PDA

View Full Version : Help repairing 16K-64K 5150 motherboard.



prime
April 18th, 2016, 03:16 PM
Hi all,

I have here a 16K-64K 5150 motherboard that is not working correctly. Though what it is doing is still an improvement over the state it was when I got it (dead).

So far I have :

I've checked the board for shorts and fitted 9x4116s in bank 0.
Checked that the +12, +5 and -5 are getting to the RAM, and that the roms have +5.
Replaced the 8237 and 8088 (which where missing), and swapped the 8255 which seemed faulty.
Diagnosed that one of the BASIC roms is bad so I have removed all 4 of them for now.
Put a known working 8 bit VGA card in a slot, & connected to monitor.

With the machine in this state I get the following :

With the IBM BIOS I get beeps : one long, one short, pause, one short.
With the Anonymous BIOS I get slightly different results depending on the presence of the VGA card.
Without I just get one short beep.
With I get one long and three short.

The DIP switches are set to :
12345678
01111111

Memory DIPS set to
12345678
11111000

The PSU is known good.


Does anyone have any clues what to try next before I start pulling chips :)

Cheers.

Phill.

SpidersWeb
April 18th, 2016, 03:29 PM
Taken from minuszerodegrees.net:


If MDA or CGA selected, then look for that type of card. If card not found, beep 1 long then 2 short. If card found, try to initialise the card. If there is a problem found when initialising the card, beep 1 long then 2 short.

Do you have an MDA or CGA card to test with?

Stone
April 18th, 2016, 03:45 PM
With the IBM BIOS I get beeps : one long, one short, pause, one short.This is an unrecognizable combination of beeps.



With the Anonymous BIOS I get slightly different results depending on the presence of the VGA card.
Without I just get one short beep.This indicates a successful boot without any problems.



With the Anonymous BIOS I get slightly different results depending on the presence of the VGA card.
With I get one long and three short.This indicates a video problem.

modem7
April 18th, 2016, 11:47 PM
You need to tell us what revision of IBM BIOS resides on your 16KB-64KB motherboard, because there are changes in behaviour between BIOS revisions. See [here (http://www.minuszerodegrees.net/5150/bios/5150_bios_revisions.htm)] if you do not know how to identify the revision.

Looking through the posts, I think that you may have the first revision, the 04/24/81 one. Why? Because I have heard the "With the IBM BIOS I get beeps : one long, one short, pause, one short" beep pattern before, and it is documented in the 'Failure' section of the web page at [here (http://www.minuszerodegrees.net/5150/basic/5150_cassette_basic.htm)]. It ties in nicely with your removal of the BASIC chips.

What revision do you have?


Put a known working 8 bit VGA card in a slot, & connected to monitor.
1. Some "known working 8 bit VGA card" do not work in an IBM 5150. Do you know if it is an IBM 5150 compatible VGA card?
2. If the 16KB-64KB motherboard has the first or second IBM BIOS revision, even an IBM 5150 compatible VGA card will not work - the third revision is required for VGA cards.

prime
April 19th, 2016, 04:34 PM
Right I've done some more testing this evening.

After making up a flash rom adapter on tripad board, which I programmed with the most recent IBM bios + BASIC, the board now boots with a single short beep, so no error beeps.
However connecting a CGA adapter (with jumpers set as they should be) results in a wildly dancing pattern on the screen (I'm using a CGA -> scart converter, but I verified that it worked with my Sinclair PC-200).
I've scoped the sync signals and there is nothing there, so my suspicion is that the 6845 is not being initialized.

I also tried again with the VGA card and still got no output.

Bearing these in mind I suspect that maybe one of the ISA bus buffers or one of the chips that enables them has failed, I'll have a play tomorrow.

Cheers.

Phill.

Chuck(G)
April 19th, 2016, 05:19 PM
You should also try the Supersoft diagnostic ROM, as some have suggested. It will initialize a display even if no motherboard RAM is operational. It does require a CGA or MDA/Herc; VGA and EGA will not work.

Flamin Joe
April 19th, 2016, 07:24 PM
+1 for the Supersoft Diagnostic ROM. Saved me a lot of time recently diagnosing an XT clone motherboard which was dead and giving no beeps. Had no idea where to begin since the normal place I would of started is the RAM but it's soldered on the board. In no time at all it pinpointed the fault being in Bank 0 so the repair now is pretty straightforward.

In case you don't have a CGA/MDA/Herc adapter to use it will still output beep codes which can be deciphered with the manual. It just can be a little bit confusing at times especially since by not using a supported display will result in it throwing display initialization failure beeps into the mix. I honestly would grab a compatible adapter if you can find one just for this purpose and keep it aside for any future diagnosing needs.

modem7
April 20th, 2016, 12:00 AM
After making up a flash rom adapter on tripad board, which I programmed with the most recent IBM bios + BASIC, the board now boots with a single short beep, so no error beeps.
Suggesting that all you need to figure out now is the video issue.


However connecting a CGA adapter (with jumpers set as they should be) results in a wildly dancing pattern on the screen (I'm using a CGA -> scart converter, but I verified that it worked with my Sinclair PC-200).
I've scoped the sync signals and there is nothing there, so my suspicion is that the 6845 is not being initialized.
Did you remember to change the motherboard's video switches to one of the two CGA settings (i.e. informing the motherboard BIOS that it needs to initialise the CGA card) ?


I also tried again with the VGA card and still got no output.
The upgrade of the motherboard BIOS to the third revision removed one possible cause of the problem.
Your VGA card not being 5150 compatible is still a possibility that you need to keep in mind.

Great Hierophant
April 20th, 2016, 07:31 AM
If I remember correctly, the first and second IBM PC BIOSes do not scan expansion cards for bootable ROMs. EGA and VGA rely on having their boot ROMs scanned, allowing them to initialize themselves. I believe they redirect Int 13h calls to their own code instead of the BIOS code. Without the scanning, most EGA and VGA cards will not work with the first and second IBM PC BIOSes.

prime
April 20th, 2016, 01:19 PM
Right some more progress, I've run the supersoft / landmark test ROM and things are "odd"

The ROM comes up with a normal display on the CGA monitor and starts doing tests but after the 3 timer tests and the DMA test the screen corrupts (see attached image).

Then I get the beep s which indicate a ram refresh test fail.

The odd things is sticking a scope on the /CAS and /RAS lines of the ROM banks (as the RAM is being written to), I get signals for banks 0, 1 and 2 but *NOT* bank 3, which I suppose would explain why the RAM test was failing.
30711

My first thought about the screen corruption was that the CGA card was faulty, but it does this with two cards at exactly the same point in the test, I think it unlikely that two different CGA cards from different manufacturers (IBM & Compaq) would have exactly the same fault.

Cheers.

Phill.

Stone
April 20th, 2016, 01:57 PM
In a word, RAM.

prime
April 21st, 2016, 12:18 AM
In a word, RAM.

Being pedantic Random Access Memory is three words :) :)

Though the RAM in this machine did come out of my spares box, and is all in good condition, pulls from boards where it was socketed and so not de-soldered. I'll pull it out and test it tonight in another machine that also uses 4116s (probably one of my Spectrums). All the RAM is listed on the list here here (http://minuszerodegrees.net/memory/4116.htm) and is 250ns or faster.

Cheers.

Phill.

prime
April 21st, 2016, 02:49 PM
Right some progress,

I checked the RAM in a Sinclair Spectrum and as expected it was all good, but still worth checking to eliminate it as a source of failure.

Working on the premise that the fault must be in the RAM or DMA circuits as the fault kicks in when DMA is enabled, I started tracing and replacing.
Having now replaced U17 an LS244, it is getting past the refresh test, so it looks like a bad LS244 was stopping the DMA address making it to the RAM to refresh it.
The diags are still reporting errors, keyboard & floppy but that is to be expected as they are not plugged in.

However there is still a problem with the display being corrupt, but it looks like I'm getting close to resurrecting this old beast.

Cheers.

Phill.

prime
April 24th, 2016, 09:13 AM
Well I've not got much further :(

What I have found though is that the screen corruption starts as soon as the ram refresh DMA kicks in. Through experimenting with hacking the Supersoft rom, I found that If I hack it to halt after the DMA controller has been programmed but before the CTC has been programmed that the screen corruption doesn't start until about 52-53 seconds after the DMA conroller is programmed. If I let it program the CTC as normal, it's almost instant.

I suspect somehow the DMA addresses are getting corrupt and the CGA is being addressed rather than the RAM, though the refresh still seems to be happening successfully.

I have also replaced U17, U18 and verified that U19 is functional, likewise the DMA controller is known working.

Any clues anyone?

Cheers.

Phill.

prime
April 24th, 2016, 04:01 PM
IT Lives!

Problem seems to have been that U15 had failed causing /AEN to always be low on the slots, meaning that the CGA card saw the addresses going past as part of the RAM DMA as valid writes and therfore the screen got corrupt.

Now to backtrack and test / restore the original chips that I replaced with new ones (in sockets) to get it back as original as I can.

Also found that a genuine IBM EGA card (configured with a CGA monitor) seems to work in Supersoft.


3078130780

Cheers.

Phill.

modem7
April 24th, 2016, 04:43 PM
IT Lives!
Congratulations.

I have added your board to the list at [here (http://www.minuszerodegrees.net/vcf_motherboard_failure_history.htm)].


Also found that a genuine IBM EGA card (configured with a CGA monitor) seems to work in Supersoft.
Any different to the photo at [here (http://www.minuszerodegrees.net/supersoft_landmark/supersoft_5150_via_ega.jpg)], which is for an EGA monitor on an IBM EGA card ?

prime
April 25th, 2016, 12:29 AM
Congratulations.

I have added your board to the list at [here (http://www.minuszerodegrees.net/vcf_motherboard_failure_history.htm)].

Worth adding that when I got the board a previous owner seems to have attempted repair, there was a socket in the DMA controller position (but no chip), and bank 0 ram had been removed and replaced with sockets.

Also might be worth adding the 5150 I did in March which was a bad Tant cap on the -12V line.



Any different to the photo at [here (http://www.minuszerodegrees.net/supersoft_landmark/supersoft_5150_via_ega.jpg)], which is for an EGA monitor on an IBM EGA card ?

Yeah almost identical to CGA output, this is a genuine IBM EGA card and I have the monitor set for CGA on it's DIP switches. I'm also using a CGA->SCART converter that's basically a PAL and a bunch of resistors.

Cheers.

Phill.

modem7
April 25th, 2016, 12:46 AM
Worth adding that when I got the board a previous owner seems to have attempted repair, there was a socket in the DMA controller position (but no chip), and bank 0 ram had been removed and replaced with sockets.
Although, no evidence that the removed items were faulty.


Also might be worth adding the 5150 I did in March which was a bad Tant cap on the -12V line.
Was there a thread on these forums for that ?

prime
April 25th, 2016, 02:03 AM
Although, no evidence that the removed items were faulty.

True.


Was there a thread on these forums for that ?
Nope :) Just looked at the info on minuszerodegrees :)

Cheers.

Phill.

prime
April 28th, 2016, 01:34 AM
Dammit!

Looks like it's not completely fixed :(

I can boot the Supersoft rom with CGA or EGA card and it produces a display and all is fine.

With the latest IBM BIOS or the Anonymous BIOS it works with the EGA card and it will start cassette basic and I can type / run programs, however put the CGA card in and I get no display and though I get the single beep to indicate POST succeeded, it doesn't seem to drop through to basic......typing "MOTOR" should click the cassette relay if basic is running, but it doesn't.

I suspect we may have another faulty buffer somewhere, will power up the soldering iron again......

Cheers.

Phill.

prime
April 28th, 2016, 02:48 PM
And it's working now.

The clue was that I wasn't getting the "no video card" beeps with the jumpers set to CGA/MDA, looks like the U23 buffer was fried and always returning 00 for the video DIP switches, so replacing it has made the machine functional again.

Modem7 you may want to update your site with the latest fix, may want to considder it as a separate fix as it was a problem independent of the first.

Cheers.

Phill.

modem7
April 28th, 2016, 10:30 PM
Modem7 you may want to update your site with the latest fix, may want to considder it as a separate fix as it was a problem independent of the first.
Updated. For the purpose of compactness, I put multiple faults of the one motherboard into one line. The exception is when additional faults develop on the motherboard down the track; i.e. a separate event.

fs5500
April 30th, 2016, 10:13 AM
Worth adding that when I got the board a previous owner seems to have attempted repair, there was a socket in the DMA controller position (but no chip), and bank 0 ram had been removed and replaced with sockets.

Also might be worth adding the 5150 I did in March which was a bad Tant cap on the -12V line.



Yeah almost identical to CGA output, this is a genuine IBM EGA card and I have the monitor set for CGA on it's DIP switches. I'm also using a CGA->SCART converter that's basically a PAL and a bunch of resistors.

Cheers.

Phill.


If you have IBM EGA Card, can you dump EGA BIOS via C000:0000 - C000:3FFF?

prime
May 1st, 2016, 02:12 PM
If you have IBM EGA Card, can you dump EGA BIOS via C000:0000 - C000:3FFF?

Sure here you go....

30979

Cheers.

Phill.

fs5500
May 1st, 2016, 02:21 PM
Wow, thanks very much.

ROM is verified!