Slashdot Mirror
Is Your Computer a Fire Hazard Waiting to Happen?
whoppers asks: "I'm sure we've all had our share of computer fans die, but what happens if your box is running while you're at work and several fans go out? My in-laws spare TigerPC AMD K62-400's power supply fan just went out about two hours ago, and the thing was blazing hot. A little poke to the blade, and it started up again, but shouldn't these things be made to stop if the fan stops for any reason? I'm starting to wonder if I should start leaving my box off when I'm away for a few hours. Since it's usually wide open, I don't see too much harm, but these cheap boxes that never get opened and cleaned have to be a hazard right? I can't afford a halon system in my office just yet. The only link I found related to this is here and should a few more people read this, here's the cached version. Does anyone have any thoughts or stories related to this?" The fact that this article appears on July 4th, when most Americans will be lighting fireworks is purely coincidental. That doesn't change the fact that the submittor raises a very good point.
A general rule of computers is: the older they get, the more dusty they are and dust bunnies and their denser cousins are highly flammable. Unless you can keep such machines clean, it' is probably safer to leave them off.
You have nothing to worry about. They test for things like the fan stopping.
Eventually, when the heat gets too high, the power supply will either shut off or destroy itself.
The UL/CSA logo (if its genuine -- many cheap power supplies don't put on genuine labels) "guarantees" the power supply won't be dangerous to you. A flaming power supply would be, obviously.
Hope that allys your fears!
If you could be told what you can see or read, then it follows that you could be told what to say or think - BoC
Fans are designed to move air - like any fluid motion, air will move along the path of least resistance. In the case of a case fan, where the case is left open, you're pulling air from a very small area right in front of the fan before exhausting it out the back. The heat generating components (CPUs, hard drives, video chips, etc) tend to be far enough away from these fans that you'll see almost 0 airflow over them.
I used to work for a major hard drive manufacturer, and would get complains from users who said our drives were running too hot. Quite often, they said "I even leave the case off, and it's still too hot!". Many times, just putting the case on solved their heat problems. By creating essentially a duct for the air to flow through, the fan was able to pull air from the front of the case, across the heat generating components, and then exhaust it out the back.
In the case of components with their own fans (CPUs, video chips), this is still important - while you've exhausted the hot air from around the component, without a properly functioning (read: case on) case cooling system, that hot air is never removed from the general area around the component, and just gets sucked back in on the intake side of the fan.
Just my $.02.
-- "Other than that, how was the play Mrs. Lincoln?"
My friend has an old IBM PS/1 that's a 486 with a Pentium Overdrive chip that he uses as a print server. While he's been away, I've had to fix the machine. It was scorching hot to the touch and I thought the power supply was going bad.
After considerable effort, I removed the power supply with the intention of replacing it with another AT one that I found. Unfortunately, the power supply had extra proprietary connectors and the replacement one didn't, so I was left to figure how to fix the original one.
I took a closer look and I saw nine (9) years (!) worth of dust clogging the power supply fan, thus blocking its motion. Ignoring printed warnings as "Caution! Shock Hazard" and "Warning: No User Serviceable parts inside", I carefully opened the power supply and removed the fan. Fortunately, the fan had a plaster connector for easy plugging/unplugging (as opposed to being soldered directly to the board).
So I removed the fan with easy and scraped (yes, SCRAPED!) the crap off of it and wiped it down. Then I applied WD-40 to the bearings to get the fan blade with more ease. I had to help the WD-40 spread by using a screwdriver to turn the fan both directions. Finally, after 20 minutes of effort, the fan blade was turning reasonably well with I'd tap it, so I put it back in, reconnected it, and reassembled the power supply.
Once the computer was put back together, I turned it on and felt for heat. Not alot. For the first time in a while, there was ventilation coming from the back of the power supply and the system was running much cooler than before.
So, let this be a lesson to you. Make sure you regularly (yearly?) clean your fans off, removing the dust before it cakes on. Make sure that you do this to any older PCs you have or are about to obtain.
Otherwise, your system will run dangerously hot and only bad things can come of that.
Which a great deal of them aren't. With most of the computers I've owned, if I took the side panel of the case off, the CPU ran a full 8-10 degrees C cooler than with the case fully closed. This is primarily because the poor case cooling meant that the air inside the case was 10 C or so hotter than room temperature, so removing the side panel let the CPU fan suck in cooler outside air to blow on the heatsink (since the CPU fan is at 90 degrees to the motherboard, it's good at sucking air directly from outside if the side panel of the case is off).
10 PRINT CHR$(205.5+RND(1)); : GOTO 10
> the new ASUS motherboads have COP : CPU Overheating Protection, which switches off the machine when temp goes baloony.
I have an ASUS board a bit over a year old. I do intensive number crunching on my home machines, with some jobs running over a week of continual 100% CPU time. Being AMDs, they tend to run kind of hot, too, so sometimes I put a room fan blowing along the wall behind the boxes' exhausts.
At any rate, one warm day I had the A/C set kind of high and the room fan aimed elsewhere, and one of the boxes overheated while I was out to lunch. But the board halted it for me. When I came home it was making a horrible alarm sound, and unfortunately I had to reboot because I couldn't figure out how to make it restart after the alarm, but at least I didn't get a fire, nor even any overheat damage to the CPU.
BTW, Linuxers/BSDers who have temperature sensors on their motherboards may want to run lm_sensors and a display such as gkrellm in order to keep an eye on your system temperatures when you are around.
Sheesh, evil *and* a jerk. -- Jade
Yeah, except you have to reach a temperature at which the plastic will create a self-sustaining fire for this to happen. What temperature do you think butane burns at? Ever tried to burn plastic with a lighter? Ever tried to burn paper with a naked CPU?
/their/ burning temperature. I think that the clear answer is "no."
The whole question hinged on wether a CPU or PS will reach a high enough temperature to raise nearby combustibles to
-Peter
I have a little story about air filters and overheating...
At work, one of the PM guys wanted to protect all the large, expensive DC motors on the manufacturing lines with an air filter on the blower motors. Well, next thing we know, there's filter fabric zip-tied to every blower opening on every motor.
The fabric didn't restrict the air flow too much. Until a week later, when the grease and dust in the air clogged them up. Then the problems really started showing up. If you look up the prices of DC motors in supply cataloges, you may notice the prices run up to $100,000 each for the large 500 horsepower models. It seems our desire to protect these babies created an intense smell of burned enamel. When you have about 300 of these motors laying around, many in obscure places, we learned its better to have dust caked up inside the motor than have an undersized filter trying to protect it.
So, the question is, are you going to change or clean this filter on a regular basis?
I've seen Watchdog cards used for this purpose. When the temp rises above X deg. it performs an orderly shutdown. I believe there are IBM desktops and servers that have this feature built in.
ARGH!! Another problem that can be solved with a component costing 10 cents - a thermal fuse. All you software engineers trying to read hardware interrupts from fan speeds and temperature sensors. What if the FAN_SENSE wire shorts with the PSU's AC output - even if the fan stops working it'll still look like it's giving 60rpm. Leave this to the electrical engineers now go back to Java or VB software peopl.
A caveman dreams of being us, the incalculable power and riches. We dream of being Q, then what?
You've never owned a 6400 or a 6500 then. The CPU/PS fans are so weak, the Mac tends to lock up on hotter days. If you look on google, you can find entire web sites devoted to publicising the problems with these machines. They were cheap, but IMHO not worth the trouble.
And the 6400/6500 case is anything but "easy to open". You need a special tool (a case splitter) -- otherwise you have to nearly rip the calluses off your fingertips (after your fingernails go first) straining to pull the front off. When you hear the sudden, loud CRACK sound, loud enough to suspect you broke something, then you've got the front off.
Like a lot of pre-Steve macs, the fasteners are all made of plastic or very weak metal, so unless you go carefully, you can easily break a Mac case where it will never close again properly. The PowerPC cases are great examples of the this. Everything, from the card guides for the NuBus cards to the tabs that hold the motherboard in place are made of very fragile plastic. If you're ever working inside a Mac made from about 94 to 99, be careful!
Macs have this ridiculously undeserved reputation for having great cases that are easy to work with and won't cut you. This is pure hype. The aluminum flashing inside a Mac case as RF shielding tends to get bent up as you open and close the case and is hard to straighten. I've been working inside some PowerMac and said to myself "Hey, what's that little maroon bead on the motherboard..." before I realised I was bleeding. On the upside, the shielding is sharp enough to make a clean cut.
SOME macs do have good cases, but most do not. Most require special tools (like torx wrenches) and a reasonable understanding of electronics, especially the Macs with built-in monitors (like not electrocuting yourself on the CRT). The recent macs once Jobs was back at the helm have much, much better cases than almost every mac that Apple ever made. If you have basic PC building skills, than a post-iMac era Apple machine should make you feel right at home. For all the earlier ones, read up on the specific model before you do anything. They're no worse to work on than cheap clone PCs, but they have their own unique gotchas that are sometimes so wierd, proprietary, and/or dumb, it's best just not to guess.
Apple kind of runs under this idea that the computer they are selling you is actually an _appliance_, which should run reliably as designed (not that it will). But if it doesn't, you're expected to "take it to the dealer" and have them fix it, like a car or television (which nobody does, because you'll get ripped off or have to bring it back multiple times). That's the only way they're managerie of case designs make any sense, is if a trained Apple Service tech is the only one whose ever supposed to open it.
I wouldn't buy any mac unless I already had the directions on how to open it up and work inside. That will give you an idea how many proprietary plastic parts you might have to buy from Apple for $10-$40 each if you muck around inside. If they're still available. That's what I advise people who ask me advice about what mac they should get, especially for people buying used macs.
Democracy. Whiskey. Sexy. Pick any two.
It's not just about your CPU failing.
If you have a serious PSU failure there is a good chance you'll have insulation breakdown on the 110V(or 240V in my case) mains section.If this breakdown happens *before* the fuse (which is generally located on the board) , you can have the situation where you can very easily reach temperatures high enough to start a fire in most things.
If you are not careful about your house wiring, or power your system(s) with cheap, thin conductor extension leads/power boards,you can go for some length of time (30 seconds or so), before the fuse blows or the breaker trips. Not good if your power supply has just turned into a 1000W bar heater.
You are in a twisty maze of processor lines, all alike.
There is a lot of hype here.
I've found that just laying my case down lowers the temp (only by 1-2degs to be honest). I figure this is due to the fact the CPU will be facing upward and so the heatsink is working more efficiently.
I've also found that adding a fan to the top of the case to pull the hot air out (on upright cases) makes a BIG difference, especially on smaller cases designed for the MATX boards. I had an athlon in a small case that was running at 60-65degs and very unstable until I cut a hole and put a fan in the top - it dropped the temp to 50-55degs and hasn't crashed since.
I think the biggest single difference I've made was adding a Coolermaster "Heatpipe" heatsink to an Athlon - that thing knocked 15 degrees off the CPU temp immediately! I had to remove it again though, as the noise was unbearable... It makes a nice-looking paperweight though!
Also little things like tidying up a rats-nest of wiring and putting dummy plates over unused expansion slot cut-outs in the case helps.
Code, Hardware, stuff like that.