CANopenNode includes optional trace functionality (non-standard). It monitors choosen variables from Object Dictionary. On change of state of variable it makes a record with timestamp into circular buffer. String with points can later be read via SDO.

Trace is disabled by default. It can be enabled using Object Dictionary editor. Include also CO_trace.h/.c into project, compile and run.

Here is en example of monitoring variable, connected with buttons (OD_readInput8Bit, index 0x6000, subindex 0x01). It was tested on PIC32:

# Enable trace first:
./canopencomm 0x30 w 0x2400 0 u8 1
 
# Press and hold the button on Explorer16 and execute SDO read command:
./canopencomm 0x30 r 0x6000 1 u8
[1] 0x08
# It displays same value, as was transmitted via PDO and visible on candump.
 
# Now get the complete history for that buttons with timestamp for each change
# and store it as a text to the file:
./canopencomm 0x30 r 0x2401 5 vs > plot1.csv
cat plot1.csv

If large data blocks are transmitted via CAN bus, then more efficient SDO block transfer can be enabled with command ./canopencomm set sdo_block 1

For more info on using trace functionality see CANopenNode/example/IO.html file. There is also a description of all Object Dictionary variables.

Trace functionality can also be configured on CANopenSocket directly. In that case CANopenSocket must first receive PDO data from remote node(s) and store it to the local Object Dictionary variable. CANopenSocket's trace then monitors that variable. Text buffer is then read with the similar command as above. But local SDO data access from CANopenSocket itself doesn't occupy CAN bus, so large data is transfered realy fast. Besides that, Linux machine has much more RAM to store the monitored data. Except timestamp is less accurate.