Prior Art & Thesis Research

The CapSense feasibility overview introduces two sensing approaches for extracting expressive data from the 555’s keys — a traditional capacitive overlay and an AC signal coupling scheme that uses the organ’s own divider signals as excitation. Both approaches build on decades of prior work in capacitive and optical keyboard sensing. This page collects that research lineage and the quantitative findings that drive our design targets.

Prior Art

This idea isn’t entirely new — there’s a thread of AC-coupled keyboard sensing running through keyboard history:

IKOR (1967) built keyboards with AC coupling between emitter pads and PCB tracks. Fixed-capacitance emitters and receivers were separated by an aperture-shielded key that modulated the coupling as it moved. This is the first known AC-coupled keyboard sensing design — the key literally acts as a variable capacitor in the signal path. Source: telcontar.net

Pye Electro-Devices (1960s) designed an operator console that detected finger touch via 1 pF capacitive coupling to a 100 kHz oscillator signal — demonstrating that sub-picofarad coupling changes are measurable with period-appropriate electronics. Source: telcontar.net

Giulio Moro’s PhD thesis (2020) — “Beyond Key Velocity: Continuous Sensing for Expressive Control on the Hammond Organ and Digital Keyboards” (Queen Mary University of London) — directly investigates the Hammond organ’s 9-contact-per-key action and shows that contact-closure dynamics carry continuous expressive information. The rate at which the contacts close (velocity) creates frequency-dependent transients that can be measured and mapped to MIDI expression. This is the closest published research to what we’re proposing. Source: instrumentslab.org

Hammond organ key-click — the sequential closure of 9 contacts during a single key press creates frequency-dependent transients that vary with keystroke velocity. Players learned to control the “click” expressively. The underlying physics is the same: the divider signal couples differently at different points in the key’s travel.

What Moro’s Thesis Measured

Moro’s work on the Hammond C-3 provides hard numbers that inform our approach. A few key findings from the thesis (which we have a local copy of):

Contact dynamics are velocity-dependent. Using optical reflectance IR sensors (Omron EE-SY1200) sampled at 44.1 kHz alongside the digital state of all 9 contacts, Moro showed that the contact bounce duration, the closing time offset between the first and last contact, and the presence of “late bounces” all correlate with both the speed and the type of touch (pressed vs. struck). Two key presses at the same average velocity produce measurably different contact patterns depending on whether the finger was already moving when it engaged the key (struck) or started from rest on the key surface (pressed).

The contact-closing region is narrow. On the Hammond, 95% of contacts close within a ~3.5 mm window of key travel (out of an 11 mm total throw). The 555’s single SPST contact per key has an even narrower engagement zone — the silver-plated plunger bridges the rails over perhaps 1–2 mm of travel. This means any continuous sensing approach needs high resolution in that critical region.

Contact bounce has a characteristic frequency. Individual contact bounces oscillate with a period of roughly 200 μs (~5 kHz), alternating between open and closed states. Early bounces last less than 5 ms; late bounces (triggered by the key rebounding off the keybed) can appear 3+ ms later and extend the overall transient. On the 555, we’d see a similar phenomenon — the silver contact plunger bouncing against the rail — but with a single contact rather than nine overlapping ones.

Optical reflectance sensing works. Moro’s platform uses McPherson’s keyboard scanner — infrared LED/phototransistor pairs (QRE1113) mounted at the back of the keys, sampling at 1 kHz per key. The sensor’s response follows an inverse-square relationship with distance, requiring a nonlinear calibration (three paired measurements per key). The achieved action-to-sound latency: 3.33 ms mean, < 5 ms worst case — well under the 10 ms threshold Wessel and Wright (2002) recommend for responsive instruments.

Note

The optical reflectance approach is worth noting as a potential Approach C for the 555 — it doesn’t require capacitive sensing at all. An IR sensor strip above the keys, similar to McPherson’s scanner or the Moog Pianobar, would provide continuous key position data without any interaction with the contact rail signals. The QRS PNOScan is a commercially available retrofit that works this way. The downside: it requires physical mounting above or beside each key and careful calibration of the nonlinear sensor response.

A velocity-based controller is insufficient. Moro’s listening tests confirmed that participants could reliably distinguish between pressed and struck touches on the Hammond — the contact-closure dynamics carry perceptually significant information beyond simple velocity. This reinforces the value of continuous sensing: even if we can only extract velocity from the 555’s single contacts, the shape of the key trajectory matters.

What Moro Built and Tested (Chapters 4-5)

Chapters 4 and 5 of the thesis move from measurement to implementation — Moro builds a complete expressive Hammond emulator and validates it with professional players.

The Bela platform. The sensing hardware uses IR optical reflectance sensors (Omron EE-SY1200) sampling key position at 1 kHz, connected to a Bela embedded audio computer achieving < 100 μs round-trip audio latency. The software side is a modified setBFree — an open-source Hammond B-3 emulator — patched to accept continuous key position as input instead of simple MIDI note-on/off. This is the platform that established our design targets.

Virtual contact emulation. Rather than hardcoding the 9 contacts of a physical Hammond, Moro creates programmable virtual contacts placed at configurable positions along the key travel. This lets the system emulate anything from a single-contact instrument (like the 555) to the full 9-contact Hammond action. A player study with 10 professional organists compared:

• C1 — single contact (equivalent to the 555’s current SPST action)
• C3 — three virtual contacts, evenly spaced
• C4 — three virtual contacts, compressed toward the bottom of key travel

During free exploration, players developed strong preferences for multi-contact configurations — the timbral variation from sequential contact closure was immediately perceived as more expressive. During structured performance with backing tracks, the differences were harder to detect, suggesting that free exploration better reveals the instrument’s expressive range. The threshold finding: minimum 3 virtual contacts are needed for perceptually useful expression.

Key state machine. Chapter 5 formalizes the key’s lifecycle into five states:

State	Description	MIDI Analog
Partial Press	Key moving, no contacts closed	Proximity / anticipatory trigger
Press in Progress	Contacts closing sequentially	Velocity calculation window
Key Down	All contacts closed, key at rest	Note sustained
Release	Contacts opening	Note Off trigger
Post-Release	Key returning to rest position	Damping / release velocity

This state machine maps directly to our ESP32 firmware design — the shift register provides the binary contact state, while the CapSense/AC coupling layer provides the continuous position that determines which state the key is in and how fast it’s transitioning.

Percussiveness vs. velocity. Peak velocity during the early ballistic collision phase — the moment the finger first strikes the key — is distinct from the average velocity over the full key travel. The Hammond’s characteristic key click is driven by this peak, not the average. For the 555’s MIDI conversion, the position derivative should extract peak early-phase velocity specifically, using the first ~2 mm of key travel after initial acceleration exceeds a threshold.

Aftertouch from over-travel. On the Hammond, roughly 1 mm of key travel remains after the final contact closes. Moro maps this post-contact displacement to continuous aftertouch, with a dynamic dead zone normalized per-key (since mechanical tolerances vary). A second player study (6 professional organists) found: continuous key position was discovered spontaneously within 0–9 minutes of exploration, aftertouch was rarely discovered without prompting, and percussiveness control was found by only 1 of 6 players initially. Between-session learning significantly improved confidence, suggesting these controls reward familiarity.

SpiceBook Open in SpiceBook ↗

References

Hammond Organ and Expressive Sensing

• Giulio Moro — PhD thesis (2020) — “Beyond Key Velocity: Continuous Sensing for Expressive Control on the Hammond Organ and Digital Keyboards”, Queen Mary University of London. instrumentslab.org — local copy
• Moro, McPherson, Sandler (2017) — “Dynamic temporal behaviour of the keyboard action on the Hammond organ and its perceptual significance”, Journal of the Acoustical Society of America, 142(5):2808–2822. The journal paper covering the contact bounce measurements and listening test from Chapter 3 of the thesis.
• Moro — Chapters 4–5 (2020) — Bela platform with IR optical sensors (Omron EE-SY1200), 1 kHz key position sampling, < 100 μs round-trip latency. Modified setBFree open-source Hammond emulator accepting continuous position input. Player study with 10 professional organists comparing single-contact vs. 3-contact virtual configurations.
• Moro — Chapter 5: Key state machine and percussiveness (2020) — Five-state key model (Partial Press → Press in Progress → Key Down → Release → Post-Release), peak early-phase velocity metric for percussiveness, flute physical model for continuous-position timbral response, player study with 6 professional organists evaluating continuous position and aftertouch.
• Moro — Chapter 6: Conclusion (2020) — “A velocity-based controller is insufficient to capture the nuances of the keyboard’s action.” Advocates for continuous key position as the minimum viable expressive input.

Optical Key Position Sensing

• Andrew McPherson (2013) — Portable keyboard scanner using infrared reflectance sensors (QRE1113) at 1 kHz sampling. Used by Moro as the sensing platform for the hybrid Hammond emulator.
• QRS PNOScan — Commercially available piano retrofit using optical reflectance sensors under the keybed. qrsmusic.com
• Moog Pianobar — Portable infrared reflectance device for grand pianos. Senses white keys by reflected IR, black keys by optical interruption. Originally output only Note On/Off; McPherson modified one for continuous key angle output.
• Bösendorfer CEUS — Premium reproducing piano with continuous key position tracking via optical reflectance sensors at 500 Hz. The first commercially available instrument with continuous key position output.

---

See Also

• CapSense Feasibility — Overview — the sensing approaches this research informs
• Analog Block Architecture — PSoC analog block multiplexing for the AC coupling approach
• Implementation Roadmap — where these design targets feed into the phased build plan
• Simulations — SPICE models validating the AC coupling physics