34 Piezo Touch and Biomimetic Sensing
34.1 Start With the Measurement Story
Prove a Tap Without Pretending It Is a Hold
Picture a robot hand lifting a glass jar. Its thin touch film gives a sharp signal when the jar first lands in the fingers, yet the signal fades while the fingers keep holding it. The control team must decide whether that fading means lost contact or normal sensor behavior.
Write two separate claims. The fast channel must notice a tap, slip, or vibration within the allowed time. A second channel must show that steady grip remains. Keep the source, time, force range, mounting, and test version with both records.
Tap the fitted finger softly and firmly, hold the load, release it, shake the cable, and repeat the test at the expected temperature. Check that a tap is not confused with electrical noise and that a long hold is not reported as a new event.
If the hand must prevent a dropped jar, keep the urgent safe response near the hand. A distant service may store trends, but it must not be the only path to releasing or tightening the grip.
This simple split does not size the sensing parts or predict every material effect. Practitioner turns the two claims into a fitted test. Under the Hood follows force through charge, the amplifier, leakage, and the limits of the model.
A piezo touch sensor listens for a physical impulse rather than a simple switch state. The useful starting point is the contact story: force, vibration, mounting, threshold, noise, and the proof that a real touch was detected.
34.2 Learning Objectives
After this page, you should be able to:
- Explain why piezoelectric sensors behave like rapidly adapting mechanoreceptors.
- Convert a dynamic force into charge and charge-amplifier output voltage.
- Choose feedback capacitance and resistor values from signal amplitude, saturation, and low-frequency cutoff constraints.
- Pair dynamic piezo channels with static-capable pressure sensors when a design must measure both events and held force.
34.3 Why This Follows Biomimetic Sensing
Biomimetic Sensing maps skin-inspired receptor roles to IoT sensor selection, adaptation, redundancy, and local processing. This page narrows in on one fast-adapting channel: piezoelectric touch and vibration sensing.
Use it when a tactile surface, robotic fingertip, vibration monitor, whisker-like sensor, or impact detector needs to capture changes without pretending that a piezo film can measure steady load by itself.
The mathematical gist. The chapter’s pC/N and 10 N tap create pC. If that force rises in 5 ms, the crystal sources 40.0 nA on average; a 1 nF charge amplifier gives a 0.200 V ideal peak. Its 100 MΩ feedback path makes ms, so a linear 5 ms tap retains about 97.5% of that peak while a slow hold leaks away.
34.4 Change-Sensing Like Skin
Human skin does not have one kind of touch sensor; it has specialists. Some mechanoreceptors, such as the Pacinian corpuscles, respond only when pressure is changing — they fire at the onset and release of a touch, signal vibration and texture, and then fall silent under steady pressure. They are rapidly adapting: they report events, not constants.
The piezoelectric sensor is an almost perfect engineered echo of that behaviour. A piezoelectric material generates electric charge when it is mechanically deformed — but only while the deformation is changing. Squeeze it and hold, and the signal fades. That makes piezoelectric films a natural fit for biomimetic touch, vibration, and impact sensing, which is exactly where flexible piezo polymers are used to build artificial skin and whiskers.
The useful design point is that this is not a failure of the sensor; it is a specialization. In a robotic fingertip, a piezo film can flag the instant an object slips, buzzes, or taps the surface, while a separate capacitive or resistive pressure sensor reports the steady grip force. That is the same division of labor skin uses: fast-adapting receptors report motion and texture, while slow-adapting receptors keep track of sustained contact.
A simple rule helps during sensor selection: if the question is "did something change right now?", a piezo or vibration channel is a good biological analogue; if the question is "what value is being held?", pair it with a static channel. Biomimetic design is the pairing, not the piezo film alone.
Intuition: a piezoelectric sensor is like a fingertip brushing a surface — it vividly feels the moment of contact and the texture as it moves, but it stops noticing a weight left resting on it. It senses change, not state.
Overview Knowledge Check
34.5 Charge Amplifier Basics
A piezoelectric element produces a charge proportional to the applied force, through its piezoelectric coefficient d:
Charge: Q = d × F d = piezoelectric coefficient (charge per newton) quartz ≈ 2.3 pC/N, PVDF polymer ≈ tens of pC/N, PZT ≈ hundreds of pC/N Charge amplifier: Vout = -Q / Cf (integrates charge onto a feedback capacitor Cf)
Worked example: a tap on a PVDF touch film
PVDF film with d ≈ 20 pC/N, a 10 N tap: Q = 20e-12 C/N × 10 N = 200 pC That charge is far too small to read directly, so a charge amplifier converts it to a voltage. With a feedback capacitor Cf = 1 nF: Vout = Q / Cf = 200e-12 / 1e-9 = 0.20 V A readable 0.2 V pulse marks the instant of the tap.
Note the output is a pulse, not a level. It appears as the force rises and disappears as the force settles, which is precisely the "event detector" character we want for tactile and vibration sensing.
Now check whether that pulse is large enough for the electronics. A 12-bit ADC over 3.3 V has a least-significant bit of about 3.3 / 4096 = 0.0008 V. The 0.20 V tap is roughly 250 counts, which is easy to threshold. A gentler 1 N tap would make about 20 pC and therefore 20 mV with the same 1 nF feedback capacitor, still about 25 ADC counts before noise and filtering. That is usable, but it leaves much less margin.
This is why Cf is a design knob rather than an arbitrary part. A smaller capacitor gives more voltage per newton but saturates sooner during hard impacts. A larger capacitor survives larger forces but shrinks the signal. If a gripper may see 1 N texture taps and 100 N accidental knocks, pick Cf so the small signal clears the ADC noise floor while the large signal stays below the input rail.
Why a charge amplifier, not a plain voltage buffer? A piezo element's own output voltage depends on cable capacitance, so it drifts with wiring. A charge amplifier reads the charge directly, making the reading independent of cable length — important for a sensor whose signal is measured in picocoulombs.
Practitioner Knowledge Check
34.6 No DC in Piezo Touch
The defining limitation of piezoelectric sensing — no response to steady force — comes straight from the physics of charge, and it is the same trade-off biology made.
Electrically, the piezo element behaves like a tiny charge source in parallel with its own capacitance and leakage resistance. When force changes, charge is displaced quickly and the amplifier sees a pulse. When force stops changing, no new charge is produced; the stored charge then leaks through the material and feedback path. The output therefore decays even if the mechanical load remains perfectly steady.
The feedback resistor and capacitor set the decay time. With Rf = 100 MΩ and Cf = 1 nF, the time constant is Rf × Cf = 0.1 s and the high-pass corner is about 1 / (2πRfCf) = 1.6 Hz. A 20 Hz vibration or tap passes well; a 0.1 Hz slow squeeze is strongly attenuated. Raising Rf or Cf moves the cutoff lower, but it also changes recovery time, noise, leakage sensitivity, and pulse amplitude.
Charge leaks away
The charge from a constant deformation slowly bleeds off through the material's finite insulation resistance and the amplifier input. So a held force gives a decaying pulse, not a steady level — piezo sensing is inherently AC-coupled.
The low-frequency cutoff
In a charge amplifier the feedback resistor sets a high-pass corner at fc = 1 / (2π Rf Cf). Below it the response rolls off, which is why piezo excels at vibration, impact, and sound but cannot weigh a static object.
Rapidly vs slowly adapting
Skin pairs rapidly adapting receptors (dynamic, like piezo) with slowly adapting ones (sustained pressure). A full biomimetic touch system likewise pairs piezoelectric films with a static-capable sensor such as a capacitive or resistive one.
Static touch partners
Held touch pressure can also come from tactile MEMS pressure tiles embedded in a compliant pad or from stretchable optical waveguides whose light path changes under pressure, strain, or curvature. Treat these as static or slow channels that complement the piezo event channel, not as replacements for its vibration evidence.
Cross-sensitivity to watch
PVDF is also pyroelectric, so it responds to temperature changes as well as force. Biomimetic designs must guard against a warm finger being mistaken for a press — a real cross-sensitivity, not a hypothetical one.
So the piezoelectric "weakness" is really a specialisation, and it is the same specialisation evolution chose for vibration-sensing receptors: report the change, ignore the constant. Building touch that feels both a tap and a resting weight means combining a dynamic piezoelectric sensor with a static-capable partner — exactly how skin does it. The validation test should therefore include both parts: a tap or vibration test to prove the dynamic channel, and a held-load test to prove the companion static channel.
Under-the-Hood Knowledge Check
34.7 Follow a Tap From Ceramic to ADC Code
A fingertip taps a piezo disc under a metal panel. At the piezo touch panel, the panel bends, the crystal makes charge, and the readout produces a short voltage pulse; a held finger soon becomes quiet even though the force remains. That physical decay is why piezo touch imitates rapid skin response better than steady pressure.
Read Figure from the applied force to the decision. The force first strains the piezo element. For this biomimetic tap, charge then enters the amplifier, whose feedback capacitor turns charge into voltage. Along the piezo sensing path, the feedback resistor lets that voltage fall, and the ADC plus threshold labels the pulse as a tap. Each labelled block can change sensitivity or timing, so “piezo detected touch” is not enough to locate a missed event.
For a 10 N tap and a coefficient of 20 pC/N, the crystal produces (Q=20\times10=200\ \text{pC}). With a 1 nF feedback capacitor, (V=Q/C=200\ \text{pC}/1\ \text{nF}=0.200\ \text{V}). A 3.3 V, 12-bit ADC has a 0.806 mV code width, so the ideal pulse spans about (200/0.806=248) codes. A 40 mV threshold leaves room above small vibrations while keeping the 0.200 V tap visible.
34.7.1 Predict What the Piezo Panel Will Report
Keep the piezo pulse polarity with the captured trace.
Mounting changes the mechanical input before the piezo equation begins. A thick steel face may spread a fingertip force across the disc, while a loose mount can ring after impact. Threshold and hold-off time therefore need tests on the final panel, not only on a bare bench element.
- Predict: A user presses slowly and holds for two seconds. Will the piezo touch output stay high? Check: No. The feedback path removes the charge, so the circuit mainly reports the change at press and release.
- Predict: The feedback capacitor doubles from 1 nF to 2 nF for the same 200 pC tap. What peak follows? Check: (200\ \text{pC}/2\ \text{nF}=0.100\ \text{V}), half the former piezo peak.
34.8 Release Checklist
Before relying on a piezoelectric biomimetic touch channel, confirm these points:
Start by the product requirement is an event, tap, vibration, texture, or slip signal rather than a steady force by itself. Then the piezo material coefficient, force range, and feedback capacitor produce ADC counts above the noise floor without clipping hard impacts. Next the feedback resistor and capacitor set a low-frequency cutoff that passes the events of interest and rejects irrelevant drift. After that cable capacitance, leakage, input bias current, and temperature cross-sensitivity are included in the validation plan. Continue by a static-capable capacitive, resistive, strain, or force channel exists when the system must know held pressure or weight. Finally validation includes both dynamic tap/vibration tests and held-load tests so the channel boundary is visible.
34.9 See Also
Start by Biomimetic Sensing for the broader receptor-role and hierarchical-processing design pattern. Then Sensor Circuit Fundamentals for the analog building blocks behind the readout path. Next Signal Conditioning for Sensors for amplification, filtering, and ADC preparation. Finally Sensor Specifications for response time, bandwidth, range, and accuracy requirements.
34.10 Next
Return to Biomimetic Sensing once the dynamic touch channel is bounded, then continue to Sensor Specifications to translate the biomimetic role into measurable requirements.