Chasing smoother contact on a 6-axis arm for precision press-fit, holding <50 µm positional error while regulating about 15 N axial force at 0.2 m/s approach with an ATI Mini45 and EtherCAT drives. Can anyone point to practical impedance/admittance tuning references suited to industrial controllers and drive-level loops, ideally with example gains and stability margins for this force/velocity regime?
On a similar press‑fit, the single change that made our admittance stable was notching the Mini45 Fz at the arm’s first mode before the loop. For your “0.2 m/s” and 15 N, we found a 43 Hz peak and used a narrow notch at 43 Hz (±3 Hz, Q≈6), which let us run Kz≈250 kN/m with about 0.9 damping and about 45° phase margin to stay under 50 µm; if you can’t ID the mode, a quick 0.1 N swept‑sine on contact finds it fast.
Project the Mini45 force onto the approach normal before the admittance and zero the tangentials during approach; that let us raise Z stiffness without exciting lateral modes — on an EtherCAT torque stack we run about 1 kHz inner torque, admittance crossover about 20 Hz, Mz about 1.2 kg, Bz about 80 N·s/m, Kz about 2.5 kN/m for 15 N at 0.2 m/s, holding about 35–40 µm. Agree with @derekT98 on conditioning; if your effective inner bandwidth is <300 Hz, drop crossover to about 10–12 Hz or it’ll chatter. Are you in pure torque mode or position+torque overlay?