PDA

View Full Version : Machine flakes out after 10 minutes (Was: one WEIRD problem...



Mike Chambers
November 25th, 2006, 06:18 AM
yeah i don't understand what in the world could be causing this one.

on my XT clone, i am using a network drive for my C disk out of a 3Com Etherlink II-TP (3C503) on IRQ 5

(i also have an NE1000 compatible card in there on IRQ 2 for use of a packet driver and TCPDRV/WATTCP based software while using the network drive.)

it all works fine together, but occasionally (and it seems to be getting more frequent??) i can no longer access any network drives. i get network error codes from the MS net client (v3.0) and if i ctrl+alt+del it simply won't reconnect to the drive.

that's not the strange part. this is. after this happens, my only recourse is to physically turn off the power and turn it back on, right? ok, so i do that. whenever i do after one of these network errors (and occasionally at other times) none of my floppy drives work AT ALL! no light on boot-up, nothing. just "Please insert a disk into drive A and press RETURN"

and if you DO hit return, the LEDs on the keyboard just light up and the system halts.

the floppies absolutely refuse to work unless i leave the whole system powered off for a good 10+ minutes!!

has anybody ever seen anything like this?? :(

any help would be greatly appreciated!

mbbrutman
November 25th, 2006, 06:35 AM
First, the editorial note.

You are being scolded for not getting a PC specific problem in the PC specific area of the forum. I moved it. You are also being scolded for being to vague in the subject line. Subject lines should give you an idea of what the message is about. You should edit the subject to make it more descriptive.

Now, as for the problem:

I don't think you have a port or interrupt problem, unless you are writing some really bad software that is stomping on areas of memory in use by DOS or the packet drivers.

I would suggest turning the machine on and then walking away from it. If it continues to work for an extended period of time, it is something you are doing to it. If it fails on it's own, then it might be overheating, or you have some other electrical problem. Be careful walking away from it .. you wouldn't want it to catch fire.

The key to your msg is that you say it has to cool down for 10+ minutes. That's a very important clue.

Mike Chambers
November 25th, 2006, 06:46 AM
First, the editorial note.

You are being scolded for not getting a PC specific problem in the PC specific area of the forum. I moved it. You are also being scolded for being to vague in the subject line. Subject lines should give you an idea of what the message is about. You should edit the subject to make it more descriptive.

Now, as for the problem:

I don't think you have a port or interrupt problem, unless you are writing some really bad software that is stomping on areas of memory in use by DOS or the packet drivers.

I would suggest turning the machine on and then walking away from it. If it continues to work for an extended period of time, it is something you are doing to it. If it fails on it's own, then it might be overheating, or you have some other electrical problem. Be careful walking away from it .. you wouldn't want it to catch fire.

The key to your msg is that you say it has to cool down for 10+ minutes. That's a very important clue.

sorry... it's kind of confusing sometimes since "Hardware" is under the "Technical Support" subforum and i was seeking tech support :)

i'll remember to put these sorts of questions here from now on though. btw, it doesn't let me change the original message's topic after posting.

as for the problem at hand, no it's nothing i'm doing. it will even happen if it's just sitting at the DOS prompt and i leave the room for a while to do something else. i can come back and say, CD to a directory on the network drive and it'll give me the error.

you're right about it possibly overheating. now that i think about it, i do have the operating speed jumper now set to 8 MHz instead of 4.77! i guess i can switch back to snail mode and see what happens.

is this a common problem? :neutral:
(for the record, as you may recall i've posted before i'm using a Turbo PC model board made by PC's Unlimited.)

maybe if switching it back to 4.77 fixes it, i could consider fitting some sort of heatsink on the CPU.

EDIT: that did nothing :/ this just makes the problem that much weirder. if it's not an overheat problem, then what in the world would i have to wait 10 minutes for? (i actually put my finger on the CPU. it's not even that hot. YAY for randomly touching heat-generating electronic devices!!)

i'm going to take out that NE1000 clone and see what happens.

mbbrutman
November 25th, 2006, 07:14 AM
Yes, I know .. too many wonderful options for placing a post. I'd leave the hardware area for stuff that is not necessary platform specific.

It doesn't have to be hot to be flakey. You might have a marginal chip, capacitor, other other component that starts flaking out when it is warm. The heat sink trick won't help you in that case because you don't know where to put it.

If the machine is definitely flaking out on it's own without your help, it's time to start problem diagnosis. Here are a few suggestions:


Boot with pure DOS (no drivers) and leave it alone. If it doesn't flake out, it is software induced.
If it continues to flake out, start stripping non-essential components. Then try to make it flake out.
A memory exerciser would be a good piece of software to run. There are other diagnostic programs out there too.
Check the output of the power supply.


Remember past posts about the evils of static electricity and how it can cause stuff to flake out down the road? Use a static strap and proper grounding procedures.

chuckcmagee
November 25th, 2006, 07:26 AM
I got a SECOND DEGREE BURN on my finger once doing the "check for heat" trick. I had accidently rotated the CPU chip 90 degrees before inserting it in the motherboard. It was fried and so was my finger tip. (2nd degree is where it bubbles up, not so fond memories). Guess I'll lick my finger first next time.

modem7
November 25th, 2006, 01:42 PM
In such a situation that's been described, I'd be bringing out my can of freeze spray (specifically produced for diagnostic purposes).
Narrow it down to a section of the board first, then start homing in on the problem component.

Be careful not to overdo it, because some components can be damaged from large thermal changes over a short period of time (i.e. thermal shock).

DimensionDude
November 25th, 2006, 03:38 PM
Not a definitive heat test, but you could try setting a desk fan (not a processor fan) to blow on the chassis and see if that increases your runtime.

Kent

Mike Chambers
November 25th, 2006, 07:51 PM
In such a situation that's been described, I'd be bringing out my can of freeze spray (specifically produced for diagnostic purposes).
Narrow it down to a section of the board first, then start homing in on the problem component.

Be careful not to overdo it, because some components can be damaged from large thermal changes over a short period of time (i.e. thermal shock).

never heard of that stuff.. its for diagnostics? that's pretty cool, and useful!




Not a definitive heat test, but you could try setting a desk fan (not a processor fan) to blow on the chassis and see if that increases your runtime.

Kent

i tried it... nothing different happened. :(

i think i've isolated the problem though. it'd not faulty hardware. it seems to only happen when i try to use either my NE1000 (IRQ 2) and my 3C503 (IRQ 5) OR use the 3C503 and my COM2 port.

mbbrutman
November 26th, 2006, 06:17 AM
Which means you didn't follow the procedure - turn it on with nothing but a pure DOS disk and see how long it lasts.

So now we're back to software ...

Mike Chambers
November 26th, 2006, 10:07 AM
Which means you didn't follow the procedure - turn it on with nothing but a pure DOS disk and see how long it lasts.

So now we're back to software ...

it works indefinately on a pure DOS disk as far as i can tell. i let it go for a few hours and i came back and it seemed to be working fine.

nige the hippy
November 26th, 2006, 12:17 PM
might be being stupid, but wouldnt an irq conflict between the network card and com2 be problematic, (irq 5). from what I gather stuff comes in on the network and doesn't get processed??
( out of the mouths of babes and fools...)

Mike Chambers
November 26th, 2006, 03:15 PM
might be being stupid, but wouldnt an irq conflict between the network card and com2 be problematic, (irq 5). from what I gather stuff comes in on the network and doesn't get processed??
( out of the mouths of babes and fools...)

actually my COM2 is set to use IRQ 4... which makes this even more weird.

Anonymous Coward
November 26th, 2006, 05:22 PM
Doesn't the parallel port or disk controller usually take IRQ5 in 8088 machines? Can you try setting your NIC as IRQ3? Also, you might want to disable one of your serial ports if you have two of them. I think that COM2 normally takes IRQ3. COM1 should be IRQ4. I usually run my NICs on IRQ3 in my 8088 systems.

mbbrutman
November 26th, 2006, 05:59 PM
The IRQ sharing usually isn't a problem unless more than one device is actually using the IRQ.

For example, parallel ports are commonly assigned IRQs. But if you never actually enable the interrupt on the device, it might as well not be assigned an interrupt. Parallel ports on older machines with older operating systems generally never use interrupts - all of the software polls the parallel port. It was a big deal when OS/2 gave people the option to use interrupts on the parallel port - it meant that during printing your machine wasn't consumed by trying to service the parallel port.

Same thing with the COM ports. You can have multiple COM ports sharing IRQs, if only one uses the shared IRQ. Most older COMM programs poll the ports directly instead of using interrupts. (An exception might be code designed to use the 16450 or 16550, which has the deeper FIFOs.)

modem7
November 26th, 2006, 11:14 PM
One of the symptoms is, "the floppies absolutely refuse to work unless i leave the whole system powered off for a good 10+ minutes!!"
And so it sounds like a hardware problem that appears in particular set of conditions (elevated temperature plus certain IRQ usage?).
Bad electronic component or bad motherboard design (e.g. inadequate component cooling) ?

Luke
November 27th, 2006, 04:11 AM
I had the same problem with floppy drive in my XT when it came, after few minutes/hours it didn't respond.
It was bad power connector, after repairing it I never had the problem again.

One of your network cards have IRQ set to 5 wich is reserved for hard disk on XT computers, you don't have HDD, but you have controller wich propably use this IRQ.
Try to set network card to IRQ 7 if you don't have parallel port in the machine.