68300

The cable capacitance, network design and stub length into each device can 
have a big effect.  From the RF standpoint you want your network to look 
like a straight line from the master to your last node with very short 
(<6") side branches if any.   These side branches especially if just using 
any wire will cause reflections which can cancel the signal at one node 
and not another.  Try using profibus or device net cable and see if this 
makes a difference.  I think Device Net cable matches your 120 ohms while 
Profibus is 150 ohms.





"Godiska, Jim" <jgodiska@...>
08/28/2002 02:11 PM
Please respond to 68300

 
        To:     68300@yahoogroups.com
        cc: 
        Subject:        [68300] 68376: CAN bus rate and length issue


Hello List,
 
I am currently using the MC68376 (using a 20 MHz external clock oscillator
chip) along with Phillips TJA1050 CAN transceiver and the typical 120 ohm
terminating resistors on the bus.  Currenly, I have a machine set up with 
a
1 MHz bit rate, 35 nodes (in a linear network with no stubs), and a length
of about 85 feet, but the CAN bus starts going nuts.  It seems that if I
shorten the network to 31 nodes (with about 13 feet less cable length), 
then
the CAN bus appear to be working.  If I try a different combination of 31
nodes out of the 35, it also works again.  If I take the 31 node network 
and
remove one 1 foot interconnect cable and insert a 26 foot cable, the bus
goes nuts again.
 
At this point I am beginning to believe that the problem is simply the 
cable
length of the network.  The obvious solution may be to drop the bit rate
down to, say, 500 kHz.  However I have a few questions first.
 
1.  I can set up the parameters in the CAN module one of two ways.
Currenlty, I am using:
 
        _TOUCAN.CANCTRL1 = 0x87;        // Low tolerance clock, 1MHz CAN 
bus
        _TOUCAN.PRESDIV = 0x00;
        _TOUCAN.CANCTRL2 = 0xF3;
 
The other option is:
 
        _TOUCAN.CANCTRL1 = 0x04;        // High data rate, 1MHz CAN bus
        _TOUCAN.PRESDIV = 0x01;
        _TOUCAN.CANCTRL2 = 0x49;

Which would be better?
 
2.  If I drop the bit rate, does should I set the paramters more for a
higher data rate mode or for more of a less accurate clock mode?
 
3.  Does anyone have a chart showing recommended cable length, nodes, and
various bit rates for the CAN bus?
 
4.  Using a split terminating resistor with a center tapped capacitor tied
to ground, would help with noise issues, but I am thinking this is not the
main cause.  Do you agree?
 
5.  I am currently using 22AWG twisted shielded cable with the shield 
being
termiated at one end only.  I can increase the wire size to 20 AWG which
should help the bus length somewhat according to one application note that 
I
found, but I do not believe it to be a significant contributor to the
problem.  Any opinions?
 
6.  In the early days of investigating CAN bus implementations, I thought 
I
saw that the maximums were 1 MHz bit rate at a 50 meter length and about 
110
nodes.  I thought that these three things could be achieved at the same
time.  Am I right or wrong about this?  Are there certain other things 
that
I need to do to meet this description?
 
Thanks in advance,
Jim
 
 


[





[Non-text portions of this message have been removed]