Hi,
I was having the memory error on my 754D, U206 and U208 were specified in the log in different test runs. I did initially replace U206 and U208 without success. All the 308pins on the QFP controller had good solder joints. While probing for solder bridges after installing the new chips, I managed to catch two consecutive pins shorted.
Turns out this short was NOT a solder bridge, it was a bad SRAM chip with the pins (25-26) internally shorted! Not any one already replaced. I had to remove 7 chips from the gang before finding the bad one.
-> Note that the 8 gang is split 4 on top side and 4 on bottom side.
It is logical that if two address or data bits are shorted (I did not check de sram chip spec), the diagnostic will fail on the first chip affected.
No more error! Yay.
Cheers!
Well, that's very interesting, because that confirms my latest observations and the direction I was heading into.
Given I was still getting errors pointing to U208, I removed my replacement, and I was still getting what seemed to be a short between two (neighboring pins) addresses. Here are my notes from about a couple of weeks ago (I haven't had time to go back in there since):
"I checked all of the address lines (say A0 through the four chips on the top of the board, so I picked one chip's pin 12 and checked the other three pin 12s, and so on for all address buses/nodes) and they seem to check out. Then, looking for shorts - "address to address," so A16 to A14, A14 to A12, A12 to A7, etc, neighboring pin to neighboring pin starting at pin 2 - I didn't find any other measurement that'd "beep," other than pin 2 to pin 3, or A16 to A14.
Beeping threw me off, as I expected it to be a solder bridge, but I now think I was dealing with something completely different. It seems to me one chip has some odd condition between its pin 2 and pin 3 (maybe leaky/near shorted ESD protection diodes?)."
With this last sentence, I meant that, upon closer inspection (resistance measurement vs. continuity), I wasn't actually seeing a true short between A16 and A14, but about 10 ohms. Which to me ruled out an actual solder bridge (too large a resistance). So I take your observations as a confirmation that these 10ohms are the sign of an internal fault with one of the chips. Unfortunately, as you note, there's eight chips sharing those address buses, so any one of them could have this condition. I just hope it's not going to be the eight!
