Why Does Displacement Respond Later Than the Force Command?

Problem

In a contact force-control setup, stepping the force command produces an immediate change at the DAC output, but the encoder shows no meaningful displacement until roughly 150 ms later. Is the delay physical, or an artifact of the measurement chain? A wrong answer leads to tuning the controller against a phantom dead time.

Likely Causes

1. Static friction (stiction). Below the breakaway force the axis simply does not move. The “delay” is the time the force ramp needs to exceed the breakaway level — it should then shrink when the force step is larger.
2. Contact pre-load and nonlinear stiffness. At light contact the effective stiffness is low; early force goes into deforming the contact rather than producing encoder-visible motion.
3. Drive dynamics and current limits. The DAC command is not the force at the load; amplifier bandwidth or current slew limits stretch the actual force rise.
4. Detection threshold. “Detectable displacement” implies a threshold; with a fine error band and a slow force ramp, the crossing time is a function of the threshold, not of dynamics.
5. Signal-chain latency mismatch. Force and position may be sampled in different loops (e.g. 4 kHz vs 16 kHz) or filtered differently; misaligned time bases create apparent delay.

Measurement Method

• Record DAC command, measured force and encoder position on one common time base at the highest available rate; verify channel alignment first with an event visible in all channels.
• Repeat the force step at several amplitudes. A stiction-dominated delay decreases with amplitude; a pure transport delay does not.
• Repeat at several preloads. A contact-stiffness effect scales with preload; friction largely does not.
• Compare the measured force signal to the DAC command to isolate drive dynamics from mechanics.

Engineering Interpretation

The amplitude and preload sweeps split the 150 ms into its components. If the delay collapses at higher steps, treat it as breakaway friction and address it with friction feedforward or a touch-down strategy — not with a Smith predictor. Only the portion that is invariant to amplitude and preload deserves to be modeled as true dead time in the loop design.

What to Test Next

• Quantify breakaway force in both directions and its variation over the stroke.
• Re-run the step tests with the force sensor signal aligned against a hardware trigger to bound the signal-chain latency contribution.
• Check whether the delay changes between the 4 kHz and 16 kHz loop configurations — if it does, the signal chain is involved, not just mechanics.