Browse Prior Art Database

Displaying Detailed Power-On-Self-Test (POST) Errors

IP.com Disclosure Number: IPCOM000106140D
Original Publication Date: 1993-Oct-01
Included in the Prior Art Database: 2005-Mar-20
Document File: 2 page(s) / 86K

Publishing Venue

IBM

Related People

Parsons, AW: AUTHOR

Abstract

This report shows how a product without a front panel display can reveal detailed POST error codes, instead of just enabling a FAIL LED. Properly written POST code and a single processor driven output, connected to an external counter (almost any lab grade frequency counter will do), are all that is needed. The following example discusses how this was accomplished on the cost-reduced 5394 workstation controller.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Displaying Detailed Power-On-Self-Test (POST) Errors

      This report shows how a product without a front panel display
can reveal detailed POST error codes, instead of just enabling a FAIL
LED.  Properly written POST code and a single processor driven
output, connected to an external counter (almost any lab grade
frequency counter will do), are all that is needed.  The following
example discusses how this was accomplished on the cost-reduced 5394
workstation controller.

      The first step was to write the Power On Self Test (POST)
diagnostic software so that each error was identified with a unique
3-digit decimal number.  For simplicity, error codes were grouped
according to function, such as initialization (010-099),
microprocessor (110-199), DRAM (210-229), and so on.

      Next, an error serializer routine was written that would pulse
one of the processor's unused Peripheral Chip Select lines for the
total count of an error code.  For an error code of 629, the routine
would pulse the chip select line 629 times.  Naturally, this is the
signal that needs to find its way to the event counter.

      The 5394 has a CE (Customer Engineer) probe connector built
onto its PC board that provides +5 volts and ground for a test probe
used in field service.  This same connector has a third pin, used
only for polarizing, so that the test probe cannot be plugged in
backwards.  Since that third pin has no electrical connection, it was
the perfect choice for connecting the error code signal to, and the
adjacent ground pin would provide the necessary ground reference for
the event counter as well.

      All of these steps were pulled together by the diagnostic
software, which was structured to call the error serializer routine
every ten seconds if a fatal error was discovered.  In this manner, a
5394 could still be evaluated after the failure occurred.  On the
other hand, if no errors were found, or errors were considered
non-fatal, then the serializer routine would be called only once
before letting the 5394 product code run.  By the way, a successful
POST, with no errors found, was coded as 999 so that the eve...