Quote:
Originally Posted by Tristan Lall
That's an interesting theory but there's still one thing I'm wondering: what was the return path through the sidecar? (How did the current get from battery to motor to sidecar to battery, or the reverse?)
Nice work getting to the finals. It's always nice to see Canadian content on Einstein.
|
My theory: the return loop was a high resistance path to ground through the CIM case and robot frame. From what I recall, most CIMs demonstrate a path back to ground through the case, even though it's often a few megaohms. Measuring resistance directly between the frame and groundpost with your multimeter often doesn't show anything, but measuring voltage across +12V and the robot frame will show you +12V, proving that some kind of return path exists...
The path likely didn't even go through the sidecar, but went from +12V battery -> breaker -> +12V PDB -> Arm Speed Controller -> RS775 (Case Short)-> Robot Frame -> CIM (High Z Case to Ground) -> Drivetrain Speed Controller -> GND PDB -> GND Battery.
Some things about this theory are still fishy:
1) You won't fault unless you are running both your arm and your drivetrain at the same time. The speed controllers shunt the fault path otherwise(?). This agrees with our situation as well. Our problems only occurred when running both the arm and drivetrain simultaneously.
2) Creating a fault loop through the high resistance of a CIM case to ground path doesn't SEEM like it would cause problems... If the return path really is a few megaohms, the voltage drop caused by the fault should be negligible. Maybe someone else has a better explanation here, but my guess is that there has to be a true
short from frame to ground occurring, albeit intermittent such that you would still pass inspection. I don't have our robot back from CMP yet, but would like to test resistance from the CIM case to the groundpost, both at robot rest, and when driving at full blast under load, with the CIM running. I wonder if those CIM cases do short to ground occasionally while running?
3) Both 610 and 2056 experienced very similar symptoms during the GTR, although we cut out usually for only 20-30 seconds at a time. This seemed to be in line with the time needed for a cRIO reboot. 781 cut out for the entire match. My guess here is maybe your fault had a much smaller resistance, and caused a more serious voltage drop - either in length or magnitude - causing your radio to kick over? Or the cRIO was rebooting repeatedly. That would explain losing comm the entire match, but does not agree at all with the information from the FTA, who said you retained communications the entire time... Strangely, our FTA said the same thing at the GTR, and told us to focus in on the Classmate possibly acting up. We could recreate our problem in the pits however, and clearly saw the yellow status light on our cRIO blink, indicating a cRIO reboot.
4) From 610s experience, we recorded varying resistances in the RS775 case short. As you rotated the motor, the resistances would jump around. If left unchecked, these resistances would get worse and worse the more we ran the arm. You could start the day with a non-shorted RS775, and by the end of the day, have developed a wicked short, just from running the arm in match conditions. At >200k the short didn't seem to cause any problems. Between 200k and 100k we saw drastically increased numbers of Jaguar CAN Timeouts - once every few seconds. Below 100k is when we would start seeing cRIO reboots.
5) Lastly, firmly tap the case of your main breaker a few times while running drivetrain and arm really hard. Draw as much current as you can through your robot while you do this. We had a minor issue here too, or at least we thought.
Sorry to hear that this... our takeaway for next year will be to find a clever way to completely isolate the RS775s... should we choose to use them.