The problem
I'm building a motion control rig for repeatable camera moves on miniature shots. The Y-axis servo oscillates when it hits loaded movement - the kind of slow, smooth pan you need for a corridor flythrough.
This is the boring part of filmmaking that nobody talks about. You can have the best miniature in the world but if your camera shakes by half a millimeter at 1:6 scale, it looks like an earthquake on screen.
What I tried
PID tuning. Hours of it. Kp at 450, Ki at 12, Kd at 1450. Every combination felt like it was almost right and then the servo would hunt at the end of a move.
Then I tried the obvious stuff:
- Bumped the power supply from 36V to 48V. More headroom, less sag under load.
- Re-routed encoder lines away from the high-current runs. Electrical noise was probably injecting jitter.
- Applied firmware patch v2.1.4 which supposedly fixes a timing bug in the feedback loop.
Result
Better. Not perfect. The repeated 20-second sweeps are now smooth enough for wide shots. Peak torque is close to spec. But there's still minor vibration in the top 10% of the speed range.
I'm starting to think the problem isn't the PID values or the firmware. It's the cable routing. The encoder cable runs through a drag chain that puts variable tension on the connector as the axis moves. At high speed, that tension changes faster than the servo can compensate.
Next step
Run another pass with additional mechanical damping on the cable path. If that fixes it, great. If not, I need a different encoder cable with more slack or a slip ring. Either way, this is the kind of problem that takes a week to diagnose and five minutes to fix once you know what it is.