PDA

View Full Version : Parity check



evildragon
August 3rd, 2012, 08:49 PM
Oh boy, I think my IBM is nearing it's demise.

I went to turn it on, and it started booting, and when it was connecting to a NTP server to set the clock, the entire screen went blank with this on it:

http://img.photobucket.com/albums/v395/Evilweredragon/parity.jpg

So I turn it off, wait a minute, turn it on, and I kept getting this:

http://img.photobucket.com/albums/v395/Evilweredragon/mem.jpg

I then turned it off, frustrated.. The memory lasted all these years and now it's dying. At this point, I'm thinking I killed the freaking thing with that Bad Apple demo I made a couple weeks ago. So now I feel I'm a bad programmer and pushed the hardware too far that it actually killed the memory, as that demo was literally, the hardest thing this IBM has EVER played.

So a couple hours passed, and I tried turning it on again, now it's booted up nice and fine. So I ran PC-Check on it and did a thorough memory test, and it passes now.

What the!? The memory was failing, now it's fine?

Chuck(G)
August 3rd, 2012, 08:59 PM
You're not a farmer are you? ("Parity check" - what a farmer gets from the government for not raising corn) :)

It could be a wonky power supply that's causing this (check your voltages). Otherwise, keep a log of the reported failing locations. Right now, it doesn't look like a genuine memory error.

evildragon
August 3rd, 2012, 09:10 PM
Ok I'll keep a watch on it.

Isn't this what the power good signal is supposed to do though? The power supply levels out and tells the motherboard things are good to go?

Crypticalcode0
August 3rd, 2012, 11:58 PM
Not completely, all it does is tell if the PSU is okay.
If it is not there at boot it will reset the CPU if it falls below spec it shuts down the CPU during operation.

Again this is taking the PSU's word for it that it's OK.

On the board of the computer this can be a different story. ;)

pearce_jj
August 4th, 2012, 12:01 AM
Also faulty expansion cards can cause this.

Chuck(G)
August 4th, 2012, 06:52 AM
Ok I'll keep a watch on it.

Isn't this what the power good signal is supposed to do though? The power supply levels out and tells the motherboard things are good to go?

PG is a very poor indicator. It's supposed to be raised when power supply levels are stable. Usually, it's something simple, such as a time delay circuit or comparator.

evildragon
August 4th, 2012, 08:15 AM
I just noticed something strange here.. When I type "mem" at DOS, it says I have 638KB conventional memory.

Uhhhhhhh? Where'd did the 2KB go?! The BIOS counts to the full 640KB and passes POST test.

Chuck(G)
August 4th, 2012, 08:21 AM
Are you using a DDO or disk capacity extender? if so, that's where the memory went.

If not, anything not part of baseline MS-DOS?

evildragon
August 4th, 2012, 08:28 AM
No extender I can think of..

I just tried ROM BASIC and it's showing short too.

http://img.photobucket.com/albums/v395/Evilweredragon/pccheck.jpg
http://img.photobucket.com/albums/v395/Evilweredragon/basic.jpg

I have an XT-IDE rev 2 installed, but would that cause ROM basic to show up short? I do have it in full operating mode and it's set to use 1KB of RAM, but im seeing 2KB loss.

Agent Orange
August 4th, 2012, 08:35 AM
I think at I'd start swapping and/or replacing some RAM about this time.

evildragon
August 4th, 2012, 08:35 AM
I don't have any other RAM to put in this machine.

pearce_jj
August 4th, 2012, 09:02 AM
Take out the XTIDE (or disable it's ROM) and check again.

evildragon
August 4th, 2012, 09:04 AM
I'm going to try that later. Right now Im watching TV with my little sister, taking a break from it.

Stone
August 4th, 2012, 09:07 AM
I just noticed something strange here.. When I type "mem" at DOS, it says I have 638KB conventional memory.

Uhhhhhhh? Where'd did the 2KB go?! The BIOS counts to the full 640KB and passes POST test.This isn't a new machine -- you've had it some time, right? And, it always had 640K before when you typed mem. A boot sector virus or other bad boy could cause this. I know I've seen malware reduce the 640K to 639K at boot and I'm not familliar enough with all the parasites to remember their specifics but it's probably worth looking into.

evildragon
August 4th, 2012, 09:09 AM
This isn't a new machine -- you've had it some time, right? And, it always had 640K before when you typed mem. A boot sector virus or other bad boy could cause this. I know I've seen malware reduce the 640K to 639K at boot and I'm not familliar enough with all the parasites to remember their specifics but it's probably worth looking into.

I've had the computer since my birth.

Boot sector virus wouldn't affect ROM BASIC.

evildragon
August 4th, 2012, 09:58 AM
I tried this real quick, and it appears the memory is definitely there as mem /d shows it all..

http://img.photobucket.com/albums/v395/Evilweredragon/mem-1.gif

Stone
August 4th, 2012, 09:58 AM
I've had the computer since my birth.

Boot sector virus wouldn't affect ROM BASIC.Who said anything about ROM BASIC?

I had referred to your 638K problem.

evildragon
August 4th, 2012, 10:04 AM
Who said anything about ROM BASIC?

I had referred to your 638K problem.

ROM BASIC was missing 2KB too.. :rolleyes:

So if it was a virus, that wouldn't affect ROM BASIC missing the 2KB also.

evildragon
August 4th, 2012, 10:56 AM
Gah! It passed another test of PC Check, then went to a black screen with:
PARITY CHECK 1
00000(S)

Then when rebooted, passes POST again.

I'm going to remove ALL cards, modules, etc, and do a thorough cleaning.

evildragon
August 4th, 2012, 03:53 PM
Reseated the memory, and went to load compushow to load a picture and got a Stack Overflow, System Halted.. Control alt delete does NOT work, everytime these errors come up.

I even reseated the expansion cards.

Didn't measure power supply levels yet.

cr1901
August 4th, 2012, 04:29 PM
Based on everything said so far (especially since you got errors without expansion cards), I'd imagine, like others have said, the memory is intermittently dying. However, the memory should be easily replaceable, provided the error isn't with the first memory bank (at which point you'd have to de/resolder the memory from the board- nothing more relaxing than an hour of sold). The IBM PC takes 4164 memory ICs (please correct if I'm wrong) as I recall, and can be found on Ebay at modest prices in bulk if you look for it:

http://www.ebay.com/sch/i.html?_nkw=4164+ram&_sacat=0&_odkw=4164&_osacat=0

BEFORE THAT however, my suggestion for now is write down any error codes that pop up, and consult the PC technical reference manual for clues on the problem- as I remember, there are sections for diagnosing hardware errors using th return codes (the numbers BESIDES 201- memory error, 301- keyboard error, etc).

It could also possibly be a power supply error- idk if the PC returns error codes for power supply failure, however (does anyone else know?).

Stone
August 4th, 2012, 04:36 PM
The IBM PC takes 4164 memory ICs (please correct if I'm wrong)...Isn't this a PS/2?

evildragon
August 4th, 2012, 05:07 PM
Yes, this is a PS/2 Model 25, and it uses TWO different types of memory, 2x256KB SIMMS, and 6 DIP chips to make up 256KB (so technically to 768KB RAM but 128KB is discarded at the end).

evildragon
August 4th, 2012, 08:12 PM
Well, it's dead.

I found my other Model 25 motherboard (that's dead, but has good SIMMS), and with that memory stick, it still locks up with parity.

This is without expansion cards and booted to a floppy.

Looks like it's dead..

Model 25 - RIP - 1987/2012

modem7
August 4th, 2012, 08:23 PM
Well, it's dead.
I found my other Model 25 motherboard (that's dead, but has good SIMMS), and with that memory stick, it still locks up with parity.
This is without expansion cards and booted to a floppy.
Looks like it's dead..
The "08A9DA" in your first post is sure to correspond to address 554 KB.
You earlier wrote, "and it uses TWO different types of memory, 2x256KB SIMMS, and 6 DIP chips to make up 256KB".
So could not the problem be in the "6 DIP chips"?

evildragon
August 4th, 2012, 08:34 PM
The "08A9DA" in your first post is sure to correspond to address 554 KB.
You earlier wrote, "and it uses TWO different types of memory, 2x256KB SIMMS, and 6 DIP chips to make up 256KB".
So could not the problem be in the "6 DIP chips"?

Oh you're right, I only switched the SIMMs..

Can you tell me then how I can calculate the memory address to KB location? That way I can easily do that on my own.

I have my friend over right now and we both kept restarting the PC-Check memory test, and it kept passing.. Then, he did the test again an hour ago, and when I went to check on it, it was at the SMC ethernet packet driver initialization screen, completely frozen, as if it froze when rebooting itself during a memory test.

modem7
August 4th, 2012, 08:39 PM
Model 25 - RIP - 1987/2012
RIP = Random Inconsistent Parity ??????


Can you tell me then how I can calculate the memory address to KB location? That way I can easily do that on my own.
08A9DA hexidecimal = 567770 decimal
567770 / 1024 = 554 KB

evildragon
August 4th, 2012, 08:41 PM
RIP = Random Inconsistent Parity ??????
Yea, lets go with that.


08A9DA hexidecimal = 567770 decimal
567770 / 1024 = 554 KB
Thank you, that is exactly what I need to know.

evildragon
August 4th, 2012, 09:06 PM
For right now I am using the CRT on my 480p DVD player (which outputs a compatible scan rate on the Y output on the Component video jack--just needs sync separated with the correct polarity)..

Watching Hellraiser on it. Never seen it before..

It is always interesting to see a movie be played on this CRT. Black and white, but it looks better than a regular black and white TV. It's progressive scan!

The way I see it, until I resolve the RAM issue, I might as well use the CRT.. It works at least.

modem7
August 4th, 2012, 09:20 PM
The way I see it, until I resolve the RAM issue, I might as well use the CRT.. It works at least.
From the Model 25 section of the 'Upgrading and Repairing PCs' book: "512K of RAM standard, expandable to 640K on the motherboard."
Perhaps try running the board with only 512 KB fitted for a while.

evildragon
August 4th, 2012, 09:31 PM
From the Model 25 section of the 'Upgrading and Repairing PCs' book: "512K of RAM standard, expandable to 640K on the motherboard."
Perhaps try running the board with only 512 KB fitted for a while.

I'm going to do that tomorrow. For right now I'm watching a movie on it's screen. (Lady just played with the cube). Can't pause it now. ;)

modem7
August 4th, 2012, 10:18 PM
I'm going to do that tomorrow. For right now I'm watching a movie on it's screen. (Lady just played with the cube). Can't pause it now. ;)
Your Model 25 is suffering, and you're watching TV !
You're not welcome in my house.

evildragon
August 4th, 2012, 10:24 PM
Your Model 25 is suffering, and you're watching TV !
You're not welcome in my house.

It's not suffering when the motherboard is removed. I AM however examine solder joints on it while it's CRT does other things.

Crypticalcode0
August 5th, 2012, 02:32 AM
couldn't you shadow the BIOS in RAM on the PS/2?

evildragon
August 5th, 2012, 08:56 AM
couldn't you shadow the BIOS in RAM on the PS/2?
No, but what would that do to help a memory issue anyway?

Crypticalcode0
August 5th, 2012, 10:26 AM
Shadowing BIOS in RAM is a speed trick, if that is done calling BIOS routines would use those RAM chips i think. anyhow it is not important unless your system does that Mine does and this had cause me some headaches in the past.

evildragon
August 5th, 2012, 03:38 PM
Ugh, so I tried without the DIPs and only on the SIMMs for 512KB.. Freezes, PARITY, but no memory read error, just PARITY.

I then try the DIPs only, same god damn thing!

So I remembered, I did try something slightly different. I had flashed the XT-IDE rev 2 with the XTP BIOS, which gives it 186 instructions.. Perhaps, though faster, it's just screwing up.

So it's now on the regular 8086 BIOS, and I'm going to see how it goes. I was at least able to boot Windows now. :/

evildragon
August 5th, 2012, 03:49 PM
Nope that didn't work. It froze again.

PC Check froze when testing block address 42A00H.

EDIT: I found my Intel 8086 CPU. It's dated 1977.. Going to see if it freezes with that too.

EDIT2: Nope, instead it crashed in a very freaky way: http://img.photobucket.com/albums/v395/Evilweredragon/interleave.jpg

It all of a sudden wants to low level format my disk? WHAT?!

EDIT3: Removed the XT-IDE Rev 2 card.. I'm now at 639KB main memory. But, where the heck is the other KB? I have NO expansion cards installed now and am STILL short a KB. This don't make any sense.

EDIT4: Well, having the XT-IDE removed fixed the issue..

Now to test the power supply and voltages..

EDIT5: Power levels are fine.. I guess I found the culprit, though not sure why 1KB is still missing, but the XT-IDE being removed IS fixing the issue.

EDIT6: Now put the XT-IDE back in, but with a different empty EEPROM. Will try memory test again before flashing this EEPROM. Technically the EEPROM isn't empty, it's out of a cars ECM, but the computer will just ignore it since it has no idea what to do with it.

RJBJR
August 5th, 2012, 05:14 PM
This is from the PS/2 Model 25 Technical Reference, page 5-29. It seems to suggest that the BIOS POST will come up with 639k during POST memory count if there is 640k present in the machine.

You can download the Tech Ref here (http://bitsavers.org/pdf/ibm/pc/ps2/PS2_Model_25_Technical_Reference_Jun87.pdf)

Interrupt 12H - Memory Size Determination
This routine returns the amount of RAM in the system as determined
by the POST routines.
The following are the memory determination assumptions:
All installed memory is functional. If the memory test during
POST indicates less, that value becomes the default.
All memory from 0 to 640K must be contiguous.
Note: The memory value returned will be the total system memory
minus the 1K block of extended BIOS memory. A 640K machine will
return 639K if all the memory is functioning properly.
ON RETURN:
(AX) - Number of contiguous lK blocks of memory

evildragon
August 5th, 2012, 05:20 PM
Ok well that make sense then..

I have another EEPROM on my XT-IDE rev 2 right now, and so far it is working. Will continue to do extensive testing.

SpidersWeb
August 5th, 2012, 05:24 PM
Chances are the PS/2 BIOS is using 1Kb for extended data. Don't worry about the 639KB.

Edit: bah I pressed submit too late!

pearce_jj
August 5th, 2012, 11:10 PM
Removed the XT-IDE Rev 2 card.. I'm now at 639KB main memory...EDIT4: Well, having the XT-IDE removed fixed the issue..

A good time to mention posts 5 and 12 :D ?

modem7
August 6th, 2012, 12:36 AM
A good time to mention posts 5 and 12
Yes. I can see why.
Today is International Day of Spelling and Grammar, and in post #12, you have used "it's" instead of "its".

evildragon
August 6th, 2012, 01:16 AM
Yes. I can see why.
Today is International Day of Spelling and Grammar, and in post #12, you have used "it's" instead of "its".

I ran my XT-IDE rev 2 for the rest of the night with a space 28c64 EEPROM, and I did not have one single crash.

It was a faulty EEPROM this entire time.

evildragon
August 6th, 2012, 10:53 PM
I ran my XT-IDE rev 2 for the rest of the night with a space 28c64 EEPROM, and I did not have one single crash.

It was a faulty EEPROM this entire time.

Ok, so it crashed again.

But I found out what's causing it cause I can make it happen on-demand.

The new AC unit that was installed in our house, is what's causing it.. Every time the heat pumps compressor kicks in, is when the IBM crashes.

Ugh, couldn't it have been a more "normal" issue?

pearce_jj
August 6th, 2012, 11:28 PM
Presumably the compressor isn't soft-start (DC-Inverter)? So it's likely brown-out at the supply; I'm thinking dried-out electrolytic caps in the PSU.

evildragon
August 6th, 2012, 11:35 PM
Presumably the compressor isn't soft-start (DC-Inverter)? So it's likely brown-out at the supply; I'm thinking dried-out electrolytic caps in the PSU.
I don't know what these new Trane compressors use, but it sounds like it starts pretty hard.

I removed the motherboard in preparation of checking things while it was outboard, and I was greeted with this while it was outboard..

Can you spot the "problem"? All I'll say is, if this was like this for a while, this poor sucker has been starving for amps on the motherboard.

http://img.photobucket.com/albums/v395/Evilweredragon/problem-1.jpg

RJBJR
August 6th, 2012, 11:43 PM
It would be interesting to plug a multimeter into the wall socket and watch voltage drop when the compressor kicks in.

What's that red wire? Looks to be an orphan or an escapee?

Crypticalcode0
August 6th, 2012, 11:47 PM
That is a 5V from the AT connector from what i can tell in the picture.

evildragon
August 7th, 2012, 12:04 AM
Yes, it's the 5v rail going to the motherboard..

So, I would assume, if this wire has been working it's way loose to the point of not even connecting, then no wonder a AC compressor is causing it to fail, it's already being deprived of amps.

I fixed it and am testing again.

But I found it interesting that the BIOS chips are getting rather warm on this motherboard.. The two chips are 118F each, the RAM is cooler than that.

mikey99
August 7th, 2012, 05:22 AM
....

The new AC unit that was installed in our house, is what's causing it.. Every time the heat pumps compressor kicks in, is when the IBM crashes.

.....

I had a similar problem with a PCjr many years ago, it would just randomly reboot at times. Spent a lot
of time trying to figure out what the problem was.....one night I just happened to be in another room
of the house, and heard the A/C kick on, just as I saw the screen on the PCjr rebooting in the other room.
Turned out the brick power supply was affected by the voltage drop.

evildragon
August 8th, 2012, 10:16 PM
So I fixed the wire, and so far no problems when the AC kicks in. So that broken 5v rail mail have been the problem.

But because the planar was so caked with dust, I took it out and decided to use a brand new tooth brush and rubbing alcohol and give it a cleaning.

http://img.photobucket.com/albums/v395/Evilweredragon/model25planar.jpg

http://img.photobucket.com/albums/v395/Evilweredragon/model25planar2.jpg

So far so good..

evildragon
August 9th, 2012, 07:16 AM
So I fixed the wire, and so far no problems when the AC kicks in. So that broken 5v rail mail have been the problem.

But because the planar was so caked with dust, I took it out and decided to use a brand new tooth brush and rubbing alcohol and give it a cleaning.

http://img.photobucket.com/albums/v395/Evilweredragon/model25planar.jpg

http://img.photobucket.com/albums/v395/Evilweredragon/model25planar2.jpg

So far so good..

Gah! Didn't work. So I replaced the BIOS EPROM with the CMOS ROM version. Same thing.

I measured the house AC voltage when the air conditioning turns on as it still seems to be the trigger, but my analog meter showed NO movement of the needle when it kicked in.

SpidersWeb
August 9th, 2012, 12:41 PM
Replace the line filter caps in PSU, or get a UPS?

evildragon
August 10th, 2012, 02:50 PM
Well this has to be the strangest thing I've ever seen. I took apart the model 25 and took a look at the power supply with an ESR meter and all the capacitors are fine. Then I noticed, once again with the XT – IDE card removed the computer does not freeze no matter what. So because it was memory related I changed the EEPROM with a different one but it still did it. I then tried with the disk on module removed and booted from floppy. Still freezes when the ac compressor kicked in. So I installed the original rev 1 XT - IDE and it works perfectly fine no crashes at all.

What the hell? Is it possible that a tantalum capacitor got damaged during soldering and is having issues?