exxos's Atari, Amiga & retro forum

terriblefire wrote: 01 Nov 2020 14:27 Well we can tell you exactly which pins are bridged if you do a complete memory detection and try editing the memory and see which byte lanes arent happy.

This is the 4th or 5th thread with exactly the same symptoms. all 4 were resolved using memory editor, complete memory detection and some magnification.

Ok, I will do a complete reflow (just in case) and follow this hints if the behavior does not change. Thx

terriblefire wrote: 01 Nov 2020 14:27 Well we can tell you exactly which pins are bridged if you do a complete memory detection and try editing the memory and see which byte lanes arent happy.

This is the 4th or 5th thread with exactly the same symptoms. all 4 were resolved using memory editor, complete memory detection and some magnification.

Reflown & performed memory edit test.
Same behavior.

Editing memory results:

Writing all 0x00:
$40000000 00 00 80 00 00 00 80 00 00 00 80 00 00 00 80 00
$40000010 00 00 80 00 00 00 80 00 00 00 80 00 00 00 80 00
$40000020 00 00 80 00 00 00 80 00 00 00 80 00 00 00 80 00

etc.

So bad lines at address $40000002 $40000006 $4000000A $4000000E $40000012 and so on--- any hint on where to look for bridges ?

And what happens when you try to write 1's ?

Try the following patterns, 0s As 5s and Fs.

It looks like a bad data bit D15 (lower half of the 32 bit data bus) so far

terriblefire wrote: 02 Nov 2020 15:46 And what happens when you try to write 1's ?

Try the following patterns, 0s As 5s and Fs.

It looks like a bad data bit D15 (lower half of the 32 bit data bus) so far

FFs -> all correctly written
AAs -> all correctly written
55s -> read back D5 at same positions where reads 80 if writing 00 (as I would expect)

Can you point me to the location of pins related to D15 (I obviously don't have any schematics or layout :)
Thanks

My bet is that D15 on the ram is shorted to power. But here is everywhere that bit gets routed.

terriblefire wrote: 02 Nov 2020 16:41 My bet is that D15 on the ram is shorted to power. But here is everywhere that bit gets routed.

Mighty Stephen !!!!!!
Correctly spotted: there was a bridge under IC7, indeed.
Not visible with the uscope, hidden enough.

I focused my troubleshoot on the cpld and did not consider the "Easy-to-solder" ram chips....

Board working 100% now, thank you very much for your support and for your spectacular project.

Cheers,

Excellent. Enjoy

@sampedenawa I also noticed this...
This will cause you to get 32mb repeated twice...

Enter some data at 40000000 and check if its repeated at 42000000

Then run Amiga Test Kit for further shorts.

I spotted this last one and It was "fixed" after my previous reflow.

Just finished to pass all memory tests, everything 100% works as it should.

If only we have had such an accelerator board at good old Amiga times .. ;)