Hi All,
A "funny" thing happened here this morning with my desktop computer.
I was having a bit of a play around with the speed controller program for the CPU fan, that came with my mother-board.
I'm not sure what I did, but for some reason, the CPU fan stalled, meanwhile the computers buzzer was letting off a raucous beeping sound, the screen froze and I ended up rebooting.
Once I restarted, it returned to normal, fan was running at it's usual speed.

I won't be mucking around with that program again, even though I hadn't changed the temperature or fan speed settings that much, at least not enough to bring on a thing like that.

What I did find later on, on the Net was a program called SpeedFan, it monitors all you fans speeds, system voltages and temperatures and it's free.
It can be found HERE.

You can buy it too, which means you can do a few other things apart from what the free version does already, but the free version is all I'll ever need.

What I did notice was in the voltages readout, is my +12V is down as far as 8.58V and the -12V is up at -16.97V, is this normal for the voltages to that far out of wack?

The plus 12 being that low would concern me. It is what powers fans and media drives. As far as I know just about the only thing they do with minus 12 these days is drive the serial ports and that can be anything from 3 to 30 although some devices do not honor that old standard. My weather station will not talk to a port that is much under +/-12v

I just configure the fan speed in the BIOS using the dynamic temperature ranges. Then without any software or OS intervention, the fan automatically adjusts its speed to the CPU needs. It goes full speed at power on and reset/reboot, and slows down once things are running. One mainboard maker I use (ASRock) slows the fan down gradually. The other (Tyan) slows it down quickly. Both then make speed adjustments as needed to keep the CPU temperature within the specified range.

I have not checked fan voltages on my computers. It is my understanding they use pulse width modulation to control the speed. I don't know what the chopping frequency is. But maybe that chopping is affecting the voltmeter reading.

Might not be a bad idea to change out the PSU, they're not too expensive. Just in case the weird voltages are a symptom of bigger (yet undiscovered) problems.

The +12 does not seem to be that well regulated but I always see something in the high 11s.

Are those voltage readings actual measurements taken with a DMM, or just the numbers that the program reports?

I have tried a couple of those "system monitor" programs, including one that Intel included with their own motherboard, and ALL of them had serious "issues" with accurate voltage measurements. Not sure what the measurement system uses as a voltage reference, either.

Measure the supply outputs yourself with a good DMM before swapping out the PSU solely on the results of some piece of software. While you are at it, check each DC supply rail with the meter set for AC, to look for excessive ripple. The first thing to go in those power supplies is the filter capacitors drying up...

If that software wasn't written for your specific motherboard, any readings you get from it are probably wrong. It's often hit-or-miss, with some functions working, but others not. For example, I had to try 3 different programs before I found one that let me monitor my motherboard's temperatures. (It lets me monitor all the voltages, fan speeds, etc, too.)

I've had good luck with the "Everest" info tool... Gives you tons of info on hardware, including voltages, fan speeds and temperatures if your motherboard is so equipped. The only computers so far that I haven't been able to get fan speeds/temps from are Dells.

It's primarily a hardware profile tool, but also gives lots of info about the OS.

Just as a question for some of our computer guru's here, what sort of temperature range should you expect from a CPU?
I still have SpeedFan installed here and I think it read as high as 55C (122F), it's as low as 22C (71F) when I start the computer up.

Of course, I guess it depends on the type of CPU and whether the system is over-clocked or not, but what sort of maximum temperature should someone start worrying at?

Thanks,
Mike.

It depends a lot on the CPU; for example, my Athlon64 X2 has a maximum design temperature of 65C while the Core2 Duo in my laptop is designed for 105C. Most chips these days have built-in thermal protection that will scale down operations if you exceed design temperature to prevent sudden death. There are many factors that determine whether said temperature is "ok", but if you're running close to or at the design temp CPU life is probably going to be shortened significantly.

If you have an Intel processor, thermal spec is easy to find:
http://processorfinder.intel.com/

Note that laptops can be pretty tricky; just because the CPU is ok doesn't mean the heat will play nice with other components. I had a Pentium 4 laptop a while ago that pretty much melted down... All was fine when you started it up cool, but as it warmed up things like USB ports close to the CPU would stop working. The cooling system was sufficient for the CPU itself, but it didn't efficiently evacuate the heat from the laptop--it let too much spread around to other components.

Trumpy, it depends how heavily you're overclocked (or underclocked, as the case may be.) Depends a lot on the processor, but 55C is right in the normal temp range.

If you're overclocked, the only temp you have to worry about is at what point your system starts locking up. If that happens, either lower the clock speed, raise the voltage, or put on a bigger fan! Underclocked computers (EG, any computer that's not yet been overclocked) can tolerate quite high temps before locking up.

I never really understood overclocking. Usually CPU speed is not your bottleneck unless you are playing games or cracking codes. Adding RAM is usually the best way to speed up a windoze machine, simply because the software is so bloated.
Easier than that is to go in and stop all those unnecessary processes that get added on, whether you want them or not. I usually try to keep the active processes down to one page.
It seems hard to do because things like Quick Time always seem to want to stay resident if you ever start them and come back when you boot. I go through the registry fairly regularly and delete almost everything in the "run" entries.

I've never heard any valid reason for overclocking, other than people looking to ruin a perfectly good computer or "just because I can". Even overclocking a little is going to increase the failure rate exponentially, and yield very little results. Errors abound on overclocked hardware.

I agree, If that chip would have tested OK at a higher clock rate, they would have sold it at the higher clock rate.

Who said overclocking was limited to CPUs?

I overclock because of economics. I paid for a 2.4GHz P4, and overclocked it to 3.2GHz. I saved about $150 over what I'd have paid for a 3.2GHz chip. My RAM is overclocked, as is my front-side-bus, so I really did see about 30% increase in performance.

Whether the CPU, RAM, video card, north-bridge, or other components is your bottleneck depends greatly on the types of programs you're running. If you're just running MS Word? Yeah, don't overclock. If you're into gaming, though, crank that sucker for every Hz you can get out of it!

This isn't the case at all. Most CPUs are intentionally under-clocked by the manufacturer for marketing, so they can set their price points where they like them.

All processors of a certain series come off the exact same Intel assembly line, often even cut from the the same wafer. Their marketing department tells them they need a certain number of chips at different speeds, and they set price points for them. As chips come off the line, they're tested. They test for 3.2GHz until they fill that queue, then test for 3.0GHz, 2.80GHz, etc. Once they get all the 3.2GHz chips they need, they don't even test for them anymore. The next chip off the line might run at 3.8GHz, but it gets a 2.4GHz code laser-etched onto it and is tossed in a box. Eventually, if the quality of the line is good, enough fast chips are identified early and every subsequent chip off the line is sold at slower speeds and not even tested. It's luck of the draw whether you get a 2.4GHz chip that can run at 3.8GHz, or whether it was on the edge of the wafer and a few layers were sputtered too thin or whatnot and it actually IS a 2.4GHz chip. Buying a faster-marked chip does not help your chances, just garuntees it will be at leat that fast.

In my case, my RAM was my limit, not the 2.4GHz chip; I'm running at 3.2GHz at default voltage and I know could have pushed by CPU way higher than 3.2GHz, but it wasn't worth dropping the RAM and FSB speed for. My last chip was an Athlon that I unlocked the multiplier on by redrawing the laser-etched connections with a graphite pencil. I overclocked the crap out of that chip; got about 30% out of it. Only got about 5% out of my RAM, though. Eventually my watercooling rig started leaking; it never really cooled it all that great anyway so I ditched it for a nice passive phase-change heat-pipe when I built my P4 PC.

Naw, overclocking doesn't change the failure rates at all. It's the heat that hurts components, not clock speed- and even then, it's the capacitors that typically fail, not the microchips. Whether you overclock or not, you just have to keep everything cool.

The errors are mostly due to people pushing their computers to the limit, not due to component failure. If going back into bios and raising the voltage or dialing the clock back down fixes the lock-up, there wasn't any damage.

The problem with that theory is you assume they are all going to run at 3.2. What happens to a whole wafer that only runs at 1.8? These people are in business to make money and they will not provide any more quality than the market will pay for.
You may, in fact, be getting a chip that was never tested at the higher clock rate or you may be getting one that failed at the higher clock rate.
I guess the real question is whether that extra speed actually translates into any real time savings in your day. In any virtual memory machine, as soon as your total loaded applications exceeds your RAM you are running at the speed of your disk drive. Most windoze machines do that by the time they boot to the desk top unless you have a buttload of RAM.

A CPU getting up to 122F, gee I'd think that would trip an auto-shutdown, my recently failed desktop did that to me. Said it did an auto-shutdown because the CPU overheated. Although, that may have been a result from the failing capacitors.

According to the system BOIS temprature readout for the processor area the temprature stayed 90F and 95F. It was an approxament value but I didn't see it reach 122F.