Background
Composite parts are easily damaged where they are held. A gripper that applies the right total force can still mark a part, because damage is a local phenomenon: it starts under a corner, at the edge of a pad, wherever the contact stops being distributed.
Force control cannot see this. The same force setting is safe on one geometry and damaging on another, and the difference is invisible until the part is inspected — or scrapped.
What was needed is the pressure at the contact surface itself, fine enough to locate a single hot spot.
Configuration
A single 4096-point sheet on the contact surface of one finger, read by one PXS064mini. No panel assembly, no merging — one sheet, one collector, one map.
- 7.6 × 8.1 mm pitch — the resolution that makes a two-point finding visible at all
- 400–6000 kPa range — covers soft-touch handling through to clamping loads
- 0.3 mm thick, flexible — it conforms to the finger pad, so it measures the real contact rather than a flat ideal
- < 1 ms response — fast enough to see the moment of contact, not just the held state
| Item | Value |
|---|---|
| Sensing film | PD-FS, 0.3 mm, piezoresistive |
| Points | 4096 (64 × 64) on one sheet |
| Point pitch | 7.6 × 8.1 mm |
| Pressure range | 400–6000 kPa, configurable |
| Accuracy / repeatability | ±5% FS / ±5% FS |
| Collector | 1 × PXS064mini (90 × 40 mm, < 100 g) |
| Sampling | 20 Hz, 12 bit |
How the measurement was taken
- Sheet laid over the finger contact pad, with a thin protective layer between film and part so the part surface is not marked by the measurement itself
- The part placed in its production position, gripper closed at the force used on the line
- The grasp repeated at several positions, because a contact map that changes with part position is telling you the pad is not conforming
- Map recorded while held, which is the state that matters for damage
What the data showed
The average pressure sat comfortably inside the limit. Two locations did not — they sat above the damage threshold for the material.
That is exactly the failure mode force control cannot detect: the total was fine, the distribution was not. The finding moved the discussion from how much force to where the contact actually sits — a geometry question, answered by changing the contact, not by lowering the setting.
Honest limits of this setup
- One finger, one face. This maps the contact surface, not the whole grip. Two-finger or multi-pad grippers need a sheet per contact
- Static grasp only. 20 Hz records closure and hold. Slipping, impact and vibration are outside what this chain resolves
- The film is 0.3 mm of extra compliance. On a small radius it changes the contact it is measuring; on tight geometries, correction or a thinner path is needed
- Curvature reads as gradient. On a curved pad, part of the pattern is geometry, not load — the two have to be separated before drawing conclusions
- Wear is real. A sheet used repeatedly against an abrasive composite surface should be inspected and recalibrated on a schedule
Why the client is not named
Naming the integrator and the part would disclose a handling process still being qualified. The customer agreed to publish the configuration, the method and the finding, and to keep the identity out.
No logo, no plant name, no quote. The contact map and the underlying dataset can be reviewed under a mutual NDA — start on the contact page.