franco@iuvax.UUCP (06/07/86)
STarter Kits are being mailed again. The botched 1MEG upgrade has been
fixed. During repair of the ST we hit upon a few things that may be useful
to other people attempting the upgrade. I will describe these things below.
This was the most thoroughly botched upgrade I have ever seen or heard about.
After the first attempt there were 2 bridges, one burned out DRAM on the
lower memory bank, two traces accidentally cut, two new DRAM pins not
soldered to old DRAM pins and several cold solder joints. But we found all
the problems in two evenings and now the ST is solid as a rock.
The moral: even a thoroughly botched upgrade attempt may be repaired with a
little patience, some PD software and a few simple tools - even if you don't
know terribly much about how the ST works.
The Tips: (caution - these worked for us but I do not know for sure whether
they will work for you)
Note - initially all we got was the "starting pattern" on the screen. The
disk drive did not spin.
1. We located the bridges using a 10X magnifying glass with the PC board held
up to the light (non component side facing the light while we looked at
the component side). The bridges were very easy to spot this way - you
will see why if you try it. Using the same technique we spotted and
repaired the cut traces.
Note - after repairing the traces and eliminating the bridges we found that
the system attempted to boot (disk drive spinning the disk) but bombed
with four bombs.
2. We located the faulty DRAM using a VTVM that also measures resistance.
We measured resistance between each pin of one of the DRAMS to ground.
We found that pin 9 measured 0 ohms while all other pins (except those
supplying power) measured about 1000 - 1500 ohms. We noticed that pin 9
of all DRAMs are connected (address line, I think). Careful inspection of
the traces connecting these pins suggested that the problem must be a
faulty DRAM (no bridges were visible). We isolated the problem chip by
cutting the trace connecting pin 9 of the 8th DRAM with pin 9 of the
9th DRAM. By measuring resistance we found the problem chip must be
one of the highest 8 DRAMs. Cutting another trace and measuring
resistance allowed us to conclude that the problem chip was among the
highest 4. After cutting two more traces we knew exactly where the bad
chip was.
Note - after replacing the bad chip (and the piggybacked one (soldering the
two chips together before putting them onto the board)) and repairing
cut traces the system booted but did not recognize the new DRAMs.
3. We found the faulty connections using a public domain memory editor as
follows. After booting, we changed the phystop variable (address $42E)
from $80000 to $100000 and the contents of the memcntlr variable
(address $424) from 4 to 5. This made the ST think it had 1MEG of memory.
Next we hit the reset button. The machine rebooted (thinking it had
1MEG of memory) and put the screen bitmap into the upper memory bank
(new DRAMs). Now we had a nice pattern of dots, resulting from the flaky
connections, on the screen. To find the flaky connections we brushed a
toothpick against each DRAM pin until the dot pattern changed. The
pattern changed two times (indicating two flaky connections).
Note - after resoldering these connections the ST booted and recognized
1 MEG of memory (and was steady as a rock).
Comment - it may seem to you that we added alot of wires (to repair traces)
to repair the ST. Actually the number of wires that we added was
small compared to the number of wires that Atari added at the
factory! Some of the Atari supplied wires connected to 68
ohm RAS resistors that are recommended for the upgrade (thank
you for helping with the upgrade, Atari!)franco@iuvax.UUCP (08/11/86)
STarter Kit mailings have resumed. Disks have been mailed to Ostrum, Nady, Schulz, Chan, Hoven, Sechrest, Doyle and Atkins.