Overheating CPU slowed down system

wayner · Post by **wayner** » Tue Apr 13, 2004 7:22 pm

This system is a Dell Dimension 8200 bought in Dec of 2001 with a 1.7 GHz CPU. There is no CPU fan but there is ducting from the 92mm case fan to the CPU heatsink.

A year or so ago I thought the fan was getting too noisy so I wanted to replace the case fan. I bought a Zalman 92mm fan with a silencer which is just a resistor which I installed.

When rebooting the PC I would usually get a notification about a previous fan failure. I ignored these messages as I assumed the motherboard didn't like the resistor.

Recently I noticed that the system was very slow. When checking out Task Manager I noticed that the "System" process was taking all of the CPU. I tried booting into safe mode changing drivers, etc. - the problem was that the system was overheating and, I assume, that the OS was slowing it down to cool things off.

I opened up the system and noticed that the resistor was extremely hot to the touch as was the CPU heatsink. The shielding of the resistor looked as if it was getting melted or worn away. This connector may have been getting pinched between the case and the 5.25" drive cage - it wasn't a problem when the case was open but the pinching would occur when closing the case.

Now that I have removed the resistor the PC runs fine. The only problem is that the Zalman fan is too noisy without the silencer so I am back to the original fan.

Why would the resistor be so hot? Is it possible there was a short? Has anyone heard of this before?

Is there any way to monitor the temp of this system? MBM doesn't give me a CPU temp.

One piece of advice - if you see a PC running very slowly with the System process using all of the CPU check out the temp of the CPU.

Bluefront · Post by **Bluefront** » Wed Apr 14, 2004 5:42 am

Well you learned an almost expensive lesson. When you modify airflow, fan size, fan speed.....anything that has a potential heat effect, you have to measure the resulting temperature change. Or suffer expensive damage.

You can buy cheap digital temperature meters with small remote probes. RadioShack, Target stores....lots of places have them.Place the probe as near to the CPU as possible. The under-side of the CPU heatsink close to the CPU socket usually works well.

silvervarg · Post by **silvervarg** » Wed Apr 14, 2004 8:58 am

Many Dell machine buildt that way actually has a temp controlled fan.
Often the Panaflo 92mm L1BX is used.

So another lesson is check what you got before you replace it with something else. Make sure what you replace it with is something better.

Still it is odd that the resistor got hot, it should not have been that hot.
Perhaps the Zalman fan does not work great with the PWM pulses of the temp controlling. That could be the reason that it draws lots of current early in the pulses. That could be the cause of the resistor heating up.

The biggest problem with the Dell machines is usually not the fan or the temp controlling. It is the way the fan is mounted and the duct. This can enhance the vibration noise a lot.
A soft mount of the fan and some work with the duct could probably make a lot of difference.

wayner · Post by **wayner** » Wed Apr 14, 2004 9:33 am

silvervarg wrote:Many Dell machine buildt that way actually has a temp controlled fan.
Often the Panaflo 92mm L1BX is used.

So another lesson is check what you got before you replace it with something else. Make sure what you replace it with is something better.

The original fan was a Datech which is a relatively cheap fan. I believe this fan is very common with Dell PCs - the Dell PC I use at work also has a Datech fan.

Interesting theory about the resistor heating up - I think you might be right!

POLIST8 · Post by **POLIST8** » Wed Apr 14, 2004 9:55 am

You could get a matching Ohm Resistor that has better heat tolerance (see: fatter resistor).

esn · Post by **esn** » Wed Apr 14, 2004 11:29 am

POLIST8 wrote:You could get a matching Ohm Resistor that has better heat tolerance (see: fatter resistor).

Yes, make sure you are using at least a 1/2 watt resistor. I'm using that in one of my systems and it is cool to the touch. The other system uses doides (two 5.1V in parallel). This drops the voltage to 6.9V assuming a 12V supply. FYI, my fans are Panaflo 80mm L1A. FYI again, in the downloads section there is a fan resistor calculator. Check it out if you haven't already done so.

silvervarg · Post by **silvervarg** » Wed Apr 14, 2004 11:36 am

I believe that the typical Zalman resistor is a 2watt resistor, so replacing it with a 1/2 watt resistor is not a great idea.
A 2 watt resistor normally can take any computer fan without becoming hot, that is why the behaviour was so strange.

wayner · Post by **wayner** » Wed Apr 14, 2004 11:39 am

silvervarg wrote:I believe that the typical Zalman resistor is a 2watt resistor, so replacing it with a 1/2 watt resistor is not a great idea.
A 2 watt resistor normally can take any computer fan without becoming hot, that is why the behaviour was so strange.

I will check the resistor and post back when I get a chance. Now if I could just remember the colour scheme

:

Bad Boys Rape Our Young Girls But Violet Goes Willingly - That's it isn't it?

scalar · Post by **scalar** » Wed Apr 14, 2004 11:59 am

Thanks to the Intel Pentium IV heat monitoring system, at least you're not buying a brand-new system. Rather than smoke itself, the P4 will automatically throttle down to keep from baking to death.

But just how slow does the P4 get when the magic-smoke containtment takes effect?

I wonder how well a fanless P4 linux router box would work, intentionally designed to not cool the P4 properly.. routers don't need too much horsepower anyway...

-Scalar

shathal · Post by **shathal** » Sun Apr 18, 2004 3:11 am

Depends.

Do be aware that there *IS* a point at which the thermal throttling *WILL* shut down your PC. This is not something that the motherboard does, but circuitry on the CPU itself.

I personally like that.

The CPU only throttles so much, before it goes "OK, that's it - best shut down, this isn't working..." and does just that.

It is quite regretable (in my opinion) that AMD's solution in this regard (as best I know) is still reliant on the motherboard being able to do the same. Also, another problem with having to talk to the motherboard is that it's an additional delay. I hope that they'll move the entire stuff to the CPU, like Intel did.

Better to lose a few minutes/hours work, than losing a lot of bucks on a new CPU, right ?

In general, it's PRETTY hard to kill a P4 via overheating. It can be done (usually, though, that's due to OC'ing it and higher voltage), but this is "long-term" damage. Short-term "it melted right there..." is pretty hard to do. I am not aware of anyone having succeeded (which is a good thing).

Just mentioning this, in case it's been forgotten. Throttling WILL shut down the PC, at some point, if the temps keep increasing. Good CPU survival method, I think

.