Another dead 5150 motherboard, need eproms

[fabs0]

Hello everyone,

I'm in the process of widening my horizon from vintage keyboards to vintage computers
That's why I got myself a 5150.

Unfortunately the board (64-256) is "dead"..
Of course I followed the diagnostic process on minuszerodegrees, which first lead me to a failed PSU, which I changed.
After that I changed ALL the tantalums on the board, just to make sure, nothing.
I also tried to piggyback the RAM chips, which did not help (but does not alwas work, as I read)

Observation:
- CPU and Clock Generator are warm, CG even HOT, is this normal for a "halted" system?

I really would try the Landmark Diagnostics, but have no idea how to obtain such an chip (and adapter?)

Now my question:
Is there anyone (preferably in europe, I'm located in Germany) who could provide me with a Landmark BIOS AND a 5150 BIOS (3rd rev)
I'm williong to pay good money for it
Is there something else I can try without logic probe or oscilloscope?

Thanks guys,
Fabian

 

[Exceter]

This isn't located in Germany but the seller indicates that they will ship to Germany: https://www.ebay.com/itm/154312527901?hash=item23edbe501d:g:p-YAAOSwEFNdp8wu

 

[modem7]

Hello everyone,

Welcome to these forums.

Of course I followed the diagnostic process on minuszerodegrees, which first lead me to a failed PSU, which I changed.

Good. That logic process will have proven the motherboard at fault.

After that I changed ALL the tantalums on the board, just to make sure, nothing.

Had there been any short-circuit tantalums, the aforementioned diagnostic process would have suggested such.

I also tried to piggyback the RAM chips, which did not help (but does not always work, as I read)

It was worth the try.

Observation:
- CPU and Clock Generator are warm, CG even HOT, is this normal for a "halted" system?

I powered up a good 5150 motherboard for 30 minutes. Most chips are barely warm at all. The 8088 and 8284 Clock Generator chip stand out as the warmest, although, by my standards, I would not say they are hot. I consider them 'very warm'. Temperature-by-finger is very subjective. I tied to use my infrared thermometer on them, but the results were quite varying. Using my finger as a guide, the 8284 is slightly warmer than the 8088.

The 8284's temperature will be independent of system activity. (In the 5150, it always generates clocks, of which the frequency never changes.)

Do you know for fact that the 8088 is in a HALT state? For example, maybe due to the nature of the fault, the 8088 is not halted, instead, executing (corrupted) code in a loop. If you were curious as to which, you could measure the 8088's status pins (see [here], a HALT indicated by {S2=LOW, S1=HIGH, S0=HIGH}.

I really would try the Landmark Diagnostics, but have no idea how to obtain such an chip (and adapter?)

As you may be aware, per [here], there are options that do not require an adapter. No adapter might reduce postage cost.

Now my question:
Is there anyone (preferably in europe, I'm located in Germany) who could provide me with a Landmark BIOS AND a 5150 BIOS (3rd rev)
I'm willing to pay good money for it

See [here].

I could send you some from Australia, but I expect that the postage time would be long.

Is there something else I can try without logic probe or oscilloscope?

You did not indicate swapping over chips from another motherboard, and so I assume that you have no second motherboard.

Remove the CPU and 8284 from their sockets, looking for pins that may be bent up under the chip (by a previous owner).

Try removing the RAM chips in banks 1, 2 and 3, in case of interference. Even with the third revision BIOS, screen activity is expected for bank-0-only operation.

You only have a multimeter. Very limiting, but see the multimeter comment in the image at [here]. It can provide a low (repeat: low) level of confidence.

 

[fabs0]

Wow, thanks for that lenghty reply

Had there been any short-circuit tantalums, the aforementioned diagnostic process would have suggested such.

Theye were faulty on the 12V, -12V and -5V line, so I replaced them all.

I could send you some from Australia, but I expect that the postage time would be long.

Found someone in the US who is willing to help me out, still a little nearer that Australia, but thanks for that offer!

edit: now the Netherlands, thanks pietja

You did not indicate swapping over chips from another motherboard, and so I assume that you have no second motherboard.

Remove the CPU and 8284 from their sockets, looking for pins that may be bent up under the chip (by a previous owner).

Unfortunately I do not have a second board. All I have is a Memory extention card to harvest (maybe) good RAM chips.
All chips have been reseated with deoxit.



In the meantime I was able to verify, that the CPU and Clock Generator have 4.77 Mhz on the clock pin, so I suppose they are good?
Also I noticed that I have this weird modified board mentioned here: http://minuszerodegrees.net/patch_wi...s_off_U101.jpg, does that change anything?

So for now I can only wait for the ROMS, on further inspection I could see that some ROMS have oxidized pins...

Anything else I can do while I wait?

Thanks!

 

[modem7]

In the meantime I was able to verify, that the CPU and Clock Generator have 4.77 Mhz on the clock pin, ...

You would have used your multimeter, per my earlier post. That provides only a low level of confidence.

... so I suppose they are good?

Clock Generator: Low level of confidence only.
CPU: No, in the same way that I cannot assume that my car is in working order by verifying that it has petrol in its tank.

Also I noticed that I have this weird modified board mentioned here: ...

Correcting a design defect of some sort.

... does that change anything?

No.

Anything else I can do while I wait?

A very good physical inspection.

If the ROM's do not help, then you may need to invest in a logic probe. Even that has limitations. For example, the logic probe can inform you that the Clock Generator is generating clocks, but it cannot inform you as to the frequency of those clocks. So I would call that a medium level of confidence.
But, referring to step 9 at [here], if the logic probe showed activity out of channel #1 on 8253 timer chip, you would have some confidence that the POST had started and had gone at least as far as step 9. It would be a very good level confidence if the probe showed the output being {mostly high, sometimes low}.

 

[fabs0]

Now things get serious, I got myself a cheapo scope

Do you know for fact that the 8088 is in a HALT state? For example, maybe due to the nature of the fault, the 8088 is not halted, instead, executing (corrupted) code in a loop. If you were curious as to which, you could measure the 8088's status pins (see [here], a HALT indicated by {S2=LOW, S1=HIGH, S0=HIGH}.

Status seems to be 1, 1, 1 which translates to "passive" (?)

But, referring to step 9 at [here], if the logic probe showed activity out of channel #1 on 8253 timer chip, you would have some confidence that the POST had started and had gone at least as far as step 9.

Just high on that line, no further activity. So the issue is before step 9..



My ROMS are already on their way, so hopefully it helps


Edit:
Got the landmark ROM, it's complaining about 8253 timer channels.
Have to look into that, replacement chip is already ordered.

Can this chip be piggybacked, or do I have to desolder it?

 

[modem7]

Can this chip be piggybacked, or do I have to desolder it?

The situation is the same as for RAM piggybacking. There always exists the possibility that an output on a faulty chip is going to clash/conflict with the corresponding output on the (good) piggybacked chip.

 

[fabs0]

Unfortunately the board is still not running

I replaced the 8253, but I have the same issues:

- no activity on the 8253 Channel #1
- Landmark fails for all 8253 Channel Tests
- Also the speaker does not beep but rather click, is that normal?
- errors for the keyboard test

But I managed to get rid of all memory errors reported by landmark.
ALL mostek chips in bank 2 and 3 were bad, are they known to be unreliable?

any ideas, how I can go on?

 

[modem7]

I replaced the 8253, but I have the same issues:
- no activity on the 8253 Channel #1
- Landmark fails for all 8253 Channel Tests

You have an oscilloscope, which is helpful.

Refer to the diagram at [here]. Do you measure:

• A 1.193 MHz clock signal on pins 'CLK 0', 'CLK 1', and 'CLK 2', of the 8253. Measure at the 8253, not the source of the 1.193 MHz clock. Signal is always present.
• Activity on the 8253's 'RD' and 'WR' pins. Activity is always present.
• Activity on the 8253's 'A0' and 'A1' pins. Activity is always present.
• Activity on the 8253's 'D0 through 'D7' pins. Activity is always present.
• ONLY during part of the POST, activity on the 8253's 'CS' pin. (Normally HIGH, with activity seen as bursts of LOW going pulses.)

- Also the speaker does not beep but rather click, is that normal?

The Supersoft/Landmark ROM should be beeping the speaker. A simplified diagram of the speaker circuitry is at [here]. Note how the 8253 plays a part. So fixing the 8253 related issue may fix the no-beeping issue.

- errors for the keyboard test

Expected if a known working PC/XT-class keyboard is not attached.
Is one attached, and plugged in the keyboard port, not the cassette port ?

But I managed to get rid of all memory errors reported by landmark.
ALL mostek chips in bank 2 and 3 were bad, are they known to be unreliable?

No, but maybe they were 'pushed'.
E.g. Previous owner had accidentally plugged them all in the wrong orientation, and the particular model of chip cannot tolerate that. (I have 'been there; done that'.)
E.g. Motherboard stored long term in a 'hostile' environment. (E.g. See the JUL14-43314-bobba84 line at [here].)

 

[modem7]

- errors for the keyboard test

Expected if a known working PC/XT-class keyboard is not attached.
Is one attached, and plugged in the keyboard port, not the cassette port ?

As a follow-on to the previous post, I enhanced the 8253 diagram that I pointed to.
In doing so, it reminded me that PCLK also goes to the keyboard interface circuitry !

 

[modem7]

Do you know for fact that the 8088 is in a HALT state? For example, maybe due to the nature of the fault, the 8088 is not halted, instead, executing (corrupted) code in a loop. If you were curious as to which, you could measure the 8088's status pins (see [here], a HALT indicated by {S2=LOW, S1=HIGH, S0=HIGH}.

Status seems to be 1, 1, 1 which translates to "passive" (?)

I was curious about that, and decided to see what 'status' I observed if I stopped the POST from communicating with the 8253 chip. On my good 64KB-256KB type of 5150 motherboard, I disabled the 8253 (by stopping the /CS pin going low). Using my logic probe, at switch-on of the PSU, I saw activity on the CPU's status pins, then about a second later, all three were static. Like for you, I did not see a final status of 'Halt'. Instead, I saw 'passive' !!!

To confirm that in another way, I enabled the 8253, booted to DOS, then ran DEBUG. Within DEBUG, I executed a CLI instruction followed by a HLT instruction. As expected, DEBUG froze. Then using my logic probe, I observed the CPU status of 'passive'.

I made a single-byte modification to the 5150's 10/27/82 BIOS, changing the byte at offset 5Ch to a halt instruction. Offset 5Ch is the second byte of the POST. I then used my logic analyser to observe the CPU's status pins around the time of the halt instruction. Pictured at [here] is the observation; the 'halt' status (03) is present, but only only for about 300 ns.

Any 8088 experts here know the reason why (i.e. why the status is not present for the entire period that the 8088 is in a halt state) ?

 

[fabs0]

Good News!
I got some life out of the board.

Again thanks to you modem7, I had a look at the posted 8253 diagram and checked all inputs.
Could see that there was no 1.193Mhz coming from the LS175 in U26.
Changed it and now I have the following:

The good:
- beeps instead of weird clicks
- all landmark test pass
- also the keyboard was affected by this

The bad:
- put in the original BIOS and now get beeps from POST
- one long and one short beep, which indicates an fault with the 8259 circuitry
- no screen output whatsoever

I'll now try to find which chips are involved and troubleshoot those.
But I'm getting there

 

[modem7]

I had a look at the posted 8253 diagram and checked all inputs.
Could see that there was no 1.193Mhz coming from the LS175 in U26.
Changed it and now I have the following:

The good:
- beeps instead of weird clicks
- all landmark test pass
- also the keyboard was affected by this

Odd that it fixed the keyboard issue. I can only deduce that the failure mode was such that it affected PCLK.

The bad:
- put in the original BIOS and now get beeps from POST
- one long and one short beep, which indicates an fault with the 8259 circuitry
- no screen output whatsoever

We have seen this before; that is, the Supersoft diagnostic indicates that it is 'happy' about the motherboard, but the POST in the IBM BIOS ROM is 'unhappy'. So it could be that Supersoft diagnostic is not testing something that the IBM POST does, or tests it in a different way to the POST.

- one long and one short beep, which indicates an fault with the 8259 circuitry

See that symptom at my web page at [here].

See the 'note' in that row. The note is there because we had someone report '1 long beep then 1 short beep' but as it turned out, there was another beep 15 seconds later. Is that the case ?

BTW. The '1 long beep then 1 short beep' symptom appears in steps 18 and 19 at [here].

 

[fabs0]

Odd that it fixed the keyboard issue. I can only deduce that the failure mode was such that it affected PCLK.


We have seen this before; that is, the Supersoft diagnostic indicates that it is 'happy' about the motherboard, but the POST in the IBM BIOS ROM is 'unhappy'. So it could be that Supersoft diagnostic is not testing something that the IBM POST does, or tests it in a different way to the POST.


See that symptom at my web page at [here].

See the 'note' in that row. The note is there because we had someone report '1 long beep then 1 short beep' but as it turned out, there was another beep 15 seconds later. Is that the case ?

BTW. The '1 long beep then 1 short beep' symptom appears in steps 18 and 19 at [here].

It's definitely only one long and one short beep (also what you described can only happen with the older BIOS, if I read correctly? )
Timer CHip should be fine, tried several and no issues in landmark.
So looks like I'm stuck with the 8259.

Any hints where I can start poking?
Seems like this guy is connected to everything :/

Thanks,
fabs0

 

[modem7]

(also what you described can only happen with the older BIOS, if I read correctly? )

Yes. I forgot that you have the later 5150 motherboard.

Timer CHip should be fine, tried several and no issues in landmark.

The speaker is beeping. With the IBM BIOS ROM fitted, if you are seeing the proper waveforms on the OUT0 and OUT1 pins, then yes, I agree that the 8253 chip can be considered good.

So looks like I'm stuck with the 8259. Any hints where I can start poking?

Give me some time to think about that, based on what I see in the POST source code.

 

[modem7]

The speaker is beeping. With the IBM BIOS ROM fitted, if you are seeing the proper waveforms on the OUT0 and OUT1 pins, then yes, I agree that the 8253 chip can be considered good.

Another mistake of mine. You may or may not see a waveform on the OUT0 pin.

Give me some time to think about that, based on what I see in the POST source code.

Refer to [here]. Comparing the SuperSoft ROM manual to the source listing for the IBM 5150's POST, both are doing the same testing (functionally) of the 8259. If anything, the SuperSoft ROM is doing more, because according to the SuperSoft ROM manual, SuperSoft is testing more than one of the 8259 registers (the IBM 5150's POST tests only the IMR). But we do not have the source code for the SuperSoft ROM, and so cannot do a low level comparison to figure out why the Supersoft ROM passes and the IBM 5150's POST fails. Maybe that source code would reveal a bug that results in the 8259 not being tested per the description in the manual.

Another possible explanation for the 8259 tests in the SuperSoft ROM passing, and failing in the IBM 5150's POST not, is a corrupt IBM BIOS ROM. The corruption would have to be multiple bits 'flipping', and such that the 8-bit checksum of the ROM remains being 00. Seems unlikely, but still, a possibility, and very easy to test. Did you get a replacement IBM BIOS ROM ?

When the IBM 5150's POST is executing, are you seeing activity on the 8259's address pin, data pins, /RD pin, /WR pin, and /CS pin?

Do you have a known working 8088 to swap in?

Do you have a known working 8259 to swap in?

 

[modem7]

I have enhanced 'TEST 18' and 'TEST 19' at [here].

 

[fabs0]

Ok, now it gets weird:
It works! But only every 20 restarts.

I get the usual long beep, short beep.
But if i powercycle the machine a lot, at some point it boots into cbasic.

I observe the pins on the 8559 and can see activity on all pins.
When it does not boot, I get activity until the beeps and then nothing.

I tried pushing and bending on different things while power on, but nothing makes a difference, it's random.

One time it booted I could see POST errors "1000 201" and "301"

Will continuw poking.

 

[fabs0]

UPDATE:
found it.
there was a damaged trace where I socketed the 8253.
fixed it and now the board starts reliably!

Only issue left is a 301 error.
I have a XT compatible board attached but maybe it's not compatible?

 

[fabs0]

Sorry for triple post, but I'm not allowed to edit...

Keyboard is also fixed!
It was the board, opened it up to find all controller chips dislodged.

Next part are the floppies and the memory expansion card, wish me luck