Corrupt memcal '424?

[Knackersjewels]

Hey guys, had a customer send me a 424 which they believe was damaged. I can't see anything physically wrong with it, but both the 27c512, and the moates G1 27SF512 read, but cannot be Identified with memcal ID tool. Is there a chance they've been corrupted somehow?

Attached 27SF512 bin, surely it's not a stacking issue or something, is it?

Mace tune on moates adapter.BIN

(64 KiB) Downloaded 39 times

[antus]

It seems to be blank below C000. Thats the first 3/4 of the chip showing blank, including the typical calibration area. I don't know what tune it is, or what is supposed to be there, but it looks like it may have failed internally and is not returning a lot of data that it should. A typical '424 tune would have content above 0x1000 or 0x2000.

[Knackersjewels]

So you agree that it's an ECM failure that's potentially damaged the memcal?

It's apparently a VS V8, with VN twin throttles, so almost stock

[antus]

I cant say what the root cause is, but I would guess failed prom chip. I doubt the pcm caused it, but your guess is as good as mine.

[Dylan]

It'll be the memcal. Start again with a different tune. You won't be worse off.

[vlad01]

I've had proms do that, they just don't read or write past a certain point.

I am sure I have told the story a few times before, but I had one that would write in two locations simultaneously on the chip until the first part being written "crashed" into the rouge 2nd copy of the contents being written elsewhere on the chip and would throw an error saying the address already has data in it. I tried everything to try to resolve it, but it was some weird internal fault of the prom that occurred after soldering it. I use sockets exclusively now as I have had at least 3 failures after soldering direct to the memcal. Haven't had a problem since.

Honestly, one of the weirdest things I have ever seen, the duplicate data writing. And both sections of data were exact copies too! Which to this day I still don't understand how a failure could result in something like that.

[antus]

Must have been a shorted interconnect right on the die. They're modular internally in banks, so the CE or PRG line or functionally equivalent which is supposed to be selectable and to one bank only from the control portion must have been shorting out to the trace next to it which went to the next bank. You can see 2 banks in the images here, I can only assume they were both on the same address and data bus, and writing at the same time when one should have been idle.

Example from https://bgprojectz.weebly.com/microscopy/eprom