2708 Programming Algorithm? - test-drb@ccmp.vtda.org

6 May 2002

On May 6,  8:21, Loboyko Steve wrote:
...
  I obtained a bunch of 2708 UVEPROM's, and I'd
like to
 be able to program them. As you may know, they require
 a different algorithm than the 2716 and on, and three
 voltages, +5, +12, and -5.

 I'm building a little board that will program 1/2 (the
 upper or lower) of a 2716 to a 2708 (I can program
 2716's).  But I can't find a description of the
 algorithm that is detailed enough. What I think I know
 is:

 1. The /OE line must be high
 2. select address and data (start at addr 0)
 3. bring PGM line from 0 to 27 volts for 1 mSec, then
 drop back to 0 (this seems very strange)
 4. repeat 100 times per address (not all at once,
 increment the address lines and loop to avoid
 "overheating" a byte.
 5. drop /OE and verify. 
My 1976 Intel databook differs slightly from that.

1. raise /CS (pin 20) to +12V
2. select address and data
3. not less than 10 microsecs later, apply programming pulse to pin 18
4. hold address and data for at least 1 microsec, then move on to next
5. repeat for required number of iterations
6. drop /CS to normal (*without* changing the address) and verify.

The required number of iterations depends on the width of the program
pulse.  Minimum pulse width is 0.1ms, max is 1ms.  The spec calls for a
total of 100ms programming time per location.

V(IHP) on pin 18 is 25V (min) to 27V (max)
I(IPL) on pin 18 is 3mA

I(DD) is 50mA (typical) to 65mA (max)   -- Vdd (+12V +/- 5%) supply current
I(CC) is  6mA (typical) to 10mA (max)   -- Vcc ( +5V +/- 5%) supply current
I(BB) is 30mA (typical) to 45mA (max)   -- Vbb ( -5V +/- 5%) supply current

Yes, these are almost identical to the read currents.

...
  3. bring PGM line from 0 to 27 volts for 1 mSec, then
 drop back to 0 (this seems very strange) 
Why does that seem strange?  It's the programming pulse, which overcomes
the barrier imposed by the floating gate on the EPROM cell.  It needs to be
high enough to overcome the barrier potential, and you don't want address
or data changing while that is possible.

-- 
Pete						Peter Turnbull
						Network Manager
						University of York