"moi" schreef in bericht news: email@example.com...
Google for NVRAM. 2k types are on the market for years already. Used a Dallas DS1220 (now Maxim) at the time. If memory serves it was even pin compatible with the 2716. SGS Thomson (STMicroelectronics) had a similar device.
moi, google for DS1213C, DS1213D, DS1235YWL. I have all of these, plus an extra C version that have been sitting here for eons, and I will never use them now. Four in total.
Like petrus bitbyter, I used them to make my eproms writable.
I can't help you with specifics, as I got too old, and blind, so you will have to research the data sheets and figure it out for yourself, but you will be able to connect one of these up as a 2716, with a read-write switch.
They are all greater than the pin count of a 2716, so you will need to tie non used address lines either high, or low, and figure out the read/write protect switching. But it will all be simple resistor pullups or downs.
But it's all doable and cheap if you know what you are doing. You will need to make up an adapter of some sort, to get down to the 24 pin count of the 2716, if you are inserting it into an existing board.
I would hope that at least one of the four chips is still alive after 57 million years for you to give it a go, no guarantees there either.
All I need is for you to cover postage and a padded bag. Email me for details.
Site Map: http://www.dontronics.com/sitemap
I have EPROM programmers here. That's not the problem.
The requirement is to be able to fast-cycle the contents of a few locations under PC program control. This is to identify by trial-and-error (normal "analysis" yielded nothing) the required "checksums" for rows of data in a target device whose processor code isn't accessible for reverse-engineering.
I could use either RAM or parallel EEPROM (2816 style) but was hoping to avoid the design cycle of coming to grips with the to's and fro's of the shared memory in an emulator and then building a solution.
almost. It is four bytes at each of 160 locations spaced regularly throughout the 2K image, so it would be possible to achieve it with glue logic. But a shared memory approach (aka emulator) is my preferred approach. I am trying to avoid an extra design project to address the underlying issue.