Supersoft Rom showing failures

[oldmanwanderlust]

I programmed a W27E257-12 with the Supersoft ROM Diagnostic to troubleshoot some 5150's. The problem I am seeing is that the ROM is failing multiple test on a known good board. I am at a loss for what the issue could be. I am using a 2364 ROM adapter and everything pins out good. I've attached a pic and hoping for some suggestions.

 

[Svenska]

It is possible that the system is not as "known good" as you think. I've been bitten by a system where almost everything things worked fine, even though there was major damage on the board (but that was not an XT-class system).

Have you tried [Ruud's Diagnostic ROM]? Both modem7 (its current maintainer) and Ruud are members in this forum. There is also the [xtramtest] ROM, which may be useful.

 

[oldmanwanderlust]

I ran both the Rudd's ROM and it halts the test after the timing test. As for the xtramtest I found some interesting results as you can see in the pic. I'm not getting any errors in the bit test however the march test is giving errors of FF, BB, FB, FE which with my understanding of this tool does not relate to a particular RAM chip. I'm wandering if there is some logic chips that support the RAM are acting up. If so, I'm very confused as to why the system functions properly with so many errors.

 

[Svenska]

I'm very confused as to why the system functions properly with so many errors.

My first attempt at upgrading a non-compatible DOS system from 256K to 640K ended with me reporting-but-not-enabling the new memory. Turns out that none of the programs I tested with used more than 256K, so everything was perfectly fine even with 60% of memory not working.

You may have a problem with the address or data bus logic. A faulty bit reading the PIT ports may not prevent the BIOS from configuring the PIT at boot, but will show up in diagnostics. A broken bus driver chip may cause conflicts for only a few ranges. A few memory segments accidentally aliasing will not be detected by simpler memory tests, and very little software uses all memory. And so on.

 

[modem7]

I programmed a W27E257-12 with the Supersoft ROM Diagnostic to troubleshoot some 5150's. The problem I am seeing is that the ROM is failing multiple test on a known good board. I am at a loss for what the issue could be. I am using a 2364 ROM adapter and everything pins out good. I've attached a pic and hoping for some suggestions.

So, before trying the diagnostic ROM out on the problem 5150's, you decided to first try it out on a known-good 5150.

In the Supersoft ROM diagnostic, let us look at just the failure the 'U36 8255 PARITY DETECTED' test. That is equivalent to the 'RAM parity error latches' test in Ruuds Diagnostic ROM. That failure is indicating: I disabled the two 'RAM parity error' latches, but one (or both) are reporting as set instead of clear. With the IBM BIOS ROM fitted in the subject known-good 5150 motherboard, that should result in a PARITY ERROR displayed on-screen, but you are not seeing that.

Looking at the tests that failed in the Supersoft ROM diagnostic, it may be I/O related. For example, the 'U35 8237A DMA CONTROLLER' test is simply an I/O read-write test of the first 8 registers in the 8237A DMA controller chip.

So my initial thought is that in the process of swapping out the IBM BIOS ROM for a diagnostic ROM, something else was changed as well, something that negatively impacted on the ability of the diagnostic ROM's to do their job.

Right now, with the Supersoft ROM diagnostic failing the way it is, if the only thing you do is swap out the Supersoft ROM diagnostic for the IBM BIOS ROM, does the 5150 motherboard still behave as expected, or are there errors.

 

[oldmanwanderlust]

So, before trying the diagnostic ROM out on the problem 5150's, you decided to first try it out on a known-good 5150.

I suppose I should clarify my original post. I originally used the diagnostic ROM on a problem main board. When I saw the numerous tests failing, I thought there was an issue with the ROM I programmed. I decided to put the ROM in a known good system to verify it was functioning as designed. To my surprise I had the same failure at the timer test. This was when I programmed another ROM and verified the adapter was working by toning out the contact points and got the same test failures.

In the Supersoft ROM diagnostic, let us look at just the failure the 'U36 8255 PARITY DETECTED' test. That is equivalent to the 'RAM parity error latches' test in Ruuds Diagnostic ROM. That failure is indicating: I disabled the two 'RAM parity error' latches, but one (or both) are reporting as set instead of clear. With the IBM BIOS ROM fitted in the subject known-good 5150 motherboard, that should result in a PARITY ERROR displayed on-screen, but you are not seeing that.

That is correct however I don’t really use it for running intensive software such as games. Basically, just use it for QBasic programming, testing ISA cards, and floppy / MFM hard drives. I do not get any parity errors with the IBM BIOS installed though.

Looking at the tests that failed in the Supersoft ROM diagnostic, it may be I/O related. For example, the 'U35 8237A DMA CONTROLLER' test is simply an I/O read-write test of the first 8 registers in the 8237A DMA controller chip.

So my initial thought is that in the process of swapping out the IBM BIOS ROM for a diagnostic ROM, something else was changed as well, something that negatively impacted on the ability of the diagnostic ROM's to do their job.

I am not aware of anything else that could have been changed. The only other thing I changed was remove the RAM expansion card in case it was causing an issue and set SW2 accordingly. This made no difference in the test. I checked all the DIP switches just to make sure I didn’t accidently bump one and change a switch setting.

Right now, with the Supersoft ROM diagnostic failing the way it is, if the only thing you do is swap out the Supersoft ROM diagnostic for the IBM BIOS ROM, does the 5150 motherboard still behave as expected, or are there errors.

I reinstalled the IBM BIOS and the RAM expansion card. Booted the system with no errors. Ran a couple test with Checkit 3.0 and they passed. Played a game of Gorilla (LOL) with no issues. Motherboard behaved as expected with no errors or crashes.

 

[modem7]

I suppose I should clarify my original post ...

At least two 5150 motherboards, one considered known-good, and one or more faulty/suspect.

Diagnostic ROM test configuration is: Power supply + 5150 motherboard (fitted with diagnostic ROM) + video card. Is that correct?

Confirm for us that this known-good motherboard has not been performance enhanced. One enhancement that is okay for the diagnostic ROM's is a V20 CPU in place of the 8088 CPU. Speeding up the motherboard beyond a certain point can cause the failure of certain tests in the diagnostic ROM's.

The only other thing I changed was remove the RAM expansion card in case it was causing an issue and set SW2 accordingly.

Note that Ruuds Diagnostic ROM (RDR) and the SuperSoft ROM do not need SW2 to be changed.

I reinstalled the IBM BIOS and the RAM expansion card. Booted the system with no errors.

Just to rule it out, does the 'failing' behaviour of RDR change if the RAM expansion card is added? I.e. Making sure that in the switch between the IBM BIOS ROM and RDR, that only one variable is changing (the ROM).

It may seem like that is an illogical thing to ask, but I remember someone here who reported a non-functional computer, that worked when an ISA POST card was added, and reverted to non-functional when the card was removed. It was if the POST card was somehow stabilising the address bus and/or data bus.

 

[oldmanwanderlust]

At least two 5150 motherboards, one considered known-good, and one or more faulty/suspect.

Correct.

Diagnostic ROM test configuration is: Power supply + 5150 motherboard (fitted with diagnostic ROM) + video card. Is that correct?

Correct

Confirm for us that this known-good motherboard has not been performance enhanced. One enhancement that is okay for the diagnostic ROM's is a V20 CPU in place of the 8088 CPU. Speeding up the motherboard beyond a certain point can cause the failure of certain tests in the diagnostic ROM's.

No performance enhancements. Original 8088 Intel CPU. No 8087 coprocessor on board. All original hardware and there are no bodge wires on the board.

Note that Ruuds Diagnostic ROM (RDR) and the SuperSoft ROM do not need SW2 to be changed.


Just to rule it out, does the 'failing' behaviour of RDR change if the RAM expansion card is added? I.e. Making sure that in the switch between the IBM BIOS ROM and RDR, that only one variable is changing (the ROM).

It may seem like that is an illogical thing to ask, but I remember someone here who reported a non-functional computer, that worked when an ISA POST card was added, and reverted to non-functional when the card was removed. It was if the POST card was somehow stabilising the address bus and/or data bus.

Yes. RDR fails the test at the same point regardless if the expansion card is in or out of the mainboard.

 

[modem7]

I cannot see how a mis-programmed W27E257, or an incompatible 2364 ROM adapter, could cause the symptoms.

I see that you have switch 2 on SW1 set to OFF, when it should be ON if no 8087 is fitted, but that will not cause the symptoms.

And the second 5150 motherboard behaves exactly the same. Correct?

 

[oldmanwanderlust]

I cannot see how a mis-programmed W27E257, or an incompatible 2364 ROM adapter, could cause the symptoms.

I agree. I did read the ROM back and compared the MD5 with the version here and they match.

I see that you have switch 2 on SW1 set to OFF, when it should be ON if no 8087 is fitted, but that will not cause the symptoms.

I verified all the SW settings and changed accordingly.

And the second 5150 motherboard behaves exactly the same. Correct?

Yes, I actually have another known good main board that I tested with RDR and the results were the same. See attached photos.

 

[oldmanwanderlust]

****UPDATE****
I figured out the issue as to why the RDR kept failing the test. Well it turns out that my eyes and ears are not as good as they may seem. With the adapter that I got for whatever reason the Output Enable pins on the ROMS are not connected together. I bridged the pads on the board and now all is good!!! Somehow I missed this when I toned out the pins making sure there were no bad solder joints. Thanks for all your help! It is truly appreciated! Now on to troubleshooting some dead boards...

 

[modem7]

****UPDATE****
I figured out the issue as to why the RDR kept failing the test.

Thank you for the update. I am going to put on my to-do-list, the activity of working out how the non-joined enable pins resulted in the diagnostic ROM's behaving the way they did.

I see your particular adapter at [here]. It is like the '23XX Adapter' described in the 'Warning #2' section of the web page at [here].

EDIT: To help others down the track, I added your particular adapter to the web page (in the 'Warning #2' section). A page refresh in your browser may be required.