EPROM erase times and lifespan
dave09 at dunfield.com
Thu Jan 29 09:31:00 CST 2009
> Working with EPROMs, one must have an eraser. I built my own a while back, using a germicidal UV
> lamp tube, a flourescent fixture, a sealed box and a programmable timer. It works well, and I
> typically can erase an batch of eproms in like three minutes.
Three minites seems awfully fast for complete erasure. For example, an Intel 1982 databook
lists erase time for the 2764 as 15-20 mins at a distance of 1 inch from the lamp (2537
Angstroms, 12000 uW/CM2).
This is pretty consistant with what I've observed - I normally bake em for about 15 mins.
I've seen some take as long as 30 mins (usually older "fixed program time" devices).
Keep in mind that although the output is digital, the actually process of an EPROM is
analog in nature - the output is decided based on the amount of charge on the gate.
This brings a few important points to understand:
1) Some EPROMS are "more programmed" than others.
Early devices with a fixed proramming time per cell normally are saturated, and well
programmed. Later "intelligent" algorithms would watch the device, note how much charge
was needed to make a 1->0 transition occur in the cell, add a margin and "your done".
Much faster - but depending on the equipment, sometimes "less programmed".
2) The comparitor is voltage dependant on most devices.
An EPROM bit near the 0/1 threshold may change state with a difference in power supply
voltage - Intel spec'd a supply of 6V when using their "intelligent algorithm" because
that would raise the comparitor voltage and cause it to require more charge to get a
zero bit. This insured that you were programmed "well past" the normal boundary at 5V.
You can sometimes read an EPROM that is suffering from "bit rot" by powering it at a
A side effect of this is that:
3) Some EPROMs are "less erased" than others.
An "erased" EPROM read at 5.2v might not be erased when read at 4.8v ... so just because
it all reads FF's doesn't mean it's really erased. The fact that you see some "mostly
erased" suggests you are a little close to the boundary. Just like programming, you need
to add a margin to insure adaquate erasure.
My guess is that "lightly" programmed devices are showing just under the 1/0 threshold
after 3 mins, but probably aren't fully erased. You might also want to try and find
out the specs for your lamp and see if it's close to a normal eraser, and adjust your
dave09 (at) Dave Dunfield
dunfield (dot) Firmware development services & tools: www.dunfield.com
com Collector of vintage computing equipment:
More information about the cctech