Path: utzoo!mnetor!uunet!seismo!sundc!pitstop!sun!amdcad!ames!hao!oddjob!gargoyle!ihnp4!ihlpf!straka
From: straka@ihlpf.ATT.COM (Straka)
Newsgroups: sci.electronics
Subject: Re: backwards eproms (was Re: New Technologies)
Message-ID: <3741@ihlpf.ATT.COM>
Date: 18 Feb 88 14:32:39 GMT
References: <7944@sunybcs.UUCP> <2407@bloom-beacon.MIT.EDU> <360@splut.UUCP> <10@ucsd.EDU> <2848@bloom-beacon.MIT.EDU> <1988Feb9.151701.6201@sq.uucp> <194@hdr.UUCP>
Reply-To: straka@ihlpf.UUCP (55223-Straka,R.J.)
Organization: AT&T Bell Laboratories - Naperville, Illinois
Lines: 41

In article <194@hdr.UUCP> eric@hdr.UUCP (Eric J. Johnson) writes:
>In article<1988Feb9.151701.6201@sq.uucp> ian@darwin.UUCP (Ian F. Darwin) writes:
>	As I wrote on the same subject while reviewing an EPROM programmer
>	from a small American company (in MicroSystems, sometime in '83 or '84):
>	
>	I also found out (without harming the board) what happens when you
>	put in the correct EPROM, but insert it upside down.
>	*The little light inside comes on.*
>	You didn't know there was a little light inside an EPROM, did you?
>	Well, there wasn't supposed to be, and it's not there anymore anyway!

Reminds me of a little experience back at National in the old days:
(Bear with me, the end of the story is kind of neat!)

Before N-channel EPROMs there were P-channel EPROMS (eg. 1702, 5404(?)).
These devices were real pigs.  You see, you are really working against
physics to get the electrons up to the floating gate.  The only reason for
P-channel at the time was because nobody was too good at making N-channel yet
(contamination, threshold stability, ...).

Anyway, you had to hit these little puppies with 55V to do the programming!
At the wafer probe stage (real, virgin device), the technique was to
start with a ~20V programming voltage, and repeatedly hit the chip with
higher and higher voltages until the 55V was reached.  If you hit it with 55V
right away, you blow away the part.  The thoery was that during this
"puhout", as they called it, you actually heat up the local junctions enough
to *modify the metallurgical junctions* in critical areas so that they could
take the extra voltage without braeking down.

One day, one of the product engineers for that product was talking to me, and
said: "Come on over here by one of the wafer probers.  I have something to
show you."  He was probing a virgin wafer, and during the "pushout" part of
the test program, you could see (black cloth around the probe station
microscope) a few of the traces on the chip faintly INCANDESCE from the
immense current draw!  It would only happen the very first time that the
device was programmed.  Just a little bit eerie.  One wonders about long term
reliability.
-- 
Rich Straka     ihnp4!ihlpf!straka

Advice for the day: "MSDOS - just say no."