EPROM baking

[Mark G Thomas]

Hi,

> On Wed, Dec 13, 2017 at 9:08 AM, Mark G Thomas via cctalk <
> cctalk at classiccmp.org> wrote:
> 
> > Hi,
> >
> > I am working on several projects requiring 2708 and 2716 EPROMs, and
> > am finding some of my chips will not erase, and some will not take
> > a program. I've also learned more in the past week than I wanted
> > to know about repairing Data-I/O 29a/b programmers.
> >
> > I vaguely remember in the 1990s baking such EPROMs in the oven, but
> > I do not remember temperature or time. I was surprised that Google
> > didn't turn up anything useful with this info.
> >
> > I'm sure someone here will have some notes on EPROM baking.
> >
> > Mark
> 
> Mark,
> 
> If this is an issue about reviving bad eproms?  I assume you have tried the
> regular stuff.
> 
> What process are you using now to erase 2708/16's?  I have a simple eraser
> unit and it seems to always work.  Some eproms go bad but I never have
> issues with erasing them.  My point is that maybe you need a better prom
> eraser unit.  

They seem to erase fine, using a PRO-LOG 9103 eraser (box, timer, tube...)

> I would avoid baking them until you have exhausted other
> options.  Not sure what others think.  This topic has come up before here,
> about putting them outside and all that.  The erasers are all over ebay,
> and the hardware store is full of the correct types of lighting, why not
> make a box that will do the job?    I assume there is more to it that
> simply erasing them.
> 
> 
> Bill

Erasing seems to work fine. It's the re-programming them that is the problem.

On Wed, Dec 13, 2017 at 02:49:39PM +0000, dwight via cctalk wrote:
> When I was at Intel, years ago, I recall the baking was only to repair 
> the retention of the EPROMs. It was not to fix random failures.
> 
> It sounds like your EPROMs have various failures that wouldn't be helped by baking.
> 
> Each time the EPROM is programmed, there is a slight increase in the leakage of the floating gate. This was typical after thousands of program/erase cycles. Baking them repaired the damage to the insulating layer that was damaged.
> 
> Dwight

I don't think these chips have been reprogrammed many times. It seems more
age related, affecting some brands/models in my spares but not others.

The failure mode is the chips erase successfully, but any attempt to
program them fails, and they still test blank and read back "ffff...".
Some of these were chips I erased years ago before putting in my spares
drawer, and some had fine working code on them, but I erased them to 
re-program with a newer version of software on them, to discover I could not.

My stash of TI and NEC 2732s seem to have the disease, but my ST, 
Mitsubishi, and several others program fine.

In the case of a bunch of 2732s, I have tried both a vintage DataI/O 29A
programmer and a modern Batronix programmer, with the same results.
I don't think I have a programmer problem.

I still swear someone in the late 80's had me baking EPROMs in an oven
to restore their programability, but I don't remember the specifics. I 
tried a few at 450F for 15 minutes, but they still won't program.

Mark

-- 
Mark G. Thomas (Mark at Misty.com), KC3DRE