Fan Failure, does it ever happen?
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
Fan Failure, does it ever happen?
I see a lot of products that say they will alert you of a fan failure. I was just wondering if anybody has ever heard of this happening?
Re: Fan Failure, does it ever happen?
There is no reason it can't happen. If you only have a few fans it's going to be a pretty rare event, but consider those big computing centers with a great number of machines. They need to replace failed fans pretty regularly. That's why rackmount servers usually let you hotswap fans.halfpower wrote:I was just wondering if anybody has ever heard of this happening?
If we assume 60 000 hours as a mean time between failure for a quality fan and you have lets say 12 fans in all. Then the mean time before one of those fans fails is 60 000/12 = 5000 hours.
5000 hours is around 208.3 days. So in average you have more than one fan fail on you per year if you run those fans 24/7.
A lot depends on the fans you are using. Some are more durable than others.
-
- Posts: 73
- Joined: Wed Jan 12, 2005 1:07 am
- Location: Melbourne, Australia
- Contact:
I had my cpu fan die on me once. It was an Everflow 80mm with sleeve bearings that came OEM on a cheap aluminum heatsink. It caused my computer (Athlon XP 1600 at the time) to freeze about 4 minutes after boot, then freeze quicker and quicker with two reboots until I realized something wasn't right.
No damage to the chip or anything, just an inconvenience. Also, that was a Palomino core, and I think I've heard that they've made big advancements since then to the point where you could rip the whole heatsink off mid game and the cpu will be able to stop in time to prevent damage (read that somewhere on THG a while ago).
No damage to the chip or anything, just an inconvenience. Also, that was a Palomino core, and I think I've heard that they've made big advancements since then to the point where you could rip the whole heatsink off mid game and the cpu will be able to stop in time to prevent damage (read that somewhere on THG a while ago).
-
- Moderator
- Posts: 7681
- Joined: Mon Dec 09, 2002 7:11 pm
- Location: Maynard, MA, Eaarth
- Contact:
Hello:
Like most things in this world, there are qualitative things that probably have a much greater influence on the lifespan of a fan! And it is much more complicated than just stating a simple rule like "all ball bearings will last longer than any sleeve bearing"!
For sleeve bearings, some of the qualitative things that I can think of are: the quality of the material (which has "soaked-in" lubricants), the surface area of the sleeves, the tolerances of the fit of the shaft, the hardness and smoothness of the shaft and the sleeve, the overall radial balance and amount of thrust forces and "end float".
For ball bearing, some of the qualitative things that I can think of are: surface area of the races, hardness of the races and the balls, the number of balls used, the tolerances of fit and consistency of all the pieces, the quality of the lubercants used, the overall radial balance and amount of thrust forces and "end float".
But, lets' not over-analize things: the only thing that really matters is whether the fan lasts long enough; indeed, whether it stays quiet enough over a long enough time period to "work" for you. The other thing to keep in mind: most fans get noisier as they wear, so it is ususally quite easy to tell when they are failing!
Another point to make here: if you have a very efficient heatsink, and you are able to run a fan at lower speeds than you otherwise would -- this would tend to make a fan last longer. This is generally where we end up, in our pursuit of quiet computers!
Like most things in this world, there are qualitative things that probably have a much greater influence on the lifespan of a fan! And it is much more complicated than just stating a simple rule like "all ball bearings will last longer than any sleeve bearing"!
For sleeve bearings, some of the qualitative things that I can think of are: the quality of the material (which has "soaked-in" lubricants), the surface area of the sleeves, the tolerances of the fit of the shaft, the hardness and smoothness of the shaft and the sleeve, the overall radial balance and amount of thrust forces and "end float".
For ball bearing, some of the qualitative things that I can think of are: surface area of the races, hardness of the races and the balls, the number of balls used, the tolerances of fit and consistency of all the pieces, the quality of the lubercants used, the overall radial balance and amount of thrust forces and "end float".
But, lets' not over-analize things: the only thing that really matters is whether the fan lasts long enough; indeed, whether it stays quiet enough over a long enough time period to "work" for you. The other thing to keep in mind: most fans get noisier as they wear, so it is ususally quite easy to tell when they are failing!
Another point to make here: if you have a very efficient heatsink, and you are able to run a fan at lower speeds than you otherwise would -- this would tend to make a fan last longer. This is generally where we end up, in our pursuit of quiet computers!
From working in PC repair for quite a few years now I can attest to the fact that fans fail, fairly commonly. It's one of the most common things we see. One thing to note is that the failure rate for fans <80mm seems exponentially higher than the failure rate for fans >80mm. I guess from spinning faster.
Cos
Cos
[quote="NeilBlanchard"]Another point to make here: if you have a very efficient heatsink, and you are able to run a fan at lower speeds than you otherwise would -- this would tend to make a fan last longer. This is generally where we end up, in our pursuit of quiet computers! [/qote]
Some one pointed out to me in another thread that many fans won't spin-up at 5V. I wouldn't have trouble believing that undervolted fans are less reliable when it comes to spinning up.
Some one pointed out to me in another thread that many fans won't spin-up at 5V. I wouldn't have trouble believing that undervolted fans are less reliable when it comes to spinning up.
Physics says that to move a stationary object requires greater power than to keep a moving object mobile. So a fan requires more energy to start spinning but not as much to continue running. Same thing for harddisks. They draw more energy at spinup (nearly twice as much) than when doing normal reading/writing. This behaivor has nothing to do with reliability (in the longevity meaning).
"Starting up reliably" is something different. That refers to whether the provided voltage gives the fan enough energy to overcome its inertia. Depending on how the bearings ended up when the fan stopped, temperature, humidity & way too many variables, a fan rated to start at 5V MAY or MAYNOT startup. That's a problem of the manufacturer stating a too low (read borderline) spinup voltage. More established manufacturers however state conservative figures. For example, Globe rates their fans for 6-12V operation. Experience says they can start at lower than 5V reliably. Again, this is a case of specs vs what you run it at, not a longevity issue.
As for fans failing on me, nearly all noname fans I got failed within the year. None of the brands (Globe, Yate Loon, Delta, etc) failed yet. After many many years. Example: I just changed out a 80mm that had become too noisy in a PSU. It turned out to be a sleeve bearing Globe fan. The PSU was 7 years old and in nearly continuous use. I'd say that's a pretty long life, way more than spec'd.
"Starting up reliably" is something different. That refers to whether the provided voltage gives the fan enough energy to overcome its inertia. Depending on how the bearings ended up when the fan stopped, temperature, humidity & way too many variables, a fan rated to start at 5V MAY or MAYNOT startup. That's a problem of the manufacturer stating a too low (read borderline) spinup voltage. More established manufacturers however state conservative figures. For example, Globe rates their fans for 6-12V operation. Experience says they can start at lower than 5V reliably. Again, this is a case of specs vs what you run it at, not a longevity issue.
As for fans failing on me, nearly all noname fans I got failed within the year. None of the brands (Globe, Yate Loon, Delta, etc) failed yet. After many many years. Example: I just changed out a 80mm that had become too noisy in a PSU. It turned out to be a sleeve bearing Globe fan. The PSU was 7 years old and in nearly continuous use. I'd say that's a pretty long life, way more than spec'd.
I saw that too back when I used to frequent THG (imho they're down the tubes now, techreport serves as my main hub now, along with all the others, of course). If I remember correctly, Intel gave an error and temporarly froze when the HS was removed during a Quake session and then once the HS was put back on, the game continued to run. The AMD, however, literally cooked! A nice plumb of smoke! I'll see if I can find the vids...phuntism wrote:No damage to the chip or anything, just an inconvenience. Also, that was a Palomino core, and I think I've heard that they've made big advancements since then to the point where you could rip the whole heatsink off mid game and the cpu will be able to stop in time to prevent damage (read that somewhere on THG a while ago).
Hot Spot: How Modern Processors Cope With Heat Emergencies
Direct link to vids: link
FYI you may have to "register" to download, but bogus info works just fine.
DrCR
_________
DrCR, don't show halfpower that, he'll freak out.
With modern CPU's (ie, anything made in the past 3 years) nothing bad happens if the fan dies. You can lose any, or all, of the fans in the system, and the only result will be a shutdown from overheating. And that's only if you weren't paying attention to the weeks of increased fan noise as the bearings died, and the hours of rising temps.
With modern CPU's (ie, anything made in the past 3 years) nothing bad happens if the fan dies. You can lose any, or all, of the fans in the system, and the only result will be a shutdown from overheating. And that's only if you weren't paying attention to the weeks of increased fan noise as the bearings died, and the hours of rising temps.
I'm not a mathematician or a statistician, but I'm inclined to question the analysis of kesv in the first reply of the thread that the MTBF of the fans would be divided by the number of fans to get the frequency of failure.
If that were the case then one out of every 45 hard drives (with a MTBF of 400,000 hours) would fail each year, even if new. Since the MTBF is 45 years, half of the 45, actually (thus, 23) would have failed after 45 years. I suspect also from common-sense knowledge that the failure rate would be low in the early years and increase to the highest level as the 45 is approached.
If that were the case then one out of every 45 hard drives (with a MTBF of 400,000 hours) would fail each year, even if new. Since the MTBF is 45 years, half of the 45, actually (thus, 23) would have failed after 45 years. I suspect also from common-sense knowledge that the failure rate would be low in the early years and increase to the highest level as the 45 is approached.
I had once mounted a case intake fan that, over the course of one month, fed enough dust into the < 1 year old VGA fan to kill it.
Also, slightly off topic, I recently had the MB case fan header die after ~2 weeks of use. I caught it quickly as I had been monitoring temps and fans closely, having just rebuilt the system. But, the fan was off long enough to let the Smart Drive enclosed HD reach 55C. A week later a diagnostic program told me that the HD was going bad. With 6 months on the warrenty, Maxtor RMA'd it. Since Maxtor had the diagnostic code, I can't be certain that heat the sole cause of the HD failure.
edit - changed VGA fan from >1yr to <1yr
Also, slightly off topic, I recently had the MB case fan header die after ~2 weeks of use. I caught it quickly as I had been monitoring temps and fans closely, having just rebuilt the system. But, the fan was off long enough to let the Smart Drive enclosed HD reach 55C. A week later a diagnostic program told me that the HD was going bad. With 6 months on the warrenty, Maxtor RMA'd it. Since Maxtor had the diagnostic code, I can't be certain that heat the sole cause of the HD failure.
edit - changed VGA fan from >1yr to <1yr