Why Does C204A3 Trip During Demo_Force_Twist_Release?

C204A3: Sudden stop detected appears on the panel when the force-twist motion exits and the controller is forced to transition from an active wrist movement to no valid motion command. The trip becomes repeatable when Speed_inc for axis 6, or wrist 3, is raised above 0.33. Force feedback can still react correctly; that does not prove the motion transition is valid.

Start with the command executed immediately after the monitored motion ends. A gripper action or wait may be where the panel reports the fault, but that node is often only where the controller exposes a stop that began at the loop exit. Adding more wait time wastes time because waiting does not create a controlled deceleration.

Check where the fault is reported

Run the sequence at the last repeatable speed and watch three points: the continuously monitored condition, the instant that condition becomes false, and the first node after the motion. Record whether the trip occurs during the move, on a wait, or on the gripper-open command.

If the error follows the loop exit rather than a particular gripper node, treat the transition as the first fault. Moving the gripper action or replacing its program node does not repair that transition.

Determine what 0.33 controls

The observed boundary is Speed_inc > 0.33 for axis 6. That value is an application variable in Demo_Force_Twist_Release; it is not identified here as a controller speed, acceleration, percentage, or physical unit. Read the expression that consumes Speed_inc before interpreting it.

1. Find every write to Speed_inc.
2. Find the motion command or pose calculation that reads it.
3. Check whether the value changes joint position, joint velocity, Cartesian rotation, or an increment applied once per loop.
4. Check how often that expression runs and whether the loop can terminate between successive motion updates.
5. Record the commanded and actual axis-6 motion at the exit, using the controller data available on the installation.

If the variable is added to a target on every pass, its effect depends on the update cycle and the controller's trajectory handling. If it is passed into a speed field, its units come from that field. The number 0.33 alone has no transferable engineering meaning.

Use 0.33 as the installation's observed transition point, not as a published limit. The decisive reading is the remaining joint or tool velocity when the continuous condition changes state.

Inspect the continuously monitored exit

A continuously checked condition can interrupt a move before its waypoint is reached. The physical arm still has velocity at that instant. If program flow falls directly into a wait, gripper action, or another command with incompatible motion requirements, the controller must reconcile the discontinuity. A demanded near-instant stop can produce C204A3; a related bad transition can also appear as an “invalid set point” error.

Check the branch structure next. An if configured for continuous evaluation behaves as a motion-interrupt mechanism while its condition remains active. An attached else branch adds another immediate transition at the exact point the condition changes. Remove the else temporarily if it is not required for the test, and place only the interruptible waypoint and any completion flag inside the motion structure.

global didWeMakeThePoint = False
movej waypointBeforeMyForceControlledOne
while Mz < .2
  waypoint
  didWeMakeThePoint = True
end

The flag distinguishes two outcomes: the robot reached the waypoint, or the monitored moment ended the move early. Use that result to select the next process action. It does not decelerate the robot by itself.

Also check for a short loop with no delay or for position updates that execute too quickly. Such structures can be rejected as an infinite loop before useful motion begins. That is a separate program-execution symptom, but it points to the same design problem: motion updates and program-flow transitions are not coordinated.

Command the stop at the loop boundary

Insert the stop immediately after the monitored movement exits. Use stopj(a) when you want deceleration in joint space and stopl(a) when you want a linear Cartesian stop. The argument a is deceleration: rad/s/s for stopj(a) and m/s/s for stopl(a).

1. Return Speed_inc to a value that runs without the trip.
2. Place stopj(a) directly after the axis-6 monitored loop. This is the resolving branch already shown to clear the fault after moment detection.
3. Select the deceleration from the robot's permitted motion data and the process requirement. No installation-specific value is given, so read the applicable controller limits rather than copying a guessed number.
4. Run at reduced speed and confirm that the stop completes before the wait or gripper command starts.
5. Increase Speed_inc in small steps while watching remaining velocity, stop distance, contact force, and recurrence of C204A3.

A stop command also has a usable operating limit. If the incoming velocity is too high for the selected deceleration and available travel, lowering the motion speed remains necessary. Increasing a blindly can make the stop harsher and can move the problem into another motion or mechanical limit.

Do not place a wait between the loop and the stop. The controller needs a motion command that defines how velocity approaches zero; elapsed time alone supplies no deceleration trajectory.

Separate contact overshoot from exit deceleration

A high increment can create a second failure path when the tool contacts a rigid object. If force is sampled only between motion increments, the robot continues through the current increment before the threshold changes program flow. Higher approach speed increases the force reached during sensing, decision, and stopping.

Test this branch with the tool unloaded first. Zero the force reading before motion while the tool has no contact. Then use a continuously evaluated condition such as Fz >= -10 around a relative downward move, with the example threshold representing 10 N upward force. One described test motion is 10 cm downward.

1. Zero the force signal with no tool contact.
2. Start at low speed.
3. Run the relative move under the continuously monitored force condition.
4. Record the peak force, the threshold crossing, and the distance traveled before rest.
5. Raise speed gradually only while the peak force and stopping distance remain acceptable.

This arrangement is a search move: it travels until a force threshold is reached. It is not continuous force control. If the process must regulate force after contact, use a control mode designed to maintain force rather than repeatedly advancing a position target.

Testing with hand pressure can confirm that moment detection changes the branch, but it does not reproduce the stiffness, energy storage, or contact geometry of the actual part. Repeat the final test with the production mechanical stack.

Release stored wrist torque without a jump

If stopj(a) brings commanded motion to rest but the arm still jumps when the gripper opens, check for stored torsional load. Axis 6 can be stationary while the arm, gripper, or part remains elastically twisted. Opening the gripper removes the constraint and releases that stored energy as unexpected motion.

Place the gripper-open action inside a force-control wizard node. Select the tool as the reference frame, allow Z rotation to be compliant, and begin with a rotational velocity of 10 degrees/second. Increase it only as needed while watching whether the arm floats during release without producing the fault.

This branch addresses mechanical unloading, not missing deceleration. Keep the commanded stop before the compliant release when the monitored move can exit with nonzero velocity. If the fault follows a wait even after moving the gripper action, return to the loop boundary; the gripper is not the fault.

Verify the repaired sequence

Use a fixed test matrix. Change one item per run: Speed_inc, stop type, deceleration argument, contact condition, or compliant release. Record the node active at the fault, the monitored force or moment, whether the waypoint was reached, remaining motion at loop exit, and whether the arm moved during gripper release.

1. Run without contact at a known working Speed_inc.
2. Trigger the monitored moment and confirm stopj(a) executes before any wait or gripper action.
3. Confirm the arm reaches rest through a controlled joint-space deceleration and no C204A3 appears.
4. Repeat with the real part and record peak force or moment.
5. Test the compliant release if mechanical twist remains after the stop.
6. Increase speed stepwise and stop at the first unacceptable force peak, stopping distance, release motion, or controller fault.

Replacing a gripper program node with rq_open_and_wait() may change gripper-command execution time, but it does not define how the robot stops when the monitored movement exits. Treat it as a separate performance test after the motion transition is stable.

FAQ