EPROM erase times and lifespan

[Scanning]

Ethan ( et al );

Never ever power up an EPROM fresh out of the eraser. If the eraser is
putting out enough UV to actually do it's job, the EPROM gets really frikkin
hot, and the die is way the hell past the safe operating temperature for the
device. Also, just as we had a " metal migration " problem with fuse PROMs
that would heal when enough metal re-grew the fuse, EPROMs have a " charge
migration " problem. Insufficient  UV erasure will not bleed all of the
charge out of the cell and enough charge will sneak back in to make the cell
a "zero" again ( if that was the state before the erasure ) or it will be
fuzzy ( not quite one or zero, depending on Vcc ) as some of the guys here
have observed.

We had hundreds ( literally ) of industrial strength UV erasures and still
had problems. We finally had to have our CAL Lab test and post stickers on
each device because the UV output was all over the place. Plus the position
and distance to the lamp made a big difference. The range seemed to be
between 30 minutes to an hour to assure the device really was erased.
Devices that would not totally erase in that time were sh!t-canned and
erasers that would need more than an hour would have the UV tube replaced.
All of our erasers had conductive foam on the bottom of the trays.

Best regards, Steven

> > On Jan 29, 2009, at 10:27 AM, Jules Richardson wrote:
> >> Presumably more expensive erasers do things in-place - i.e.
periodically
> >> do a blank-check during the erase cycle?
> >
> >  I've never seen nor heard of an EPROM eraser that does this, in nearly
30
> > years of using EPROMs.
>
> I have never seen that either... the ones I've worked with
> professionally have had a variety of features including timers,
> conductive plastic rails at 0.6" spacing for ESD reasons and easy
> loading, but I've never seen any sort of eraser that applies power to
> the chips in any way.
>
> I myself did some experiments a couple of years ago at the South Pole
> when I didn't have a proper UV-A bulb (I just had a UV inspection
> lamp)... I could get the chips to superficially erase, but when I
> built a rig out of sockets and a trim pot to turn down Vcc to the
> EPROM for testing, I could reliably read the former contents at
> voltages below 5V even when the chips appeared empty at exactly 5V.
> In the end, I gave up on trying to use the lamp I had.  Fortunately I
> had a couple of 1MBit FLASH chips and was able to muddle through, but
> in the future, I'd probably pack some 27F256s or pack a real EPROM
> eraser and use ordinary EPROMs.
>
> -ethan