Computer Mecho's- need advice

[rodderz]

Gday peoples

Lately I've been having a few dramas with the computer being a bit of a pig to start up....bit like an old car on a cold morning and a dodgy carby

Turn the PC on, sometimes while the hard drive cranks over the monitor wont switch on from standby. Usually it happens when it's cold for some reason, sometimes you have to reset it 4 or 5 times to get it to turn on.

When it does turn on, it freezes at random points..... either at the loadup screen (right after you turn the machine on), when the windows logo comes up, at login, or at the Windows XP desktop. Mouse freezes and you have to reset as cntr-alt-del wont work

Once it has fully loaded however, the internet is connected, there are no issues whatsoever. I'm thinking it could be the operating system not starting correctly or hardware has trouble feeding power or something continuously....

Hardware details:

Pentium(R) 4 CPU
2.40GHz
256MB RAM

Cheers

Something might be wrong with the RAM.

[Paris Hilton]

Yup, I would suggest corrupted RAM, however I was having problems similar to this (But not as severe) and it turned out that my CPU was not seated in the socket the whole way. How many Sticks of RAM do you have? If you only have one, then you might have to borrow some, however if you have two or more sticks, try turning the system on with only one of the sticks. If the system runs fine then you know it's one of the other sticks, however if they all do it then you're looking at a different problem. The fact that once it starts it starts fine also leads me to belive that it may be the PSU...

[rodderz]

Not sure on the sticks...it originally had less RAM (forgot the figure) and got it updated probably 2 years ago so he may of added some higher MB into the slots

I will take the cover off later and have a gander at the internals...could be just one of those things that wear out over time.

[Paris Hilton]

Might be a good idea to clean out all the dust that accumulates in there over the years while you're at it.

[Paxton]

Get a Can of Air from Dick Smith and spray the computer out. Spend a greater amount of time on the fans and CPU than anything else. Also, Re-seat the RAM, and then do a Defrag on the HDD. That should help speed the computer up.

-Andrew

[needfordspeed]

Originally Posted by paxtonandrew

Get a Can of Air from Dick Smith.....
-Andrew

What's a can of air worth these days? My computer is dusty as from the inside.

[M14A-Mclaren]

Originally Posted by paxtonandrew

Get a Can of Air from Dick Smith

You can get cans of air? and have to pay for it? thats funny as.

[Paris Hilton]

Not much, but I wouldn't bother...I just use a vacuum cleaner, but be careful not to hit anything with the nozzle. Also might want to be weary of static electricity when you're doing it

do you have hard drives in a raid configuration?

download memtest 86, you create a floppy with it that you boot off and it runs a diagnostic on the ram. http://www.memtest86.com

[rodderz]

How do you use that memtest...i have downloaded it but the instructions on the website and readme file are pretty vague

Originally Posted by rodderz

How do you use that memtest...i have downloaded it but the instructions on the website and readme file are pretty vague

Extract the contents of the zip file to a folder, insert a floppy that is blank or has nothing important on it then run install.bat and tell it the letter of your flopy drive. Once its done restart with the floppy in and it should boot into memtest, leave it for a few hours and see if it gives any errors.

[M14A-Mclaren]

I had to use that too, takes frickin ages, I used it on a CD because I dont have a floppy disk drive, cant remember what I did now sorry.

[rodderz]

Tried the memtest found no errors...i'll give the insides a clean tomorrow and see if theres any crap in there

[Paris Hilton]

It's the PSU then, lucky, they are nowhere near as expensive as RAM...

[RED_EL_XR8]

Originally Posted by Selekta

It's the PSU then, lucky, they are nowhere near as expensive as RAM...

What vague piece of assumed knowledge and trance music brings you to the this conclusion without, a single actual practical test. If you sprout off enough compenents to replace you may get it eventually.

[Sourbastard]

Originally Posted by Selekta

It's the PSU then, lucky, they are nowhere near as expensive as RAM...

Who do you work for? Dial a Pyschic?

There is alot to do before proclaiming the CPU is the issue and throwing it out.
Ive had the exact same issues occur due to USB printers being hooked up during bootup.

My advice, disconnect all peripherals first, printers, usb devices, scanners, anything, id even test out different mice and keyboards if you have access to them. Then see if the problem reoccurs on boot up. Chances are this is a compound problem rather then one single cause.

[Rodp]

Software memtesters aren't the be all end all, in fact I wouldn't trust it to find anything but an obvious problem. Once you're at the point of running memtest, your PC seems stable so if it was a mem issue, you'd likely get 3 long beeps at POST when it couldn't boot and your PC wouldn't be stable once it was up and running considering it's such a small amount of mem.

Unless you have a can of air, I wouldn't rummage around the innards of your PC either, I'd only even use the can of air in small doses to blow off large deposits of gunk away. I've seen servers covered in a 2 inch layer of coal dust and still work fine. While dust minimises the effectiveness of components to dissapate heat, it's unlikely it's causing this issue.

Need a lot more information to make a reasonable edumacated guess. When you first turn it on, do you hear a sequence of beeps, do the LED's on the keyboard flash? If you do hear beeps then it's your system board giving you an audible indication of why it believes it can't POST. If it doesn't beep, could be sys board, PS or proc issue. (I've seen proc's pop out of sockets when they're hot, but not when they're cold and I could count on one hand how many proc's I've replaced due to a legitemate issue on one hand in fifteen years)

When it locks up mid boot, does it write an event out to the system log?

If you don't have any gear to prevent ESD (reasonable static strap), I wouldn't take the cover off, particularly if you have warranty on the box. This sort of issue is a classic symptom of a component damaged by ESD in the first place.

[rodderz]

Originally Posted by Rodp

Software memtesters aren't the be all end all, in fact I wouldn't trust it to find anything but an obvious problem. Once you're at the point of running memtest, your PC seems stable so if it was a mem issue, you'd likely get 3 long beeps at POST when it couldn't boot and your PC wouldn't be stable once it was up and running considering it's such a small amount of mem.

Unless you have a can of air, I wouldn't rummage around the innards of your PC either, I'd only even use the can of air in small doses to blow off large deposits of gunk away. I've seen servers covered in a 2 inch layer of coal dust and still work fine. While dust minimises the effectiveness of components to dissapate heat, it's unlikely it's causing this issue.

Need a lot more information to make a reasonable edumacated guess. When you first turn it on, do you hear a sequence of beeps, do the LED's on the keyboard flash? If you do hear beeps then it's your system board giving you an audible indication of why it believes it can't POST. If it doesn't beep, could be sys board, PS or proc issue. (I've seen proc's pop out of sockets when they're hot, but not when they're cold and I could count on one hand how many proc's I've replaced due to a legitemate issue on one hand in fifteen years)

When it locks up mid boot, does it write an event out to the system log?

If you don't have any gear to prevent ESD (reasonable static strap), I wouldn't take the cover off, particularly if you have warranty on the box. This sort of issue is a classic symptom of a component damaged by ESD in the first place.

Thanks Rod and others for the help in this thread

I ran the memtest again today, for nearly 6 hrs while I was out, and it did 50 passes and found no faults. However, when it just kept going and not stopping I just tried some of the options. One of them I tried was an error report on Bad RAM sectors, where it listed 1000's of errors.... confusing....

As what you said about the dirt not affecting much inside the box, a few months back i had a dvd burner installed and some errors fixed (hard drive wiped by service pack 2)and it would most likely have been cleaned back then

The good thing is though, today there are that many people who know their way around a computer and parts can be bought so much cheaper, that upgrades can be done for a few $100 instead of over a grand. I just hope by the time I upgrade it's not crashed again as today we signed up for adsl which should start in a few weeks!

[Rodp]

Originally Posted by rodderz

Thanks Rod and others for the help in this thread

I ran the memtest again today, for nearly 6 hrs while I was out, and it did 50 passes and found no faults. However, when it just kept going and not stopping I just tried some of the options. One of them I tried was an error report on Bad RAM sectors, where it listed 1000's of errors.... confusing....

I'm not familiar with memtest86(?), does it say what the errors are? Most half reasonable memory sticks can resolve single bit errors on the fly and it's entirely reasonable to see a few from time to time. A memory address stores both the data and a parity bit so if the data has a single bit error, it can work out which bit is bung and correct it using parity. These single bit errors would probably show up as a fault but would pass a RAM test. If the frequency of these single bit errors is high, then at best you've got a potential problem down the track. Unsure what is considered a reasonable threshold of generic RAM in a PC, I've been spoiled by our equipment telling us that it's exceeded its threshold.

I must say that my PC specific knowledge is reasonably thin, but most Intel hardware shares a reasonable commonality. My machine is an Athlon 3500+ and I've got a 450W PS after having lockups and blowups of lesser PS's. I believe the Intel procs are less critical about a decent PS but I wouldn't discount it. However, you're getting the opposite symptoms to what I would expect from a underperforming PS.

Is your machine a prefab made by a large PC vendor, or is it a shop bought machine? The one (major) advantage of buying from a large vendor is the assurance that it was assembled using ESD precautions. If you ever wanted an example of the sort of strange failure an ESD'ed component can undergo, this is a classic.

As suggested by someone else, I'd start removing all external devices except a monitor during a period where you know your machine is unlikely to boot. If you got yourself a strap and a non-conductive surface and you felt comfortable I'd then start stripping non-essential devices from inside the machine to a bare bones box and attempt to boot it. System board, proc, 1 stick of mem, video card and that would be it. Then adding cards and devices one by one to see if you can isolate the problem.

Never know your luck, I've solved many problems by reseating components.

[Paris Hilton]

You don't need an Anti-Static wrist strap, just make sure you ground yourself by touching the case before you get to work. Also, the best way to solve hardware problems is to use trial and error...Go ahead and rip it apart, unless you're stupid about it, you'll be fine.

[Rodp]

Originally Posted by Selekta

You don't need an Anti-Static wrist strap, just make sure you ground yourself by touching the case before you get to work. Also, the best way to solve hardware problems is to use trial and error...Go ahead and rip it apart, unless you're stupid about it, you'll be fine.

Unless you want to keep one hand on the case the whole time then *you do need* an anti-static wrist strap. If you touch the case before you do the work, then remove your hand from the case, that action alone builds up enough static charge to fry an IC.

The best way to solve hardware problems is to diagnose the fault using all the information available then taking the most likely path. Unlike myself, I doubt the OP has a warehouse full of parts he can use to go down the 'trial and error path'. I know if I took that path I'd be out of a job.

[parawolf]

Originally Posted by Rodp

Unless you want to keep one hand on the case the whole time then *you do need* an anti-static wrist strap. If you touch the case before you do the work, then remove your hand from the case, that action alone builds up enough static charge to fry an IC.

The best way to solve hardware problems is to diagnose the fault using all the information available then taking the most likely path. Unlike myself, I doubt the OP has a warehouse full of parts he can use to go down the 'trial and error path'. I know if I took that path I'd be out of a job.

Actually the difference between having your hand on the case, and using a proper anti-static wrist strap is about 1Meg Ohm. You still need to disapate that static build up without transferring it directly to the chassis.

If you put a ohm meter between the alligator clip and the wrist strap you will find a resistance of about 1Meg Ohm.

[Rodp]

Originally Posted by parawolf

Actually the difference between having your hand on the case, and using a proper anti-static wrist strap is about 1Meg Ohm. You still need to disapate that static build up without transferring it directly to the chassis.

If you put a ohm meter between the alligator clip and the wrist strap you will find a resistance of about 1Meg Ohm.

Unless you weren't aware, a static strap has a current limiting resistor in it which is rated at, oh... 1 megohm.

[parawolf]

Originally Posted by Rodp

Unless you weren't aware, a static strap has a current limiting resistor in it which is rated at, oh... 1 megohm.

Indeed - i'm just making the point, 'earthing' yourself to the chassis is not the same as using a anti-static wrist strap or similar device

[Steffo]

If you want to see if the RAM is at fault, try playing Solitaire. It is the most RAM sensitive application Windows has, and is renouned for going pear shaped with the slightest issue with your RAM. If it works fine, that's not the issue.

Personally, I'd see this as an opportunity for a decent upgrade...

Throw out the P4, its mobo and the 256 stick of RAM... get an Athlon64, 1GB of DDR400, a good Skt939 m/b with PCI Express (like an Asus A8N-E), an decent PSU (like an Antec TruePower 430W) and you've got it made.

But yeah, sounds like it could be either the RAM, motherboard, CPU or Power Supply being dodgy.. though I'd blame it on the first three before considering the PSU.

[Paris Hilton]

Oh come on. It has problems starting up, but once it is going, it runs fine. The power supply is not supplying enough power, hence the system is not starting up as stable as it should, because during the POST every component of the system is switched on, hence why it is called a POST, it tests everything. If it was the CPU, you'd have constant stability problems. At first I suggested the RAM, but that was because I hadn't read the question throughly. As for the static wrist strap, if you feel there is a danger, by all means use one, but I never have and i've never had a problem.

[RED_EL_XR8]

Originally Posted by Selekta

Oh come on. It has problems starting up, but once it is going, it runs fine. The power supply is not supplying enough power, hence the system is not starting up as stable as it should, because during the POST every component of the system is switched on, hence why it is called a POST, it tests everything. If it was the CPU, you'd have constant stability problems. At first I suggested the RAM, but that was because I hadn't read the question throughly. As for the static wrist strap, if you feel there is a danger, by all means use one, but I never have and i've never had a problem.

So the mystic powers let this ailing PSU continue, it could be the PSU yes but it could be any of 50 other things, if "jumping to conclusions" becomes an olympic sport you should really consider entering in the trials. Diagnosis, process of elimination, measurements with test equipment are the ways to diagnose faults, blind guessing will just end up with purchase of a lot of parts that weren't needed.

[Rodp]

Originally Posted by Selekta

Oh come on. It has problems starting up, but once it is going, it runs fine. The power supply is not supplying enough power, hence the system is not starting up as stable as it should, because during the POST every component of the system is switched on, hence why it is called a POST, it tests everything. If it was the CPU, you'd have constant stability problems. At first I suggested the RAM, but that was because I hadn't read the question throughly. As for the static wrist strap, if you feel there is a danger, by all means use one, but I never have and i've never had a problem.

This is starting to stray far from the OP, but I guess most threads do.

Firstly, there are many reasons why a PC won't POST and very few of them have anything to do with the supply of voltage on the rail. Secondly, POST doesn't test everything. In fact it barely tests anything. A POST is a very rudimentary test for a PC's ability to boot a ROM. There are many combinations of failures that equiment can have that will past POST at boot.

It would appear you still haven't read the question thoroughly because once it's going it still isn't fine innitially. Try reading the question again.

Lastly, because you've experienced no problems with ESD does not mean it isn't a problem. In fact, that's the biggest problem with an ESD failure... most of the time it isn't a hard failure but a latent defect. As part of ISO9000 certification, I'm required to sit through a boring environmental course on ESD packaging, handing and procedure every 6 months (for the past 15 years). Once a month I have to submit my straps and mats for testing. I've seen the effects of ESD under a microscope (it's part of the course) the surface of the IC is reasonably flat until it's touched by a finger at a higher static potential, then the IC looks like a crater on the moon. The same IC is on a board and will happily past POST, however, it has a largely diminished capacity (latent defect).

The company I work for estimates losses of about $600 million a year due to inadequate handling of static sensetive components. The equipment I work on is what keeps multinational companies running. If I was to walk up to a non-stop server, touch the case then start working on it, I'd be lucky if I wasn't fired on the spot. If I was to incorrectly handle a component that has been removed from a server and is proven to have failed, I still risk instant dismissal.

http://www.esda.org/esd_fundamentals.html

I suggest you read the above.