Image Map Image Map
Page 1 of 2 12 LastLast
Results 1 to 10 of 16

Thread: EPROM failure question

  1. #1
    Join Date
    Jan 2011
    Location
    Vancouver, BC
    Posts
    4,482
    Blog Entries
    3

    Default EPROM failure question

    Can an EPROM fail in such a way that you can read it in a reader, but it doesn't actually work in circuit?

    Just dragged out my Hyperion again for one last go over. As I mentioned before it produces a cursor and can do graphics but no text. Whether the character generator EPROM is installed or not the behavior is the same.

    I still think it's likely this video attributes controller, but am trying to eliminate other, easier to obtain parts that it light be.

  2. #2

    Default

    Broadly, once an eprom is starting to suffer bit-rot it will become equally unreadable by the system it is in and also by any EPROM reader / programmer.

    Bits which 'rot' return from logic 0 to logic 1, so, if you read a byte multiple times and several of the bits consistently read zero they are certainly meant to be zero.

    Now, if you read from the same EPROM address a hundred times and some of the bits sometimes read low and sometimes read high, those are probably bits which were originally programmed as '0'. If you make that assumption and store whichever of the 100 'read' values has the most zero bits in it, you may just possibly have recovered the original programmed value of the byte. Repeat this for every memory location in the EPROM.

    Note this dodge really only has any chance of working if the eprom is just hovering on the brink of data loss, Once the data is really gone, it's gone.

    This reinforces the point that any 40 year (or more) old item of hardware with EPROMs in it is probably getting perilously close to EPROM failure - the best thing to do before that happens is to remove any EPROMs, read them, save a backup as a file and then over-programme the EPROM with the same data. No need to erase it first, just answer 'Yes' when the programmer software says 'Device is not erased, do you wish to patch?'. By taking the time to do this, you set the 'data fade' clock back to zero and the data should be good for another 40 years or so.

    Edit: One thing I forgot to ask, does your original EPROM have a four figure hexadecimal number written or printed on it? If so, that may be the checksum of the (correct) code. If you read the EPROM and checksum the code with your programmer's software and it comes up with the same number as the one written on the label, it would be reasonable to assume that the EPROM contents are OK.
    Last edited by SiriusHardware; August 20th, 2019 at 09:44 AM.

  3. #3

    Default

    Quote Originally Posted by falter View Post
    Can an EPROM fail in such a way that you can read it in a reader, but it doesn't actually work in circuit?.....
    Yes i've had it before, Eprom that can be read just fine and UV erased and burned again but no matter what i put the eprom in the eprom would not work in circuit, I've also had a couple of NOS Eproms that did the same, It wasn't a compatibility issue, They just did not work in circuit.

  4. #4

    Default

    Presumably, Malc, you are talking about an EPROM which consistently didn't work in any microprocessor system. I am assuming that the specific EPROM Falter refers to did once work in the device which it is in, and then stopped working at some point.

    Another 'trick' (apart from the one I already outlined above) which may aid recovery from a borderline / flaky EPROM is to read it much more slowly than usual - unfortunately you can't usually vary the read speed of a conventional EPROM programmer so it would be a job for something like an Arduino, one of the larger ones with plenty of I/O pins.

  5. #5

    Default

    Quote Originally Posted by SiriusHardware View Post
    so it would be a job for something like an Arduino, one of the larger ones with plenty of I/O pins.
    Here is a good video of someone building this type of programmer.

  6. #6

    Default

    Quote Originally Posted by SiriusHardware View Post
    Presumably, Malc, you are talking about an EPROM which consistently didn't work in any microprocessor system.
    Yes and no

    I am assuming that the specific EPROM Falter refers to did once work in the device which it is in, and then stopped working at some point.
    Me Too, Eproms do fail, Though it may not be the cause of Falter's problem.

  7. #7
    Join Date
    Jan 2013
    Location
    Marietta, GA
    Posts
    3,235

    Default

    Quote Originally Posted by falter View Post
    Can an EPROM fail in such a way that you can read it in a reader, but it doesn't actually work in circuit?

    Just dragged out my Hyperion again for one last go over. As I mentioned before it produces a cursor and can do graphics but no text. Whether the character generator EPROM is installed or not the behavior is the same.
    That is kind of an odd way for an EPROM to fail, but it is possible (Although, I would suspect corrosion on the legs or in the socket). If the contents of the EPROM are not blank, then just try burning a new one and find out.

  8. #8
    Join Date
    Jan 2011
    Location
    Vancouver, BC
    Posts
    4,482
    Blog Entries
    3

    Default

    Thanks guys. I'll try burning a new one in case. I tried putting a random 2732 in from something else (original was a 2732 obviously)... didnt seem to have any effect... basically getting blank space where characters should be.

    The thing that makes me think it's more likely the attributes controller is that it handles cursor blink.. the fact that the cursor doesn't blink makes me a bit suspicious.

  9. #9

    Default

    Quote Originally Posted by SiriusHardware View Post
    This reinforces the point that any 40 year (or more) old item of hardware with EPROMs in it is probably getting perilously close to EPROM failure - the best thing to do before that happens is to remove any EPROMs, read them, save a backup as a file and then over-programme the EPROM with the same data. No need to erase it first, just answer 'Yes' when the programmer software says 'Device is not erased, do you wish to patch?'. By taking the time to do this, you set the 'data fade' clock back to zero and the data should be good for another 40 years or so.
    I have found with the bright UV light source in my eraser is takes 3 minutes to blank an eprom, I read that if you place one outside in the sun, it takes roughly 7 days. But I have not read a figure until your post that says how long it takes otherwise, assuming the clear window has a light occlusive sticker on it.

    When I got my SOL-20 one of the four UV eproms had a missing sticker over the window, so it had some room light, but disturbingly, the other stickers were just thin paper that let some light in. I replaced them with aluminium foil stickers. So these are over 40 years old and it worries me they might be on the way to rom rot.

    I would be interested to know if the "40 year fade" is about what most others have found with UV eproms or did the manufacturers ever say how long it could take ?

    So I'm going to have to "refesh" the ones in the SOL soon it would appear. It makes the idea of the mask or fused ROMs very attractive.
    Last edited by Hugo Holden; August 20th, 2019 at 02:25 PM.

  10. #10
    Join Date
    Jan 2011
    Location
    Vancouver, BC
    Posts
    4,482
    Blog Entries
    3

    Default

    So is the assumption that if the EPROM contains valid data it must be working? Can it not have some other kind of electrical failure that retains the data but prevents the chip from operating?

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •