Factory testing of CAN bus physical condition

Simon Wilton

2003-01-21 10:05:34 UTC

I think that you might be able to break it into parts.

1. The resistance measure that you already use core to core without the
harness powered. There are likely to be a characteristic set of resistances
for a 'normal' system (or each system configuration). That deals with
shorts.

2. A core to core capacitance measure might also produce a characteristic
set of readings that will be significantly altered by breaks.

3. Assuming that you are using fault tolerant tranceivers, ping of all
modules using a test module that disables CAN_H and CAN_L in turn should
prove the integrity (the responding module is powered, and has good
connection on the bus between the test connector and itself). However, it
doesn't tell you much about what faults exist.

The active terminations are more of a puzzle. If they are integrated with
modules you have established connection to them by default if the module
responds. If not, the best I can think of is to allow them to be powered
with all modules quiet, inject a pulse and watch for returns. Rather than a
full-blown TDR, you could do a custom tester for this (you should be able to
get away with go/no-go for the absence/presence of returns). It gets a bit
more difficult if you have legitimate stubs, as you will get returns from
those anyway. The test set gets a bit more sophisticated if you want it to
gate out the legitimate returns, or compare the pattern of returns with a
known good system. It should be possible.

It really comes down to what you can spend developing a solution against
what you save in training and per unit in production.

Hope this helps

Simon
----- Original Message -----
From: "Meyer Glen W" <***@phoeintl.com>
To: <***@vector-informatik.de>
Sent: Monday, January 20, 2003 3:39 PM
Subject: [CANLIST] Factory testing of CAN bus physical condition

Post by Meyer Glen W
I'm wondering what manufactures are using to test if the CAN bus has
physical problems such as shorts and opens.
With most CAN busses I could simply look for the approximately 60 ohms of DC
resistance (two terminating resistors in parallel) across the two bus lines
and know that the bus had good connection to both terminators. But now I'm
using ISO 11783 with active terminators and would like to test the 4 lines
(CAN High, CAN low, power, ground as well as the shield). I'd like a simple
way of testing the harness. A method that a factory assembler could use.
- Using an oscilloscope or CANscope to look at the communication and
determine if the voltages are incorrect or the rise time doesn't look right
- Using a Time Domain Reflectometer (TDR) to indicate the distance to opens
or shorts.
- Replacing the terminators with a test harness that checks the continuity
of the wires.
- Simply checking the voltage of the two CAN lines. This would only tell me
if there are shorts. Opens may not be detected.
- Polling/pinging each module to make sure it is able to respond.
- Watch for bus traffic messaging problems
The first two methods would involve considerable training for most factory
assemblers.
The third method may take more time then available since the assembler only
has a few minutes to perform several tasks.
The remaining methods won't necessarily tell me that the bus is physically
correct.
Does someone have a better solution? I'd like to test only through a stub
since this is the easiest method of connection (service connector) and this
connection is made in the normal course of the assembly process.
Sr. Design Engineer
Phoenix International
A John Deere Company
1750 NDSU Research Park Drive
Fargo, ND 58102
701-451-3618
Fax 701-451-3747
--
Archives, unsubscribing etc. see canlist homepage
at http://www.vector-informatik.com/canlist/

--
Archives, unsubscribing etc. see canlist homepage
at http://www.vector-informatik.com/canlist/
Problems to canlist-***@vector-informatik.de