• Please review our updated Terms and Rules here

Help fixing my Olivetti M24 SP

freakedenough

freakedenough

Experienced Member
Joined
Jul 30, 2016
Messages
94
Location
Vienna, AT
My M24 recently started to freeze with some garbage on the screen, also when idle-ing in the DOS-prompt.
There are 512KB of memory soldered onto the motherboard and 128KB socketed. The socketed DIP16 ones were already replaced by me.
However, I noticed that the issue persists as long as the M24 is warm/turned on. If i leave it turned off for a while, it boots as normal. If i reset it or cold boot while it is still warm, it results in garbage on the screen, no POST at all or a crippled "DMA CONTROL FAIL:04" error message.

Could this be caused by the soldered memory? I don't want to solder if its not necessary. I've for sure tried to remove all ISA-Cards except for the graphics card, left it running only with the soldered memory, yet the issue persists.

Idle-ing:

Freeze:

Reset (while being still warm):
 
Maybe unplug the graphics card, clean it's contacts and reseat it.

And also consider, that a PC has not only system memory, but also video card memory.

Please run also the M24 customer test, alternatively also Checkit 3.0 mainboard tests.
 
Within the last days I did the following:
- Removing the graphics card and cleaning the contacts with a contact cleaner.
- The memory configuration jumpers were incorrect. I have 512KB in BANK0 and 128KB (socketed) in BANK1. The jumpers were set for the other way around, resulting in a warning of customer test. (DIP0, SW1-4 and SW8 BANK1 enabled). However, this was the case for the last 30 years now, so its imo not the root cause.
- In the end, customer memory test passed without errors. Same for CheckIt 3.0.
- I've got a Monotech MicroRAM yesterday which allows me to enhance (if needed) base RAM and use UMB areas. I've configured my M24 to 128KB BANK0 RAM (despite of having 512KB soldered on it) and 0KB in BANK0 and configured the Monotech card to fill up the space from 128-640KB.
- Reseated all DIPs of BANK1 into their sockets after cleaning them with a contact cleaner (not using the Monotech card at all).
- Replaced the DIPs in BANK1 with 256kx1bit DIPs I took from a 286 (not using the Monotech card at all).
- I've found this post here:
http://www.vcfed.org/forum/showthre...ort-POST-diagnostic-codes&p=559927#post559927
Trixter said:
According to ftp://ftp.oldskool.org/pub/drivers/...T Personal Computer 6300 ROM BIOS listing.pdf it is a failure found while testing the DMA chip in the first 64K. So yes, you have a bad chip in the first 64K. As for piggybacking, I haven't had good success with that, but I haven't had a lot of success troubleshooting drams anyway, so it couldn't hurt to try. You can also try the finger test, to see if one of the chips is substantially hotter than the others.
Click to expand...
- Thus I've tried piggbacking all of the 18 soldered chips with 256kx1bit DIPs I took from a 286 while having BANK1 empty and disabled (not using the Monotech card at all), one by one, with no outcome.

All of the above didnt fix the issue. Currently the M24 POSTs correctly in any case but locks up randomly while booting to DOS or using mTCP apps. A freeze is permanently reproducible when starting CLOCK.EXE in Win2.11 and 3.0 under PC-DOS 5.0 - the mouse pointer can be still moved around for another minute, but Windows and the keyboard is locked up, except for CTRL+ALT+DEL. On the other hand I can start as many other applications, including CONTROL.EXE, until RAM is full. Since this is strange, I already disabled the RTC via jumper, but no change. If I boot from MS-DOS 4.01 floppy and start Win from there, also CLOCK.EXE works. Maybe this is some weird IBM-DOS compatibility issue and the initial problem is resolved. Today I will try again to switch to MS-DOS 5.0 which previously already locked up while booting from Setup Disk 1 in 10 out of 10 attempts (which is why I used IBM DOS 5 instead, to enable me access to UMB area for the Monotech card). Fingers crossed.

Which chips in BANK0 are the first 64K? Since I do not trust in piggyback that much and due to the DMA:04 error, I assume I have to desolder 4 chips, but I do not know which ones:
https://hosting.photobucket.com/albums/a166/ballsandy/Computer related/IMG_1830.jpg
https://stason.org/TULARC/pc/motherboards/O/OLIVETTI-8086-M24-M24SP-1.png
 
Last edited:
Try to get Olivetti or AT&T branded MS-DOS. Originally these machines were delivered with MS-DOS 3.20/3.30, maybe that fixes your issues. I only could provide you german versions of 3.30 and 5.0 olivetti branded. (and they are on 1.44mb image)

Or try original Microsoft branded.
 
If you have a German Olivetti branded MS-DOS 5.0 I would be happy if you could share it with me. I have 3.20 on my own, but it lacks support for the 1GB XTIDE partition.
 
The fact that your issue doesn't happen when the system is cold, but then starts to happen as it warms up, suggests either faulty components or bad solder joints. It might be helpful to reflow the solder joints for at least the DRAM chips.

I've configured my M24 to 128KB BANK0 RAM (despite of having 512KB soldered on it) and 0KB in BANK0 and configured the Monotech card to fill up the space from 128-640KB.
Click to expand...

I don't think that will work? Won't the physically soldered 512K RAM be the RAM that responds to the address lines?
 
I could install MS-DOS 5, but I had to remove autoexec.bat and config.sys from the setup disk, to avoid loading "nlsfunc" and "keyb gr", otherwise it locks up. But OK, I know that the M24 keyboard is very specific.
However, in the end I got this error when trying to run CLOCK.EXE on the first attempt. Afterwards it remained unchanged to PC-DOS - it just freezes, but I can still move the mouse pointer.

All other stock apps like Write work...
Crazy, isn't it?

Which are the first 64KB of RAM on the M24? Which 4 chips in Bank 0?
 
OK that's weird. I kept the PC running while I typed the above post, and the clock appeared after 4-5 minutes, but it was not ticking. Menu of the clock app (about screen) worked, also minimizing to an icon.


Is there maybe some other issue than with the RAM? CPU?

Edit: now it locked up completely.
 
Okay, found the cause for this. I tried once again to remove all device drivers one by one, and after removing CLOCK402.SYS the issue was resolved.
I believe the inital issue with a freeze and garbled screen got resolved by playing around with the jumpers, resocketing the DIPs and using contact cleaner.
Oh man, I should have read the readme of this driver way earlier, it would have saved me a lot of time.
*** DISCLAIMER *** The SOFTWARE clock driver may not work correctly with
unsupported PC6300 DOS Versions, such as DOS 5.0 or higher, or non-AT&T DOS
versions. SOFTWARE applications which load into a memory range that conflicts
with the SOFTWARE clock driver may cause unpredictable results. Any SOFTWARE
applications which continue to read the HARDWARE clock instead of the DOS
clock WILL continue to see the wrong date.
Click to expand...

Still i would love to know which DIPs of BANK0 are "the first 64KB of RAM". And does anybody know if there is an alternative to CLOCK402.SYS?
 
Last edited:
That's something I can confirm. "keyb gr" from Olivetti DOS 5 hangs on M24. I use "keybgr" from Olivetti DOS 3.30 on DOS 5 and it's fine.
 
It just freezed again while idleing in the DOS prompt, but the cursor is still blinking. No reaction on CTRL+ALT+DEL.
It drives me crazy...is there something wrong with my config maybe?



Code: 
DOS=UMB,HIGH
DEVICE=C:\DOS\USEUMBS.SYS A000-B000 C000-D000 E000-F000
DEVICE=c:\dos\DOSMAX.EXE /R+ /N+ /P-
SHELL=c:\dos\SHELLMAX.COM C:\DOS\COMMAND.COM C:\DOS /E:512 /P
FILES=40
BUFFERS=30
BREAK=ON
DRIVPARM=/D:0 /F:2
;DEVICEHIGH=C:\DOS\CLOCK402.SYS 2020
DEVICEHIGH=C:\DOS\ANSI.SYS
DEVICEHIGH=C:\DOS\SETVER.EXE
;DEVICE=C:\DOS\PALMZIP.SYS

Code: 
@echo off
rem LH C:\DOS\NE2000.COM 0x60 0x4 0x360
LH C:\DOS\MOUSE\MOUSE.COM
LH C:\DOS\SHARE.EXE/f:4096
LH C:\DOS\wd8003e.com -w 0x60 0x03 0x280 0xD000
set mtcpcfg=c:\mtcp\sample.cfg
PATH=C:\DOS;C:\WIN211;C:\WIN30
set TEMP=C:\WIN30\TEMP
C:\DOS\mode.com co80
PROMPT $p$g
cls
 
Last edited:
freakedenough said:
It drives me crazy...is there something wrong with my config maybe?
Click to expand...

If it works on a cold boot, but stops working the longer you leave the system on, then my guess is a hardware/component failure.

You might be able to isolate which component by doing basic troubleshooting: Remove everything except the floppy drive, hard drive, and video, and you should boot Olivetti DOS 3.x with a clean/empty CONFIG.SYS and AUTOEXEC.BAT, and do your testing with that. If it still locks up, you have hardware issues on the motherboard that you'll need to start troubleshooting. If it works fine for over an hour, start adding hardware and config.sys/autoexec.bat entries back ONE AT A TIME and test extensively, and maybe you'll find something interesting that can narrow things down.
 
I've done that already for sure, it didn't help.
Today I bought 18 DIP16 sockets and did the crazy 576 solder joints work and removed the old RAM, installed the sockets and new chips. I hope it was worth the 3hrs of work. With the first set of new chips (100ns -10 instead of 120ns -12) it didn't POST at all. I was already afraid that I killed it... :wow:


Right now its up for 20 minutes without a freeze. Longer than previousely.
 
Last edited:
Well, it ran in the DOS prompt for 2 hours and then in IRCjr for another 50 until I went to bed, it didn't freeze anymore. It did prior to the new RAM after like 5 minutes. I hope it is now permanently fixed. Fingers crossed.
 
Today it ran already for 4,5 hours without a freeze. I consider my issue as resolved. :bigups:
 
Last edited:
@luke.6502 I cant answer you via PM due to error "luke.6502 has exceeded their stored private messages quota and cannot accept further messages until they clear some space."

let me share the Service Manual with you. It describes the voltages on Chapter 2-48 on PDF Page 63.
https://www.retrospace.net/infoseiten/upload/files/1_m24servicemanual.rar



I assume the Service Manual will help you further with getting it back alive. 15V is really uncommon, but yes, the cable with orange and black wire goes straight to the CGA card and it has 15V.

If any questions, feel free to write me another PM after deleting some old messages ;)
 
freakedenough said:
Today it ran already for 4,5 hours without a freeze. I consider my issue as resolved. :bigups:
Click to expand...

Awesome, glad to hear it :)

But can you clarify something you wrote earlier:

With the first set of new chips (100ns -10 instead of 120ns -12) it didn't POST at all.
Click to expand...

What did you do to get past that initial issue? Put the original 120ns chips back in?
 
You must log in or register to reply here.
Back
Top