Forum Post

[OpenArm Troubleshooting] Gravity Compensation Feels Wrong After Tool Change (Intermediate)

Arm sags or feels heavy after a gripper swap? Validate payload mass and CoM before retuning gains that were already correct.

Forum / Posts Index / OpenArm

Post

A very common OpenArm problem shows up right after a new gripper, tool plate, or end-effector accessory is installed. The arm may hold position poorly, sag in one region, or feel unusually heavy in teleop even though the controller stack did not otherwise change.

How are you diagnosing OpenArm gravity compensation that feels wrong after a tool change?

Please share how you separate payload-model mistakes from friction, balance, or controller issues, and what validation steps tell you the new tool is modeled well enough before normal operation resumes.

If you reply, include one exact symptom after the tool swap and one exact payload or compensation check that exposed the root cause.

Related troubleshooting path: Repeatability worse after fast recalibration · Torque saturation and current spikes under load

Module: OpenArm · Audience: builders-integrators · Type: question

Tags: openarm, gravity-compensation, payload, tool-change

Comment 1

Intermediate context: After a gripper swap the arm slowly sank about 6 mm over 10 seconds near horizontal pose, even with no command input. It felt like bad gains at first, but the pattern was pose-dependent and repeatable.

Comment 2

Intermediate context: The useful check was measuring tool mass and center offset with a hanging scale and 100 mm balance fixture. Real CoM was 18 mm forward of the model, which explained the gravity bias error.

Comment 3

Intermediate context: Once payload and CoM were updated, we revalidated with static hold current and slow teleop arcs across the workspace. We only released after hold error stayed under 0.4 mm in five representative poses.

Comment 4

Intermediate context follow-up: Follow-up question: what are your current go/no-go limits for static hold drift (mm) and hold current after payload model updates?

Quick Symptom Selector

Pick your closest symptom to follow the right troubleshooting path.

Not selected yet.

Quick FAQ

What is the first payload sanity check after tool swap?

Validate tool mass and CoM physically before changing gains; model mismatch is the usual root cause.

python tools/payload_id.py --tool new_gripper
python tools/hold_drift_test.py --poses p1,p2,p3,p4,p5 --seconds 10
When is compensation production-ready?

Release only after static hold drift, hold current, and low-speed arc checks all pass.

Can I copy these commands as-is?

Use them as a checklist template first. Confirm interface names, fixture IDs, and safety conditions in your own cell before execution.