• Please review our updated Terms and Rules here

IBM 5170 Boot Troubles

JonathanHunt

Member
Joined
Feb 7, 2020
Messages
19
I have an IBM 5170 type 339 I'm trying to get running. This has the type 3 motherboard at 8mhz with 512k, and has the 11/15/85 BIOS.

Power supply and hard drive spin up, but no video init and no sound from speaker.
Remove dead CMOS battery and inspect, no visible corrosion anywhere, all looks good. No battery installed but based on schematic it should still boot
Checked power supply voltages, everything AOK.
Installed POST Card, consistently get 03, (Verify shutdown byte in 146818 RTC chip, after 80286 register test and ROM checksum OK)
Pulled all cards and 80287, same result.
Burned and installed Landmark Diag ROMS from minus zero for AT.
Never initializes video, but throws many audio codes:

8254 Timer channel 0, 1, 2
DMA Controller 1,2
8042 Parity Detected

and then several cannot initialize monitor codes

Pulled the RAM off the motherboard, tested it all with a DRAM tester, passed with no problems.
With no ram on board, Landmark throws the same codes plus 16k critical memory region (understandable with no RAM)

If I let the landmark run, after a 2 runs the computer loses the ability to generate tones and just generates clicks from the speaker instead, one for on and one for off.
Once its in this state, even if I power cycle, it will still do this. If I let it rest for 10 mins or so it will go back to normal tone generation.

Anyone seen symptoms like this in the past? Any tips on how I can pin down the issue? Could it actually be the RTC chip?

Thanks!

JH
 
No battery installed but based on schematic it should still boot
True.

Burned and installed Landmark Diag ROMS from minus zero for AT.
Never initializes video, but throws many audio codes:

8254 Timer channel 0, 1, 2
DMA Controller 1,2
8042 Parity Detected

and then several cannot initialize monitor codes
Per [here], we do expect that some video cards will not work with the AT version of the Supersoft/Landmark Diagnostic ROM's.

The Supersoft/Landmark Diagnostic ROM's can sometimes be misleading. So take the audio codes 'with a grain of salt'.

If I let the landmark run, after a 2 runs the computer loses the ability to generate tones and just generates clicks from the speaker instead, one for on and one for off.
Once its in this state, even if I power cycle, it will still do this. If I let it rest for 10 mins or so it will go back to normal tone generation.
Which suggests that a chip somewhere fails when warm/hot.

Anyone seen symptoms like this in the past? Any tips on how I can pin down the issue? Could it actually be the RTC chip?
If the MC146818 RTC chip (or one of its support chips) is 'dead', then we expect that the Supersoft/Landmark Diagnostic ROM's will fail on the PROTECTED MODE CPU and REAL TIME CLOCK INTERRUPT and CMOS RAM TEST tests.
You can see that at [here].
Could you have misinterpreted the audio codes ?

Something else to check: Refer to the circuit at [here]. When the motherboard is powered on (with no battery connected), do you measure about 5 volts on pin 24 (Vcc/Vdd) of the MC146818 RTC chip ?

temp_rtc7gj047nhg9526.png
 
It was indeed the case that the EGA Video 7 Card set to CGA mode would not initialize. I replaced it with a known good original IBM Color Graphics Adapter and the Landmark ROMS did initialize the video, but the display was garbage as shown in the image below after 15 or so seconds. The pattern is consistent and scrolls right for a time, then slows down and scrolls left. The pattern is random at each boot but does not change as the tests run. I also tried this with a known working clone CGA card - it displays nothing except some brief flashes of white when the display is finally intialized

20210927-172556.jpg

I have also recorded audio of the beeps from the ROM to see if perhaps I am misinterpreting the codes. The on-board RAM has been reinstalled for this recording.

https://drive.google.com/file/d/1v29s8GjOyhwAbd0Rv9p6Tirk7u4jdZfn/view?usp=sharing

I have also checked the voltage and all of the address pins for the MC146818. The voltage is good at the chip, 5.13v and there is normal looking activity on all data lines which go to the chip, and the oscillators look good.

I'm really not sure what to make of it at this point. There's clearly something wrong on the bus, but it's not bad enough to prevent the CPU from running code out of ROM or generating tones (despite that it says the timer channel 2 is bad - it has to generate the tones so its working at least partly).

Any additional thoughts or things to try?

Thank you!

Jonathan Hunt
 
I read three symptoms:

1. With 11/15/85 IBM BIOS ROM's fitted, consistently see POST code 03 ("Verify shutdown byte in 146818 RTC chip") on a POST card.
2. With Supersoft diagnostic ROM's fitted, diagnostic is unable to initialise a known working IBM CGA card (a card known to work with the AT version of the Supersoft diagnostic ROM's).
3. With Supersoft diagnostic ROM's fitted, after two runs, motherboard appears to no longer operate at all. Appears to be heat induced.

There could be one, two, or three causes.

You may be able to track down the cause of the heat induced symptom by way of electrical non-conductive freeze spray ([example]). Even forced air can cool things quickly. I remember that one member found the CPU on their IBM 5170 motherboard to be the cause of a heat induced symptom.

I have also recorded audio of the beeps from the ROM to see if perhaps I am misinterpreting the codes. The on-board RAM has been reinstalled for this recording.
I opened up the MP3 file using Audacity - easier to count beeps.
Then looking at pages 40 and 41 (beeps for AT version) of the Supersoft manual, I came up with:

6/1 = Cannot initialize monitor
6/2 = ?????????
6/2 = ?????????
5/9 = Cannot initialize monitor

1/4 = 8254 timer chan 0
1/5 = 8254 timer chan 1
1/6 = 8254 timer chan 2
1/7 = 8237A DMA controller 1
1/8 = 8237A DMA controller 2

???
???

2/1 = 16K critical memory region
3/8 = EGA/VGA memory (<-- expected because CGA card fitted)

1/4 = 8254 timer chan 0
1/5 = 8254 timer chan 1
1/6 = 8254 timer chan 2
1/7 = 8237A DMA controller 1
1/8 = 8237A DMA controller 2

1/10 = 8042 parity detected

2/1 = 16K critical memory region
3/8 = EGA/VGA memory (<-- expected because CGA card fitted)

Etc.


I see '16K critical memory region'. Maybe you didn't refit the motherboard RAM properly.
Interesting that the '8042 parity detected' did not appear until what I assume is the second pass.

It all does suggest some kind of bus problem (I/O ???)

I have also checked the voltage and all of the address pins for the MC146818. The voltage is good at the chip, 5.13v and there is normal looking activity on all data lines which go to the chip, and the oscillators look good.
Expect too to to see activity on the /CE pin (Chip Enable) when the IBM POST or Supersoft diagnostic interacts with the MC146818.

Have you tried removing all cards except for the video one, just in case a card is interfering in some way with bus operation ?

Maybe identify-then-fix the cause of the heat induced symptom, then see if the other symptoms remain.
 
Back
Top