic, butÖ
Power supplies have always been a problem with computers. Apple and IBM both had power supply problems with over heating at one time or another. And some brain child at CBM, (Commodore Business Machines), decided that they could save money with the C64 by cutting the specification of all the power supply components in their power supplies by 10%. It would save them thousands. So on the C64, they redesigned the PS and in a moment of true genius, they took the air holes out of the power supply pack. The net result was every one, without exception, of their PSs overheated and that created a need for 3rd party power supplies that were Ĺ the price of getting another design defect from CBM.
There was a computer store in FTL, and remember this was about 1983 or 85, that went out of business because they purchased their C64's from CBM when they first came out and K-Mart was selling them for less than this company purchased them for. And that happen to more than one business.
If you purchased one before they were in K-Marts, then you didn't get one. It was such a joke. We were using PETS in the VAQ department. Jim Butterfield, whom I have met, even knew of the problem.
You 'guys' never had a power supply that was totally sealed? No vents in the power supply? That's how you could tell.
Of course we won't mention the 1541 K-Mart swaps ;-)) I actually had 4 1541ís daisy chained at once. Of course you had to first turn on the computer, and then each one of the 1541ís as I recall. If you didnít, you would lock up the buss.
I wrote a talking terminal for it using SAM. Anyone ever use a talking terminal on CIS?
varya;bcolflesh - They were off the market a long time before you purchased your C64's[/edit]
put them in an insulated box ( an old fridge works ok and has the shelves to make a "rack system" ) and force cold air from a reasonable sized air conditioner through the box ...pick one up for around $1,000.oo ( even less if you get one that got dropped or has damaged case ..which wont matter to you ) ...run it from an in box thermostat control ...
Why not just get an old fridge, drill a couple of holes in it for LAN / Power wires (re-insulate the holes w/ plumbers putty or something) and let the fridge run on it's warmest setting? Then, you would have nice cool computers. Turn off the freezer and put your monitor, keyboard, KVM switch, and mouse in there! Haha, that would be a pretty cool (no pun intended!) solution!
I'd be concerned about possible moisture problems with the fridge concept.
If I'm not mistaken, most fridges have drip trays underneath of them to allow moisture taken from inside to evaporate. Also, I think most moisture inside a fridge is a result of the defrost cycle of the freezer. With that off, there shouldn't be any inside at all. I think...
Coming from a background similar to raywood's, and having some experience with accelerated life testing as well, I gotta say this is just a bad plan; Semiconductors need cool air, and electrolytic capacitors will dry out and fail in extreme heat. 50 degrees C is the maximum allowable temperature for commercial-grade ICs, and that means case temperature. Obviously, you can't maintain an IC's case temperature at 50C with any practical amount of forced air which is also at 50C. I'll second the warning on disk drives, too.
If I had to put my machines it the attic, I would build an insulated enclosure in the attic, and then run ducting to 'steal' air-conditioned air from the coolest area of the house below, through the enclosure, and return it to the house or into the air-conditioning system return. This must be forced-air, and the fan must run constantly. Actually, I'd use two fans with alarms to be sure the machines couldn't fry due to to a single-point failure.
re: "B+". Man, it's been a long time since I used that term! :) </aside>
I guarantee that at 95F, the junction temperature of the ICís will well exceed 150C. I know because I clocked a bunch of 68HC11ís once in ALT, the procedure called for them to be biased only, they got so hot the lids came right off the chip carriers. It melted the epoxy. I had never seen that before then.
We may be getting to the point where it would be cheaper to just put a small sub-room in the attic and just put an AC unit in the window. How much could it take to put up 3 walls and a door?
Not a lot. I could knock that up in a weekend. No window though - although I have the ex-bathroom roof vent.
What does it cost to run an air-con in the UK 24/7?
It's about 1400 watts as I recall.....
And yet, might the solution be as simple as thinking like an over-clocker and using a system that recirculates chilled water in tubes inside the box? I once saw an episode of The Screen Savers on G4/TechTV where the entire computer was water-cooled. It also included some software that monitored the temperature in the box and could send various types of notices (e-mail, pager, SMS, etc.) when certain temperature levels rose too high. If this sounds feasible I'll be happy to do some research to find the appropriate links for you?