PDA

View Full Version : Taiwan 8088 memory failure on Bank 1 - what did I miss?



SpidersWeb
March 12th, 2017, 06:48 PM
So I had a spare Turbo 8088 style board laying about - just another Taiwan special.

Anyway in the process of building an XT with it, I noticed it had 512KB of RAM on board but the RAM test only showed 256KB.
This isn't the end of the world because in the end a jumper switch + an AST Six Pak Plus, it now has 640KB.

But I'm mostly curious about what I missed / don't understand here when trying to diagnose the bad chip on the board itself.



Supersoft Diagnostics ROM showed bit 0 of bank 1 as bad.

- replaced the chip
- swapped the chips around in the bank
- replaced the socket

No change from all of these.

- tested CAS was linked with appropriate chips
- tested RAS was linked with appropriate chips
- replaced DRAM in bank 0 bit 0 (just in case it was holding the data line high or low)
- tested all address lines had continuity
- tested Data line had continuity (and all other appropriate pins)

No problems detected or changes in results.

If there was a problem with CAS, I'd expect a string of chip failures. Ditto for RAS.
If there was a problem with the data line, I'd expect all chips to fail.
If there was a problem with addressing, I'd expect it to not be limited to bit 0.

After adding the AST card (after all of the tests above):

When the on-board memory was not limited to 256KB, it also caused memory errors on the AST card including right through the 512KB to 640KB region - and with random bits and almost random locations. (I expected errors 256-512)
When the on-board memory was limited to bank 0 only, the memory test passed all the way up to 640KB.

Just has me scratching my head.

Chuck(G)
March 12th, 2017, 06:50 PM
Did you try writing to and reading back from the memory that was supposed to be faulty?

SpidersWeb
March 12th, 2017, 07:02 PM
Did you try writing to and reading back from the memory that was supposed to be faulty?

Not through debug (or similar), just the SuperSoft tests.
What could this show me? I can move the jumper back and take out the AST card to run new tests.

SomeGuy
March 12th, 2017, 07:48 PM
With the way that is behaving, I would be tempted to test the RAM in another machine, just to rule that in/out for sure.

If it works with just one bank, perhaps test each bank one at a time.

I'm not sure I would fully trust the SuperSoft ROM if there is anything odd about the board (yea, those Taiwanese boards are usually fairly standard) it might need its BIOS to initialize it. If it boots, try a DOS based RAM tester. - Which reminds me, is there a good DOS-base RAM tester that runs in only 256k of RAM?

Chuck(G)
March 12th, 2017, 11:52 PM
Not through debug (or similar), just the SuperSoft tests.
What could this show me? I can move the jumper back and take out the AST card to run new tests.

It might show you if the memory was really present--and what state (i.e. bad bits) it was in.

Stone
March 13th, 2017, 04:26 AM
Which reminds me, is there a good DOS-base RAM tester that runs in only 256k of RAM?I have a few short, 256K cards that I use for machines that only have 256K. 512K is enough to run the RAM tester and other diagnostic programs.