Its possible its nothing more than compatibility issues with the proms and the programming tool, some of them older ones required special setups and voltages to record to them, its hard to say for sure.

TrashPanda wrote on 2022-11-17, 15:12:

Its possible its nothing more than compatibility issues with the proms and the programming tool, some of them older ones required special setups and voltages to record to them, its hard to say for sure.

Yeah, one can probably assume that those modern programmers are made primarily with modern chips in mind at least. They might not be all that well tested with obsolete 80s technology. Also got the impression that things weren't as standardized then, that different 28 EEPROMs for example perhaps might work differently depending on maker, version, etc.

Xicor X2864AD is rated at 10000 write cycles and has a 100-year minimum quoted data retention ability.

That would imply great reliability, AFAIU.

EDIT: Could it be timing related somehow ? There are different speed ratings.

darry wrote on 2022-11-17, 15:31:

Xicor X2864AD is rated at 10000 write cycles and has a 100-year minimum quoted data retention ability.

That would imply great reliability, AFAIU.

With numbers like that and how hardy them older Proms are im siding with the programmer not being fully compatible with them as the issue here.

It is likely the programmer socket simply couldn't make proper contact. It is very hard to clean the pins at the narrow side where programmer socket holds onto... but there is a fix, move the chip side to side when in the socket, and it'll push away the crap in those parts and the problems should disappear.

There shouldn't be any aging that happens while chip isn't in use, only that the data in the cells will slowly leak away and eventually, in 100 years, according to the datasheet, you'll have data integrity errors but a rewrite will fix that for the next 100 years 🤣.

Tiido wrote on 2022-11-17, 18:25:

It is likely the programmer socket simply couldn't make proper contact. It is very hard to clean the pins at the narrow side where programmer socket holds onto... but there is a fix, move the chip side to side when in the socket, and it'll push away the crap in those parts and the problems should disappear.

There shouldn't be any aging that happens while chip isn't in use, only that the data in the cells will slowly leak away and eventually, in 100 years, according to the datasheet, you'll have data integrity errors but a rewrite will fix that for the next 100 years 🤣.

It's not hard at all with the right tool 😀 Used a 2mm fiberglass brush. Looks pretty much like an architect's mechanical pencil, with a glass fibre bundle instead of the lead. Bad contact is definitely not the issue here.

Good to hear. Those chips haven't seen many writes during their lifetime either. The automation system copies the content from the EEPROM array to battery backed SRAM at powerup (which probably only happened once, when the ship was commissioned). Programming changes are done live in SRAM, and once you're happy, you write it back to EEPROM by issuing a command. It's only done now and then, when there's been changes or additions to the ship's systems, or bugs/errors have been fixed. Sometimes there are years between the occasions. Everyday changes like adjusting a setpoint or masking an alarm only happens in SRAM, it's nothing you do an EEPROM write for . You can basically view the EEPROM as a backup for the very unlikely event of system power loss during SRAM backup battery replacement, or hardware replacements requiring you to completely power down the system.

Played around with a few more 2864:s tonight, there were some from Intel too, that behaved more or less the same. Also leaning towards the programmer compatibility issue theory now. Also came to mind that the USB ports in my laptop has behaved flaky with larger loads like USB hard drives etc sometimes, will try with my main rig before digging any deeper in this once I'm back home 😀