Identifying bad RAM on Amiga 1000 WCS board

[Ethan Dicks]

Hi, All,

I've been doing component-level diagnosis of a bad Amiga 1000 WCS
board and since I was unable to find this information anywhere, I
thought I'd post it to the list so that it's in the hands of more than
one person.

For an Amiga 1000 that starts up with a turquoise screen and never
asks for Kickstart, it means that the WCS RAM test has failed.  Common
causes are one or more bad 4464 DRAM chips on the WCS board or a bad
PAL.  I don't happen to have the PAL equations but I did spend some
time with a sick Amiga 1000, a Fluke 9010A and a cheap digital scope.

There are hand-drawn schematics floating around but they don't appear
to match the production hardware in either part placement or
completeness (the schematics describe 2 PALs, DAUGCAS and DAUGEN, but
the production hardware has two additional PALs, DPALCAS and DPALEN,
for one specific example).

If one has a Fluke 9010A and 68000 pod, one can test the WCS RAM by
pressing [RUN UUT] and turning on the Amiga and waiting a second or
two for the ROMs to set the right memory map bits to make the WCS
writable.  One can then do simple [READ] and [WRITE] tests to the
Amiga at $FC0000-$FFFFFF and even run a [RAM SHORT] on part or all of
that range (a RAM SHORT test on 256Kbytes will take more than a few
minutes).

The memory itself is a bank of 8 4464/50464 64Kx4 DRAMs at U1B-U1E and
U2B-U2E, arranged sensibly in two banks of 128Kbytes.  The chips in
row 2 are the lower half ($FC0000-$FDFFFF) and the chips in row 1 are
the upper half ($FE0000-$FFFFFF).  The individual bits are arranged as
follows:

U1E/U2E $000F D0-D3
U1D/U2D $00F0 D4-D7
U1C/U2C $0F00 D8-D11
U1B/U2B $F000 D12-D15

For those that want to trace individual bits the order on each DRAM is
pin-3, pin-2, pin-15, pin-17 which is slightly off the given order on
the 4464 datasheet of 2,3,15,17.

By way of verification, the WCS board I'm repairing failed the RAM
test with bad bits at $F000 when I pulled the defective chip from
position U1B.  The same chip failed testing in a Ming HT-21 "Handy
Tester" DIP logic and DRAM tester (but passed when tested as a 4416,
because the fault was not in the first 25% of the memory cells).

-ethan