For a 'user' manual, the key point made here is: check your connections, or swap to better connections ahead of time.
Case 1
Most of our electronics problems were solved in 2009 by
these guys (we don't buy them there; that's just the first place I found a picture). Combined with the appropriate crimp tool, our electronics board is not only modular, it is incredibly easy to fix a connection and we haven't had a mysterious random shutoff since 2008. We mainly went to the mini Anderson PowerPoles when we realized that we kept loosening the wires within the default KOP crimps while changing a motor or victor. The other issue we had, even with good electronics toolbox organization, was that we would recrimp a wire with the wrong-sized connector (which caused
this little guy in Atlanta 2007 -- luckily our post-match checkups caught this before it caught fire). So for a random subsystem shutdown, check the leads. It seems like an obvious solution, however finding the problem connection is sometimes VERY challenging.
Case 2
The other thing to check is your BATTERY connection. The battery wire leads should be SOLID, NOT MOVING, and COVERED. Picking up a battery by its wires will eventually deteriorate the connection and cause the described random shut off. The "fix" your team did may have been to simply replace the battery. As dumb as that may sound, we've had many matches (ranging from FTC to VEX to FRC) where a bad battery connection has been the culprit of random system-wide shutdowns; the cause could be traced back to battery mishandling.
If the drive train stops responding, the problem may be a combination of a bad battery connection combined with improperly-designed drive train or control code. For example, the drive train may shut off during long sweeping turns if the code only tells one side of the drive train to go forward, yet the motors on that side must pull an excessive amount of current to generate the torque necessary to make the robot go forward and turn (depending on wheel traction and wheel base). If the battery has a bad connection, it may only allow 30-50 amps of power; this is enough allow the drive train to go full-forward and other robotic systems to function, yet in a sweeping turn the drive train may pull more current than that connection allows is physically capable of handling. Thus the drive train stops responding. The solution here is twofold, though the first is a stop-gap measure while the second is a must-fix issue: disable long-sweeping turning in the code, and inspect/fix the battery connection.
PWM Cables
Use clear nail polish. It's non-conductive, comes of cleanly, and is easy to cleanly apply. Stick the PWM cable into the Victor/Jaguar connection, verify that it works properly, then put a small dot of clear nail polish where the black plastic PWM connector touches the plastic on the Victor/Jaguar. The FRC electronics designers finally got something correct with adding lock tabs to the digital/analog sidecars, yet IFI/Luminary Micro still miss this (IMO, critical for anything industrial) function.