5150 - No Post, but runs Rudds Diag just fine.

[dabone]

I've got a 5150 64-256k board, that powers up, but nothing, leave it for forever and not a beep.

I burned a copy of Rudds Diag Rom and stuck in on a adapter and the machine fires right up.



A couple of weird things...
One, Floppy controller ALWAYS passes, even when there is no controller in the system.
Two, Switch 2 is wonky.. I can set all off, and all on and it reads it correctly, but when I set sw2 to 640k for the ram card, (Removed in this pic),
But when I set it to the 640k setting of 10110 it shows 10111.

I also tried burning a new copy of the 10/27/82 bios (That's the one installed in the machine) in my eeprom adapter, but still same no post.

Ram test is working, I think, I put a bad ram card in and it failed at 318K, but 2 other ram cards test just fine up to 640k.

I can also try supersoft, but haven't gotten around to it.

 

[modem7]

I've got a 5150 64-256k board ...

I brought out an IBM 5150 motherboard of type 64KB-256KB, and then substituted the the 2022-12-12 version of Ruud's diagnostic ROM in place of the IBM BIOS ROM.
The only card present is an MDA.

A couple of weird things...
One, Floppy controller ALWAYS passes, even when there is no controller in the system.

You have discovered a bug. I see the same.

Two, Switch 2 is wonky.. I can set all off, and all on and it reads it correctly, but when I set sw2 to 640k for the ram card, (Removed in this pic),
But when I set it to the 640k setting of 10110 it shows 10111.

Switch 5 in switch bank SW2.

You have discovered a bug. I see the same. Experimentation reveals that the shown state of switch 5 always mimics the state of switch 1.

I also tried burning a new copy of the 10/27/82 bios (That's the one installed in the machine) in my eeprom adapter, but still same no post.

Presumably, you have very good confidence in that 'burned' copy.
And presumably, you have no other 5150 motherboard in which to test it.

I can also try supersoft, but haven't gotten around to it.

Do that.

I burned a copy of Rudds Diag Rom and stuck in on a adapter and the machine fires right up.

It could be that the IBM BIOS ROM is testing something that Ruud's diagnostic ROM does not.

I think we can exclude a ROM addressing issue. Both the IBM BIOS ROM and Ruud's diagnostic ROM are 8 KB in size, occupying the same address space.

Do you have one of [these] ?

 

[Timo W.]

I've got a 5150 64-256k board, that powers up, but nothing, leave it for forever and not a beep.

I hate to ask, but what is "forever"? With 640k, the RAM test at power-on takes *very* long and during that, nothing happens. The 5150 does not show any text during RAM test. The only thing you might see is a blinking cursor.

 

[dabone]

I hate to ask, but what is "forever"? With 640k, the RAM test at power-on takes *very* long and during that, nothing happens. The 5150 does not show any text during RAM test. The only thing you might see is a blinking cursor.

I turned it on and left it for around 20 mins while I cooked some dinner.

 

[dabone]

And, modem7, I do have another 5150, and I'll double check the new bios on it, but I've used the same adapter on another 5150 before.
(Currently I have 4 5150s.)

 

[dabone]

I do not have a Parallel port post code reader. I do have a ISA post card, But haven't bothered with it, I didn't think these early bios machines supported it.

 

[dabone]

Ok, so tested the original bios, and the multirom adapter I'm using for testing in a working 5150, both worked fine.
(Here's the Multikernal adapter I'm using, since the 5150s rom sockets are compatible with the 64s roms sockets. Rom adapter.

Supersoft looks good starting out, but corrupts the screen with the CGA ram test, 2 different CGA cards that I've been using for years show this behavior.



All the tests except for Floppy read pass, which is expected because there are no drives attached to the controller.
For Supersoft I removed the extra ram board again, and configured SW2 for 256k.
Using the original bios, I still get no post with this config.

 

[dabone]

I dug out a Hercules clone card and a 5151 monitor, and still get the same kind of corruption after the mda test.




Power supply does need a refurb, when the floppy drives were connected, I get memory errors, So I'm using a known working psu in it currently (XT 135 watt)

 

[dabone]

And, why does U28 Pass with a MDA card??? There's still nothing in the socket.

 

[dabone]

Ok, it's run 70 passes with Supersoft rom, most times U28 passes, randomly it fails. So there's something screwy.

 

[modem7]

I do not have a Parallel port post code reader. I do have a ISA post card, But haven't bothered with it, I didn't think these early bios machines supported it.

The parallel port POST board can be used if I need to develop custom code for someone with a 5150/5160.
For example, insert {spit out number XX to parallel port POST board} at various points in the IBM 5150 BIOS, in order to see where in the early stage it is stopping.

( An ISA POST card is usually unsuitable for that task - see [here]. )

 

[modem7]

And, why does U28 Pass with a MDA card??? There's still nothing in the socket.

Ok, it's run 70 passes with Supersoft rom, most times U28 passes, randomly it fails. So there's something screwy.

With the SuperSoft diagnostic, I occasionally see the U28 test pass even there is nothing in the U28 socket. I ignore that, putting it down to a peculiarity of the diagnostic. I expect that the diagnostic is simply calculating the 8-bit checksum of the F4000 to F6000 address space, and if the result is 00, displaying "PASSED". Ideally, the author should be checking for 55 and AA at the start of the address space first.

 

[MicroCoreLabs]

You sent me a boxful of free vintage books, so let me return the favor and send you one of my MCL86+ boards. You would just need to purchase a Teensy 4.1 microcontroller board.

It may be overkill for debugging your motherboard issue, but it could quickly isolate exactly where the fault is. We can load any/all BIOSes and test ROMs and can have it perform isolated reads and writes to any address or range that we wish. It uses the Arduino Serial.print() to display the results to your nearby PC.

Here is a recent project which uses the MCL86+ to first configure all of the motherboard ICs and then runs a 68000 emulation which accesses the IBM motherboard I/O. We could skip all of the emulation stuff and just configure the ICs (or try to), then write and read from them until we see where the bit error(s) are. The board sits in the 8088 CPU slot so it has access to everything...

IBM PC 8088 replaced with a Motorola 68000

I was wondering what the IBM Personal Computer would have been like if they had chosen the Motorola 68000 instead of the Intel 8088, so I used my MCL86+ to emulate the 68000 and find out! The MCL86…

microcorelabs.wordpress.com

 

[modem7]

Supersoft looks good starting out, but corrupts the screen with the CGA ram test, 2 different CGA cards that I've been using for years show this behavior.

I dug out a Hercules clone card and a 5151 monitor, and still get the same kind of corruption after the mda test.

Interesting that it is mostly bit 5 going HIGH.


I have seen 'corruption' with some third-party cards. That was years ago. These days, I stick to IBM video cards.

With an AST 3G card that I have, things are opposite: 'corrupted' display to start with, which comes good after the video test.

 

[modem7]

Do you have a logic probe and/or oscilloscope ?

For example, with the IBM BIOS ROM in place, you could see if the POST is getting as far as step 9 at [here].

 

[dabone]

Do you have a logic probe and/or oscilloscope ?

For example, with the IBM BIOS ROM in place, you could see if the POST is getting as far as step 9 at [here].

I'll dig those out tomorrow.

 

[dabone]

I did dig out the logic probe, and the CPU is being halted, but I'm not sure how to start troubleshooting from that link you gave me.. example, port b of the 8255, what should they be after the bios inits them?

 

[modem7]

And, modem7, I do have another 5150, and I'll double check the new bios on it, but I've used the same adapter on another 5150 before.
(Currently I have 4 5150s.)

Did you do that test; verify the programmed EPROM, and adapter?
I included "adapter" because for all I know, you have a few, and pulled one out that you have never used before, and it is faulty.

Or perhaps just bring over a known good BIOS ROM from one of your other 5150's.

We just need to ensure that whatever you are substituting for the {original BIOS ROM is under suspicion}, that it is a proven/known good substitute.

You may even want to try out the {original BIOS ROM is under suspicion} in one of your other 5150's.

I really think that this needs to be 'knocked on the head' before much more effort is expended.

I did dig out the logic probe, and the CPU is being halted, ...

How did you establish that? I used to think that one could just look at the three status lines on the 8088, but then I discovered otherwise. See the warning text at the far right of the diagram at at [here]. At [here] is a logic analyser capture.

Or is it that you don't see pulses on pin 13 of the 8253 (i.e. the step 9 that I referred to), and assumed from that that the CPU must have been halted?

... but I'm not sure how to start troubleshooting from that link you gave me..

You weren't meant to. I just gave you one example of how a logic probe and/or oscilloscope can sometimes be used in diagnosis. Now we know that you have a logic probe.

But confirm for us that you are not seeing pulses on pin 13 of the 8253.

... example, port b of the 8255, what should they be after the bios inits them?

You will be referring to 'step 6'. The source listing indicates FC hex.

 

[dabone]

""Did you do that test; verify the programmed EPROM, and adapter?""
Yes, I grabbed a different 5150 and tested the current eprom adapter, and the original u33 bios rom, both worked correctly in the other 5150 (Both have a 64-256k board)

""How did you establish that?"" (Cpu Being Halted)
I did a quick glance with the logic probe, and all address line activity on the cpu stops after a small initial burst. So my bet is that the bios is failing a check and halting.

I didn't go any farther last night.. Sleep was needed.

On the 8253 Checking pin 13 this morning, yes it's getting to step 9. But pin 10 is always high, so failing before step 19?

""It was stated before that every time that DMA channel 0 is 'active' (i.e. a refresh cycle), the RAS pin of all motherboard RAM chips is pulsed. So, for example, row 53 is being refreshed at the same time on all of the motherboard's 4116 RAM chips."" Step 11...

I also have RAS activity to the ram, (I don't really have a space for the scope near this without a lot of rearranging, so logic probe only at the moment.)

 

[modem7]

On the 8253 Checking pin 13 this morning, yes it's getting to step 9. But pin 10 is always high, so failing before step 19?

And because you are not hearing 'beep 1 long then 1 short', it must be before step 18 (assuming the motherboard has only one fault). An earlier post informs me that you have a speaker connected.

If we assume that the CPU is purposely being halted, looking between steps 9 and 18, failure of steps 10 (TEST DMA CONTROLLER) and 13 (BASE 16 KB RAM TEST) result in the POST halting the CPU.

I rule out 'IF REQUESTED, LOAD DIAG. CODE'. Even with faulty keyboard interface circuitry, I cannot see it being faulty in a way in which 65h is returned to the CPU.

You are seeing RAS activity (before all activity stops). That could be RAS as a result of RAM refresh, or RAS as a result of RAM access by the POST. It doesn't really matter - it informs us that the POST is going past step 10.

That leaves (at this time), test 13 (BASE 16 KB RAM TEST) as the prime suspect.

I will create modified 10/27/82 IBM BIOS ROM's for you to try.

For now, can I get you to do two things:
1. In case of interference, remove all RAM in the socketed banks. Yes, the issue at [here] comes into effect, but even so, something is expected on-screen on a good motherboard.
2. Because you lack a parallel port POST code reader, try TEST5083 at [here]. If you see the expected result, it means that I have a way of signaling to you in code.