Implementation Roadmap

For Players

This is the build plan. Phase 0 is maintenance — getting the organ healthy before we touch the electronics. Phase 1 is a single afternoon with an oscilloscope and a soldering iron, testing whether our sensing approaches actually work. Phases 2 through 4 progressively build out the MIDI conversion, from a single-octave prototype to full expressive control. Each phase ends with a decision gate: clear criteria that determine whether to proceed, pivot, or stop.

Phase Progression

graph LR
    P0["Phase 0\nMaintenance &\nAssessment"]
    P1["Phase 1\nBench\nTesting"]
    DG1{"Decision\nGate"}
    P2["Phase 2\nSingle-Octave\nPrototype"]
    P3["Phase 3\nFull Keyboard\nIntegration"]
    P4["Phase 4\nExpression\nLayer"]

    P0 --> P1
    P1 --> DG1
    DG1 --> P2
    P2 --> P3
    P3 --> P4

    style P0 fill:#92400e,color:#fef3c7,stroke:#d97706
    style P1 fill:#164e63,color:#cffafe,stroke:#06b6d4
    style DG1 fill:#374151,color:#f3f4f6,stroke:#9ca3af
    style P2 fill:#1e3a5f,color:#bfdbfe,stroke:#3b82f6
    style P3 fill:#1e3a5f,color:#bfdbfe,stroke:#3b82f6
    style P4 fill:#065f46,color:#d1fae5,stroke:#10b981

---

Phase 0 — Maintenance and Assessment planned

The organ must be electrically healthy before we start probing it with oscilloscopes and soldering CapSense boards onto it. A failing electrolytic capacitor during a bench test would contaminate every measurement.

Tasks

• Clean all Orbit III potentiometers with DeoxIT D5
• Free the stuck Sine Wave volume pot (penetrating treatment or replacement)
• Fix the sticky lower manual C6 key (likely felt/bushing, not electrical)
• Complete an extended 30+ minute burn-in test, monitoring for audio dropout, hum escalation, or thermal issues
• Photo-document all circuit boards with the console open — before touching anything
• Acquire the Orbit III service manual (eBay listings available)

Don't skip the burn-in

The “3-week phenomenon” is well-documented in the Wurlitzer community: unused organs appear fine during a quick power-on test, but electrolytic capacitors that have been sitting uncharged for years can fail after 20–30 minutes of sustained operation. The failure mode ranges from increasing hum to catastrophic capacitor venting. Run the burn-in before investing time in MIDI hardware.

Exit Criteria

All pots clean, no stuck keys, organ plays for 30+ minutes without degradation. All boards photographed. Known issues documented in the known issues log.

Cross-references: Initial Assessment, Known Issues

---

Phase 1 — Bench Testing planned

One afternoon with an oscilloscope, a few jumper wires, and the Cypress PSoC eval boards. The goal is to determine which sensing approach — CapSense overlay, AC signal coupling, or both — is viable for the 555’s electrical environment.

1. AC coupling probe test

   Solder a single insulated pickup wire near one contact rail on the lower manual. Connect to an oscilloscope probe (high-Z, AC coupled). Power the organ and press the corresponding key slowly while observing the coupled signal. If amplitude changes measurably as the key descends toward the rail, the fundamental physics of Approach B is confirmed. Connect the pickup to a CY8C29466 ADC input and log readings vs. key position.

2. CapSense baseline test

   With the organ powered OFF, connect a CY8C20236A or CY8C29466 to a single sensor pad mounted under one key. Verify that finger/key proximity produces clean capacitance changes with an acceptable noise floor. This establishes the best-case baseline.

3. EMI characterization

   Power the organ ON and repeat the CapSense measurement. Quantify the interference: how much does the 0–12 V square wave on the nearby contact rail shift the CapSense readings? Add shielding (ground plane, driven shield via the PSoC’s CMOD pin) and measure the improvement. The delta between organ-off and organ-on readings determines whether Approach A is viable without heroic shielding.

4. Decision gate

   The bench test results map to one of four paths:

   AC Coupling Works?	CapSense Works?	Path Forward
   Yes	No	Approach B only — AC signal coupling with PSoC analog blocks
   No	Yes	Approach A only — Traditional CapSense overlay with shielding
   Yes	Yes	Hybrid — AC coupling for position, CapSense for proximity/aftertouch
   No	No	Shift registers only — abandon continuous sensing, focus on fast on/off with debounce

Note

Even the “neither works” outcome is valuable — it means the MIDI conversion proceeds with the proven 74HC165 shift register chain and we save the cost and complexity of the expression layer. The shift registers are the foundation regardless.

Cross-references: CapSense Feasibility — Proof of Concept, CapSense Feasibility — EMI Problem

---

Phase 2 — Single-Octave Prototype planned

Build and validate a 12-key prototype that demonstrates the complete signal chain: key contact detection, continuous position sensing (if viable from Phase 1), debounce, and MIDI output.

Hardware

• ESP32-S3 development board (USB-MIDI native, BLE-MIDI capable)
• 2x 74HC165 shift registers (16 digital inputs for 12 keys + spare channels)
• CapSense and/or AC coupling hardware for 12 keys (determined by Phase 1 outcome)
• Single-octave mounting bracket for the lower manual

Firmware

• SPI scan loop running at ≥ 1 kHz for the shift register chain
• Debounce algorithm (Moro’s contact bounce data: ~200 μs oscillation period, 5 ms total settling)
• USB-MIDI output first (lowest latency, simplest debugging), then BLE-MIDI and DIN-5
• Per-key three-point calibration stored in ESP32 NVS flash (from Moro’s calibration workflow)

Latency Budget

Every millisecond is accounted for. Moro achieved 3.33 ms mean action-to-sound latency; our target is < 5 ms total.

Stage	Budget	Notes
Key contact scan (SPI read)	< 1 ms	74HC165 chain at 10 MHz SPI clock
CapSense / AC coupling read	< 1 ms	I2C at 400 kHz or PSoC ADC sample
Debounce + state machine	< 1 ms	Software, runs on ESP32 core
MIDI TX (USB)	< 1 ms	USB Full Speed, 3-byte message
Total	< 4 ms	Well under 5 ms target

Exit Criteria

12 keys produce correct MIDI note-on/off with < 5 ms latency. If continuous sensing is active, velocity correlates with keystroke speed. No false triggers during 10-minute continuous playing.

Cross-references: Approach, CapSense Feasibility — Design Targets

---

Phase 3 — Full Keyboard Integration planned

Scale the single-octave prototype to all 153 inputs: both manuals, pedalboard, Orbit III, and expression pedal.

Hardware Scale-Up

• 19 daisy-chained 74HC165 shift registers (151 digital inputs / 8 = 19 chips)
• CapSense / AC coupling scaled per the chip allocation analysis:
  • Approach A: 15x CY8C20236A on I2C bus (or 8x CY8C29466 on SPI)
  • Approach B: CY8C29466 analog blocks for bandpass/envelope detection
  • Hybrid: combination of both
• Expression pedal: 2 ADC channels on ESP32 (volume axis + pitch slide axis)

Firmware

• Key state machine implementing Moro’s five-state model:

  State	Shift Register	CapSense / AC	Action
  Partial Press	Open	Position changing	Track velocity, prepare note
  Press in Progress	Transitioning	Rapid position change	Calculate velocity from peak early-phase derivative
  Key Down	Closed	Position stable	Sustain note, monitor for aftertouch
  Release	Transitioning	Position returning	Send Note Off, track release velocity
  Post-Release	Open	Position returning to rest	Reset state

• Per-key calibration for all 151 keys (three-point linearization, stored in ESP32 NVS flash)

• MIDI channel assignment per the approach design: Ch 1 (Upper), Ch 2 (Lower), Ch 3 (Pedals), Ch 4 (Orbit III)

• Expression pedal mapped to CC#11 (volume swell) and Pitch Bend (lateral slide)

Orbit III Dual Contacts

The Orbit III’s 25 mini-keys already have dual contacts — Contact 1 for note-on and Contact 2 for second-touch modulation routing. If CapSense proves viable, it could replace Contact 2 entirely, providing a smooth continuous aftertouch curve instead of a binary second-touch switch. This would give the Orbit III’s monophonic synthesizer voice aftertouch expression that the original designers couldn’t achieve with mechanical contacts.

Exit Criteria

All 153 inputs scanned correctly. MIDI output on all three transports (USB, BLE, DIN-5). Latency < 5 ms sustained across full keyboard. Expression pedal produces smooth CC/Pitch Bend curves. No crosstalk between manuals.

Cross-references: Input Inventory, CapSense Feasibility — Chip Allocation

---

Phase 4 — Expression Layer planned

The shift registers give us on/off. The expression layer gives us how. This phase extracts continuous expressive parameters from the key position data and maps them to MIDI.

Velocity from Position Derivative

Standard MIDI velocity (0–127) derived from the rate of key descent. The position derivative is computed over the key’s travel, but the critical insight from Moro’s Chapter 5: use the peak early-phase velocity — the maximum derivative during the first ~2 mm of ballistic collision — not the average over the full stroke. This captures the percussive attack that Hammond players use to control key click.

Percussiveness Detection

A second velocity-like parameter that captures the sharpness of the initial attack. Computed from the jerk (derivative of acceleration) or from the peak-to-mean velocity ratio during the early phase. Maps to a MIDI CC for downstream sound engines that support percussive vs. legato articulation.

Aftertouch

If the 555’s key mechanism permits enough post-contact over-travel (Moro measured ~1 mm on the Hammond), the continuous position signal after contact closure maps to MIDI Channel Pressure or Polyphonic Aftertouch. A dynamic per-key dead zone accounts for mechanical tolerance variation.

MIDI CC Mapping

Parameter	Source	MIDI Message	CC#
Velocity	Peak early-phase key descent rate	Note On velocity	—
Continuous position	Absolute key depth	CC	74 (Brightness) or custom
Percussiveness	Peak/mean velocity ratio	CC	1 (Mod Wheel) or custom
Aftertouch	Post-contact over-travel	Channel Pressure	—
Expression pedal (volume)	Vertical pot axis	CC	11 (Expression)
Expression pedal (pitch)	Lateral pot/switch axis	Pitch Bend	—
Release velocity	Key ascent rate	Note Off velocity	—

Exit Criteria

Velocity responds perceptibly to keystroke dynamics. Percussiveness parameter distinguishes struck from pressed touches. Expression pedal produces musical results in a connected synthesizer. A player can feel the difference between the 555’s MIDI output and a generic keyboard controller.

Cross-references: CapSense Feasibility — What Becomes Possible, CapSense Feasibility — What Moro Built and Tested

---

The Principle

Giulio Moro frames the challenge as “widening the bottleneck.” Every stage in the signal chain — from finger to air — has a bandwidth. Traditional MIDI narrows the keyboard-to-sound-engine channel to a few discrete events (note-on, note-off, velocity). The Hammond’s 9-contact action carries more information than a single MIDI velocity byte can encode. Continuous key position sensing widens that channel back open.

For the 555, the architecture is layered:

• Shift registers are the foundation — reliable on/off detection for every key, every time. This works even if everything else fails.
• CapSense / AC coupling adds continuous position data — the raw material for expression.
• The expression layer (Phase 4) transforms that position data into musically meaningful parameters.
• Decision gates between phases keep the project honest. If a sensing approach doesn’t work, we fall back to the layer below without losing anything.

The goal isn’t to turn the 555 into something it isn’t. It’s to let the player’s intent — already present in the way they touch the keys — pass through to the sound engine without being flattened into a binary switch.

---

References

• Approach — ESP32 + 74HC165 shift register architecture
• Input Inventory — All 153 inputs cataloged
• CapSense Feasibility — Sensing approaches, EMI analysis, chip allocation, design targets
• Initial Assessment — Current organ condition
• Known Issues — Documented problems and their status
• Giulio Moro, “Beyond Key Velocity” (2020) — instrumentslab.org — local copy