2708 Programming Algorithm?

From: Dwight K. Elvey <dwightk.elvey_at_amd.com>
Date: Mon May 6 19:43:43 2002

>From: ajp166 <ajp166_at_bellatlantic.net>
>
>>>> I just remember that all the 87xx parts, 8741, 8755, 8748/49, etc,
>were all
>>>> 5-volt parts. further, I'm not at all sure that the 8080 had
>below-ground
>>>> signal levels, since they were intended to be attached to bipolar
>parts, e.g.
>>>> 8212, etc, which would have been intolerant of that. What I've got
>in my lap
>>>> is the 8080A data, which may, actually be different, but IIRC, the
>8080 needed
>>>> the negative bias supply so it could swing to ground and the +12 so
>it could
>>>> swing to a reasonable high level. My only contact with the 8080 was
>on boards
>>>> made by Intel, and, while I poked around with a 'scope and other gear
>from
>>>> time to time, I don't recall ever finding an address, data, or
>control signal
>>>> that wasn't TTL compatible.
>
>
>Wrong! the 8080 and 8080A had basically the same levels and drive. the
>issue
>of negitive voltages on the output is a red herring.

 I guess I'm wrong here. I just remember something about negative swings
but it was a long time ago.

>
>The 8708 accoring to the 1978 8048 manual is a THREE voltage part with
>the exact pinout as 2708. It's of course an intel number to allow the
>"kitting:
>practice that intel did do back then... "FAE>>> ya gotta use 8xxx parts".

 They did make changes in some cases with the leading number.
The 4702A was different than the 1702A in that the negative
rail could be -10V instead of -9V, although, the 4702A would
program and function at -9V. Maybe this is what was confusing
me.

>
>Now for a note, looking at the 1979 Intel component data book the 2708
>is
>listed and save for a faster programming method the 8708 is identical per
>notation in the data book!
>
>Recommended programming pusle width is .1 to 1 millisecond and the
>programming loop should not program any location for more than a total
>of 100mS. Though I remember programmin them using a 1ms pulse and
>doing a read to see if it took, programming it 5 times more for over
>program and looping till it took and moving on to next location. If a
>location
>took more than 90 hits is was flagged as bad. That seemed to get the
>best life out of the parts according to my notes. FYI: over eraseing
>them seemed to kill them too.

 These can be baked if they are in the ceramic package. This
brings them back to life. When I was in the lab, we did this
often. I don't recall the temeperature we used.
 The 2716's were more robust. We left some under the UV light
for weeks and had no data retention problems ( although, we
only required them to hold for about 6 months instead of 10 years ).
Dwight

>
>Programming voltage is nominal 26V pulsed! All other votages are static
>(Vcc, Vdd and Vss) with *ce/we being driven as needed for read or write.
>
>
>> That's not the point. The point is that the inputs and outputs are
>TTL (0V/5V) level and not a negative voltage. FWIW IIRC even the 8008
>had a fan out of more than one. I have the manual and can look if it
>matters.
>
>
>Correct it was 2 LS loads. And the old 4004 was ttl if used with the
>correct supply
>voltages -10 and +5 wich was typical of the PMOS logic.
>
>Allison
>
>
Received on Mon May 06 2002 - 19:43:43 BST