Experimental Physics and
Industrial Control System

[email protected] (Claude Saunders) · Mon, 5 Jun 1995 13:36:51 -0500

Here at APS we have been using a large number (40+) of PEP Modular
Bitbus controllers for about 3 years. After resolving some initial
problems related to "# of outstanding messages", "node offline command", 
and out-of-sequence messages, we have a solid driver which works with
both the older XYCOM Bitbus controller (i8044 based) and the newer PEP
(i80c152 based). All slave nodes are i8044 based.

I have been experiencing some very subtle problems with my bitbus
links run by the PEP controller that I don't see with the XYCOM
controller. I was wondering if anyone else was using this board and
having similar problems? 

Basically, the board returns a 0x91 packet (protocol error) fairly
regularly. This is normal when, for example, a CRC check fails (which
should be very rare). Also, from time to time, the transmit FIFO seems
to get stuck on full, resulting in an infinite board reset loop.
Neither of these problems occurs with the XYCOM controller.

It has been difficult to trigger on these problems and catch them on
an analyzer. I do have a couple of protocol snapshots for the case
when a 0x91 packet gets returned. These snapshots make no sense.
Basically, the PEP seems to spontaneously perform a "protocol
synchronization sequence" with a particular node for no apparent
reason. Admittedly, this could be an artifact of the analyzer.

I have spent a great deal of time studying RS485 (and our fiber version
thereof), SDLC, NRZI encoding, and the Bitbus spec. Short of getting
the complete firmware source from PEP, I am almost out of ideas here.
If there are any SDLC and/or UART experts out there, I would love some
fresh ideas.

  - Claude Saunders (APS Power Supply controls)

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System