Detecting Force

I'd like to start with a discussion about the trace shown here. This was obtained using a short metal rod pressing on a loadcell, in turn compressing a piezo film sensor fixed to a very rigid block (i.e. the loadcell is mechanically in series with the piezo film, which can't bend or stretch).

The upper (blue) trace is the output of a 100 lb compression load cell, with effective sensitivity of around 7.5 mV per Newton.

The lower (red) trace is the output of a piezo film sensor type NDT1-220K (MEAS p/n 1005935-1), connected to a charge amplifier with 100 mV/nC sensitivity, and a lower frequency limit of 0.1 Hz.

Note that the loadcell trace stays always above the initial baseline that exists at left-hand edge of screen.

With the piezo trace, it "follows" the force on the initial push, then "undershoots" on release of that force, and starts rising again from a "false" baseline, with this behaviour repeating and getting "worse". As I keep cycling the force, the piezo signal eventually spends as much time below the baseline as above – in other words, it becomes perfectly AC-coupled with zero net DC result.

The implication here is that the piezo signal cannot tell you much about the true force condition at any instant in time. It can give an indication of the magnitude of the AC (dynamic) part, but even here, the dynamic portion is not necessarily "accurate" compared with the true piezo sensitivity. The magnitude of the dynamic component is attenuated slightly, and the extent of this attenuation is dependent upon the exact frequency of excitation, relative to the cut-off frequency of the electronic filter. Note that the excitation here was approx 2 Hz while the filter frequency was 0.1 Hz, more than one decade away, but there is still amplitude error. On top of this, there may also be a phase error introduced by the filter.

This set-up was special, in that the piezo film sensor was pretty well clamped between two rigid blocks (the loadcell above, and a thick aluminum block below). In many other practical cases, the film may be fixed to a substrate that can undergo some bending, in which case the high "d31" piezo sensitivity to strain along the machine direction of the film would certainly generate additional contribution to the final output signal. It is not really possible to separate out the contributions from thickness-mode and stretch-mode excitation. In fact, even on the trace shown here, it is possible that the initial high peak from the piezo sensor (which seems stronger relative to later compression events in the trace) could have some additional contribution from machine-direction stretching.

Conclusion: it's really tough to measure "force" directly, using piezo film.

In many cases, it may be better to design a system that allows the film to stretch as a result of applied force.