[time-nuts] Agilent 53132A Needs Help

Yuri Ostry yuri at ostry.ru
Tue Jan 20 12:42:29 EST 2009


Tuesday, January 20, 2009, 17:26:00, Richard M. Hambly wrote:

R> One of my 53132As, an Agilent unit, s/n KR01202209 fail the power-on self
R> test with a FAIL:ROM error message.

Cannot say anything about your particular counter, but very often such
error is due to 'leaked' EPROM chip, that change value of some memory
cells over time. Last years I seen such problems 5 or 6 times with 15+
years old equipment, and in most times original EPROM image still can
be read out if you have a EPROM programmer that allow to set arbitrary
Vcc for a chip in programming socket.

Some background: Erased EPROM cell (actually small piece of
metallization between two layers of silicon oxide, acting both as a
capacitor and as a gate of MOSFET transistor on underlying layers)
reads as logical "one". When it is charged during programming, it
start read as "zero".

If some cell have small defects in insulating oxide, or just got a hit
of some high energy particle, part of charge can be lost and
"programmed" bit that should read as "zero" starting to read as "one"
under normal conditions (nominal Vcc).

There is a chance (very good chance, according to my own experience)
that you can find such "partially discharged" bits by lowering
(gradually) Vcc and saving read images to disk for further comparsion.

Usually I start from 5.0V, make 10-20 reads, saving each one to
separate file in a 5V0 directory, then switch to 4.9V, and do the
same, saving to 4V9 directory, and so on... Usually it is enough to go
below to 4V0...

When you analyze saved images later, first compare all files in each
directory to each other, you can find some bits that reads unstable at
given voltage. Then compare images between nearby voltages and if
there is any changes, it may be your "lost" zero bits.

If you go too low, some EPROMS that was written before and then erased
to program current image may show you some of former programmed bits
as zeros - you need to be careful. There was some "erased" EPROM chips
that read as blank under 5V but read out their previous content (and
CRC perfectly match) when read out at Vcc little below 3.8V (not all
brands of EPROM operational at that voltage, though)... 

If there is a CRC on a EPROM label, it may be very useful in
determining that your recovered image is really good. Some devices do
CRC check on startup and you can feel yourself safe enough if checksum
error is gone.

Always keep your original EPROM chip intact and do not expose it to a
UV or sunlight (if there is no label that cover their window) until
you are completely sure that you have correct image on hand. Use spare
EPROM of same type for experiments.

BTW: Looks like it is a good idea to have images of EPROMS and
calibration EEPROMs (if any) for all equipment in a safe place.

Best regards,
 Yuri, UA3ATQ/KI7XJ                mailto:yuri at ostry.ru

More information about the time-nuts mailing list