EPROM data life
marmotking
span. This has started to make me nervous. As I look around me,
everything I have seems to have an EPROM on it. Everything from
VAXstation motherboards, ascii terminals, to CD-ROM drives, to video
cards on old PC's that I use for imaging TRS-80 disks. Has anyone
considered the necessity of reading these EPROMs before they go bad?
Am I worrying about something that's not really a problem?
Wesley
that it is possible, but there's no definitive data out there. The biggest
lifespan issue seems to be in the number of erase/write cycles one can go
through.
Are you sure you aren't getting EPROMs confused with PROMs or EEPROMs?
I think the best thing we can all do is get any data that exists transferred
into something that can be put into an emulator or otherwise run on a modern
computer. And repeat that cycle every 10-20 years so it always will work on
the most modern hardware. As much as I hate to think about it, the old
hardware won't last forever...
Wesley
"marmotking" <tr...@marmotking.com> wrote in message
news:5829796b-1768-4ff6...@e23g2000prf.googlegroups.com...
marmotking
PROMs or EEPROMs. The two links that I stumbled across which got me
spooked were:
[1] http://www.classiccmp.org/oldfaq.txt
"3.1 Do EPROM's go bad?
Definitely. They apparently are considered to reliably contain data
for (on the outside edge) 15 years. This amount can be considerably
reduced if, for example, the sticker over the window has dried out and
fallen off. Luckily EPROMs were not used too extensively but they're
out there. An EPROM writer/reader is a relatively cheap investment
and an easy fix. Even if an EPROM has "forgotten" it's data it is
still fine for being "re-educated"."
[2] http://www.cpushack.net/EPROM.html
"They have no limit to read cycles and can store data for over 20
years."
15 years, 20 years even double at 30 years and a lot of this stuff is
starting to reach a questionable age for reliability. Although,
you're right, it seems a little less than definitive. I suppose it
can't hurt me to go through and image all the EPROMs on these boards
when I find them. I'd really hate to see some good old machine die
only due to fading EPROM.
Michael Black
I think it's very much the erasability of the eprom. Done with the
proper tools, you can erase them easily and fast. But I've seen all
kinds of schemes for erasing them without a proper eraser, and they
all count on length to make up for a weak UV source.
Even covered up, I suspect the window receives some UV, and accumulated
over time that can at least do a partial erase.
The real problem is that an eprom is more likely to be in a small run
unit. When something is making a big enough run, say the Microsoft
ROM in the Color COmputer, the setup fee for making mask programmed
ROMs is relatively insignificant, while the cost per unit goes way
down compared to eproms. For small runs, the per cost goes way up,
but if you are making relatively few, the overall cost is less than
going for mask programmed roms.
So you end up with roms that are most likely to go bad containing
code that is most likely harder to track down. Either you can't get
the code, or you can't find someone with a similar unit to copy the
eprom. Multiply that by those who do have the eprom may being in
the same situation, and that makes the situation worse.
The only eprom I can think of having that is important is the eprom
out of my Color Computer disk controller. The controller, and the
"dos" in the eprom, are third party. I think it was jdos, an early
version, and it made use of both sides of a floppy disk. But,
it used an odd scheme if I'm remembering properly.
I had to take the eprom out when I got a CoCo III because the old
eprom was not compatible, and the minute I took it out I lost access
to the floppies I'd written with the old "dos". I should have done
something at the time, but I never did. I can't even remember what
odd arrangement jdos used in its earlier incarnation.
So if I really want to recover those floppies, assuming they are still
readable, I have to hope the eprom is still readable. Or track down
the scheme that was used, and hope I can read the floppies on some
fancier system.
Michael
John
"marmotking" <tr...@marmotking.com> wrote in message
news:5829796b-1768-4ff6...@e23g2000prf.googlegroups.com...
I had bits start changing in an EPROM that had been in use for
10 years. Fortunately I still had the data file available so I
erased the EPROM and reburned it.
Matt J. McCullar
running life depends on several factors:
1) Manufacturer. At least a dozen companies made (make) EPROMs, perhaps
double that. One of my favorite types is the 2716 because I see it so often
on classic arcade game motherboards. Some companies made these chips better
than others did. I like those made by Hitachi, Texas Instruments, and Intel
because those still seem to hold up well. On the other end of the spectrum,
those 2716C chips made by Motorola always go immediately into the trash
because I have so much trouble with them now.
2) Environment. Some chips age better than others. Some chips collect
corrosion, others resist it. Heat and humidity. Again, it's a function of
who built the chip.
3) "Bit Rot." I have little evidence to support my theory, but my
experience is this: Once an EPROM has been programmed and that data file has
been left to sit in the chip for years on end, it may be difficult later on
to completely erase that chip. It wants to remember that particular data
pattern so much that it just won't accept any changes. Of course, I'm not
talking about the entire bit matrix, just a few stubborn bits. But if just
one bit is bad, the chip is useless.
4) Software archives. The older an EPROM type is, the more likely that the
device is no longer supported by the manufacturer. (The manufacturer may be
gone, too.) That being the case, if your circuit board's EPROM fails, it
should be rather easy to yell for help on the Internet and ask for someone
to e-mail you the data file so you can just program a newer EPROM and
replace the old.
Matt J. McCullar
Arlington, TX