386 board with a 8254 failure reported by Supersoft/Landark test ROMS

[KLund1]

Title about says it all. The board is a no-name clone. I have read this is programmable timer. So I can't just solder in a new one, if I could find it on the board.
Or is there a fairly easy fix for this I don't know about?
Thanks

 

[Ruud]

I have read this is programmable timer. So I can't just solder in a new one, if I could find it on the board.

Ii is a programmable timer indeed but I don't see the connection with (de)soldering it.

First question: what ROMs did you use? I personally only know the ones for the 8088 and 80286. If you used the 80286 one on your 80386, there is a possibility that the test can produce incorrect results.

Let's assume that the output of the test is correct and the 8254 is broken. Then there are two possibilities:
- You can find the IC as so on the board, it is a 24 pins DIL IC.
- The 8254 is embedded inside a larger square IC. Several companies produced chipsets for PCs and (even knowing the name) I have no idea what IC it could be.
In the first case you have to desolder it and to replace it with another one.
In the second case I wish you good luck in finding the correct IC and a replacement.

 

[Chuck(G)]

I suspect that the 8254 diagnosis is a red herring, particularly if it's not present as a discrete device. Channel 0 of the 8254 is associated with DRAM refresh; other channels are dedicated to time of day and speaker sound generation. Some early LSI chipsets economized on channel 0 circuitry because its function is pretty well fixed and not even implemented on some designs with alternate RAM refresh schemes.

 

[KLund1]

Thanks all for the replies!!
Here is a pic of this board. Let me know it any better/closer pics might be helpful.
With the original ROMs it does not post. No screen response. It shows a green and red LED in the lower left corner.
So I pulled out a PC diagnostics program package that was NOS until I opened it. The packaging says it works with AT-386. But there are so many 386 board/bios combos out in the wild you never know what diagnostics might work. But it dose not hurt to try the test ROMs
The diagnostic ROMs show as SuperSoft ver 1.1. I think there is a later version? I could burn those to some eproms I have.
But I has hoping that these would at least point me to where this board has a problem. The test pasess everything except the 3 8254 tests. After more reading, the 8254 test 1 for memory timing is critical to the function of the board. Without it nothing can happen, I think.
I could not find a 24 pins DIL IC on the board. It might be part of the 'CHIPS' VLS square chip. But there is an empty 24pin socket. Perhaps it was pulled long in the past?
Thought, suggestions?
Again many thanks.

 

[Chuck(G)]

If that's a 14MHz crystal just south of the empty socket, it could well be the case that you're missing an 8254,

There are some sanity tests that you can perform to verify this. Ground is pin 12 and Vcc is pin 24. Test to see if pins 9,15 and 18 are tied together. Similarly, pins 11 and 14 should be tied together and tied to Vcc.

If those check out, you need to pop an 8254 in that socket.

 

[KLund1]

Thanks,
Yes, those pins are tied together. But the crystal shows KDC 10.000
If the 8254 is missing, will any 8254 work? don't they need to be "programmed'?

 

[Chuck(G)]

No, the 8254 is a programmable device in the sense that a program sends commands to it--no ROM, masked or otherwise. Basically, it's three 16-bit counters with some interface candy, such as the ability to latch a count and issue interrupt requests with some buffering. Every 8254 is just like the next, save for minor things such as speed ratings.

An 8254-2 should be fast enough.