IBM XT 5160 POSTs but won't boot

[seanneko]

I've got an IBM XT 5160 which is having an issue booting.

It was working perfectly one day but then the following day no longer worked without anything having been changed.

It powers on and successfully completes the memory test, then does a seek on the floppy drive. It then goes to a blank screen with a flashing cursor where nothing further happens. It doesn't even attempt to read either the floppy drive or hard drive. Occasionally it will boot into BASIC, but most of the time it will sit on the blank screen forever.

I've removed all expansion cards other than video and floppy controller. I've also replaced both the floppy controller and floppy drive with known working ones from my 5150. The DIP switches are all configured correctly. I've burnt a new copy of the BIOS onto a couple of EEPROMs and installed it in the motherboard. None of these things made any difference. It still had the exact same symptoms.

I'm pretty confident that the problem is on the motherboard given that I've replaced everything else. What sort of things could cause this issue?

 

[VintageVic]

have you ruled out issues with power supply?
Is the hard drive spinning - maybe it will give a hint ?

Did you try the burned bios eepproms on another motherboard just to make sure they are ok?

 

[seanneko]

The PSU is putting out good voltage levels so everything there is all good.

I don't have another 5160 to test the BIOS on, but since it did run through the memory test I'm assuming that it's probably okay. I also tried the diagnostic ROM from this page:

http://minuszerodegrees.net/supersoft_landmark/Supersoft Landmark ROM.htm

The diagnostics ran correctly and all tests passed with flying colours. Still has the same boot problem when going back to the standard BIOS though.

 

[modem7]

I've got an IBM XT 5160 which is having an issue booting.

In case relevant for later, which motherboard: 64-256KB or 256-640KB ?
In case relevant for later, which video card ?

I'm pretty confident that the problem is on the motherboard given that I've replaced everything else.

Well, except for video card and power supply.
It is hard for me to believe that the power supply could cause the particular symptom.
I do not think that you should rule out the video card just yet.

It powers on and successfully completes the memory test, then does a seek on the floppy drive.

At [here] is a breakdown of the 5160's power-on self test (POST).

Your "seek on the floppy drive" will correspond to the 'DISKETTE ATTACHMENT TEST' action.

It then goes to a blank screen with a flashing cursor ...

Possibly a result of the 'CLEAR SCREEN' action.

Before 'CLEAR SCREEN' is 'LOOP POST'. If you set switch 1 on motherboard switch block SW1 to the ON position, does the POST run in a loop ?
If it does, that would suggest to me that your "It then goes to a blank screen" is a result of the 'CLEAR SCREEN' action.

Looking at what follows the 'CLEAR SCREEN' action, and there is not much, the 'ENABLE NON MASKABLE INTERRUPTS' may be where things are going wrong. For diagnostic purposes, it is possible to stop that action by grounding a certain pin of a certain motherboard chip. Do you have the means to do that if we tell you which pin/chip ? If with the pin grounded, the motherboard does the 'DO BOOTSTRAP' action, then we have narrowed the problem cause to a non-maskable interrupt related one.

So, four questions asked.

 

[seanneko]

In case relevant for later, which motherboard: 64-256KB or 256-640KB ?
In case relevant for later, which video card ?

64-256KB motherboard with 256KB installed.

It currently has an IBM CGA card. I forgot to mention in my first post, but I have swapped it out for both a clone CGA card as well as a Hercules card. The fault still remained regardless of which video card was installed.

Possibly a result of the 'CLEAR SCREEN' action.

Before 'CLEAR SCREEN' is 'LOOP POST'. If you set switch 1 on motherboard switch block SW1 to the ON position, does the POST run in a loop ?

I just gave that a shot. Correct, it runs the POST in a loop now when switch 1 is turned on.

Do you have the means to do that if we tell you which pin/chip ? If with the pin grounded, the motherboard does the 'DO BOOTSTRAP' action, then we have narrowed the problem cause to a non-maskable interrupt related one.

I'm reasonably handy with a soldering iron so I should be able to ground the pin. Which one should I be looking at?

Thanks for the help.

 

[pinkdonut666]

it is also "possible" that it is still a ram issue. on my Tandy 1000TL, the system woulkd actually pass its POST, RAM check wouldn't give any memory errors, but the machine was flaky and unreliable. Crashing seemingly at random.

I ended up running Check-IT (i was lucky the system would still boot) and only running check-it did it show me i had LOTS of memory errors. ended up replacing an entire bank of 41256 ram ICS and finally the system behaved normally.

all I'm getting at, just because the POST thinks the ram is okay, it 'might' not be. The ram chips in these old machines are EXTREMELY cranky in their old age.

 

[modem7]

I'm reasonably handy with a soldering iron so I should be able to ground the pin.

What I use is like what is shown at [here], except that there is a hook at each end. One hook onto the ground pin of the IC, and the other hook onto the target pin.

Which one should I be looking at?

Refer to the diagram at [here], in particular, the U87 AND gate at the right. The 'ENABLE NON MASKABLE INTERRUPTS' action of the POST takes the bottom input leg of U87 HIGH, enabling the output of the U72 OR gate through to the NMI line of the CPU. So one of the pins that can be tied to ground to achieve the test is the bottom input leg of U87, and that pin is pin 9.

 

[seanneko]

I tried grounding pin 9 of U87, and... no change It still gets stuck on the same flashing cursor.

I was curious if I could get it to boot into BASIC, so I pulled the floppy controller out and powered on. It of course gave me a 601 error, but after pressing F1 it successfully booted into BASIC. I can get it to repeat this over and over. Correct me if I'm wrong, but this means that it's completing POST but for some reason failing during the bootstrap.

I then put back in the MFM controller but left the floppy controller out. When it attempts to perform the bootstrap the HDD LED lights up for a second (but the drive doesn't appear to make any seeking noises), then the LED goes out and it again sits on the flashing cursor screen forever. It doesn't enter BASIC.

It seems like the problem is on the drive and/or controller, but as I said I've swapped both out with known good ones, so I don't think that's the case.

 

[VintageVic]

have you tried to remove hard drive controller and boot with floppy only ?
does it still behave the same ?

edit:
ah yes, in your first post you say that you just did that.

 

[modem7]

I was curious if I could get it to boot into BASIC, so I pulled the floppy controller out and powered on. It of course gave me a 601 error, but after pressing F1 it successfully booted into BASIC. I can get it to repeat this over and over. Correct me if I'm wrong, but this means that it's completing POST but for some reason failing during the bootstrap.

I draw the same conclusion.

I then put back in the MFM controller but left the floppy controller out. When it attempts to perform the bootstrap the HDD LED lights up for a second (but the drive doesn't appear to make any seeking noises), then the LED goes out and it again sits on the flashing cursor screen forever. It doesn't enter BASIC.

Note that Cassette BASIC is not some kind of failure fallback for all booting code. For example, when a DOS boot floppy is being booted from, if the boot code finds COMMAND.COM missing, execution is not then passed to Cassette BASIC.

Some things that are common between the floppy 'system' and the hard drive 'system', and are not required for the POST, are:
* +12V. An unstable +12V ?
* ISA bus

The hardware tests done by the POST, and by the Supersoft/Landmark diagnostics are basic. The idea is that, if needed, you execute 'proper' diagnostics after booting to DOS.

As pinkdonut666 suggested, a possibility is a RAM related problem (e.g. RAM addressing). During the POST, RAM usage is minimal. The interrupt vectors are created, and the BIOS Data Area (BDA) written to. All that is in the very low RAM addresses. When DOS starts to boot and load, RAM is starting to be used seriously.

 

[seanneko]

I checked the PSU. It's giving me a steady 12.3v and 5.15v. They don't fluctuate at all during the boot process.

I've also done some testing with the RAM. I removed bank 3 and powered on. No change. I also removed bank 2. No change. Removed bank 1. No change. Then I removed bank 0 and reinstalled 8x spare 64K chips I had lying around, and 1 of the ones I removed during testing to give the full 9 chips for bank 0. Again, no change.

I'm guessing my computer was a Friday afternoon special. The RAM all looked as though it was factory installed, but I came across a good 10 or so bent pins that obviously happened during installation. The pins would have still been making contact since the computer was working up until recently, it just looked pretty sloppy.

Sometimes it does give some slightly different behaviour. On one occasion it seemed to want to read the floppy drive, but the drive made some sort of ratcheting noise. It sounded like the head was trying to move past the end of its travel. I had to power off to get it to stop. Remember, the FDD is confirmed working in another PC. On another occasion, the HDD sounded like it read something, but then the drive LED went out and it didn't do anything more. I can't reproduce this behaviour. It doesn't do the same thing after rebooting.

Is there some way that I can test the ISA bus? I have a multimeter, but no oscilloscope. The video card seems to work perfectly. Would that indicate a fully working ISA bus?

 

[modem7]

I checked the PSU. It's giving me a steady 12.3v and 5.15v. They don't fluctuate at all during the boot process.

I have a multimeter, but no oscilloscope.

A multimeter will not show how 'clean' the voltage is, and so your measurement only provides a certain level of confidence. You did not earlier indicate trying a spare PSU, and so I guess that you do not have one, nor an adapter such as shown at [here].

I've also done some testing with the RAM. I removed bank 3 and powered on. No change. I also removed bank 2. No change. Removed bank 1. No change. Then I removed bank 0 and reinstalled 8x spare 64K chips I had lying around, and 1 of the ones I removed during testing to give the full 9 chips for bank 0. Again, no change.

Imagine some kind of addressing problem. For example, writing to an address actually writes to multiple addresses.

Sometimes it does give some slightly different behaviour. On one occasion it seemed to want to read the floppy drive, but the drive made some sort of ratcheting noise. It sounded like the head was trying to move past the end of its travel. I had to power off to get it to stop. Remember, the FDD is confirmed working in another PC. On another occasion, the HDD sounded like it read something, but then the drive LED went out and it didn't do anything more. I can't reproduce this behaviour. It doesn't do the same thing after rebooting.

Out of curiosity, has this motherboard been affected in the past by a leaking NiCad battery ?

The video card seems to work perfectly. Would that indicate a fully working ISA bus?

No. Different cards utilise different amounts of the ISA bus.

Because your video card is working, what you have is a certain level of confidence (and only for the slots that the video card has been used in).

Is there some way that I can test the ISA bus? I have a multimeter, but no oscilloscope.

With a multimeter, I think that the best you can do is look at the motherboard schematic, then for each contact in slot 1:
* Very continuity of that contact to the particular pin of the particular IC that the contact is indicated as being connected to.
* Very continuity of that contact to the same contact in each of the other slots.

And if that is all good, all you have is a certain level of confidence.