The Reverse Path

The approach page describes the MIDI conversion as bidirectional. This post explains why.

The obvious direction

Every organ MIDI conversion starts the same way. The organ has key contacts. The contacts make and break electrical connections when keys are pressed. A microcontroller scans those contacts and translates the state changes into MIDI Note On and Note Off messages. This is the MIDI-Out path, and for the 555 it covers 153 inputs across four playing surfaces plus the expression pedal.

This direction is well understood. People have been building organ MIDI encoders since the 1980s. The shift register topology is standard. The firmware is straightforward: scan, debounce, compare to previous state, send MIDI if changed. The input inventory catalogs every contact, the implementation roadmap lays out the build sequence, and the bench probing plan defines the measurements needed to set the firmware constants. The path from organ to MIDI is a solved problem.

The interesting direction

The reverse path takes MIDI messages coming into the ESP32 and converts them into the voltages and switch closures that the organ’s analog circuits already understand. The organ does not know or care that the signal originated from a DAW, a foot controller, or a sequencer. It sees the same electrical conditions it has always seen from its own switches and potentiometers.

This works because the 555’s control surfaces are electrically simple. The expression pedal is a pair of potentiometers. The tab stops are SPST switch closures. The Leslie speed select is a reed switch coil driven by the console switch. Every one of these control mechanisms can be replicated electronically: a DAC generates a voltage that parallels a potentiometer, a relay parallels a switch closure. The organ’s circuits accept these signals identically because the electrical interface is the same.

The output inventory catalogs 28 or more controllable targets. Some are continuous (expression volume, Orbit III VCO pitch, Wah-Wah cutoff, LFO parameters) and need DAC channels scaled through op-amp gain stages. Some are binary (Leslie speed, tab stops, tremolo, vibrato) and need relay closures driven through ULN2803 Darlington arrays. The output stage hardware page specifies the interface for both types.

Why bother?

The practical answer is that the ESP32-S3 is already bidirectional. The same USB, BLE, and DIN-5 interfaces that send MIDI-Out also receive MIDI-In. The same firmware event loop that processes key scan results can process incoming MIDI messages. The marginal cost of adding MIDI-In to a system that already has MIDI-Out is a handful of DAC channels, relay drivers, and op-amps. Maybe $45 in prototype parts.

But the deeper answer is that MIDI-In makes the organ a different instrument.

With MIDI-Out only, the organ is a controller. It sends messages to external gear. Useful, but the organ itself is unchanged. With MIDI-In, the organ becomes a destination. A sequencer can automate the Leslie transitions. A DAW can sweep the Orbit III’s Wah-Wah filter on a timeline. A foot controller can select Orbit III presets without the player reaching up to the tab stops. A second keyboard in another room can play the organ remotely, hearing the sound through the Leslie speaker in the organ’s room with all the room interaction that entails.

The most compelling example is the expression pedal’s pitch bend axis. The lateral slide drives the Orbit III’s Slide circuit through a potentiometer. MIDI-Out captures that gesture as a Pitch Bend message. MIDI-In can inject a voltage at the same Slide circuit input to produce the same pitch bend from an external source. A DAW could record the player’s pitch bends during a performance and play them back through the same physical circuit, with the sound passing through the same analog VCO and the same Leslie speaker. The organ replays its own performance. The round-trip is complete.

The Leslie as first target

The Leslie control page makes the case for the Leslie as the first MIDI-In target, and the reasoning is simple: the Leslie is the most dramatic effect on the organ, the control interface is the simplest (two relay channels driving reed switch coils), and you can verify it works by ear from across the room.

The 555’s Leslie uses a TRIAC-based speed controller with magnetically actuated reed switches. The console’s Off/Slow/Fast switch energizes coils that flip the reed switches, which configure the TRIAC to drive the motor at the corresponding speed. The MIDI interface just needs to energize those same coils. Two GPIO pins on the ESP32, through a ULN2803 driver and a pair of small relays, replicate the three switch states. CC#80 at 0-42 means off, 43-95 means slow, 96-127 means fast. That is the same convention Hammond-Suzuki uses on the Leslie 2101mk2.

The spin-up and spin-down transitions need no firmware ramping. The motor accelerates and decelerates through its own mechanical inertia. The musical character of the Leslie transition is a property of the motor’s torque curve and the rotor’s moment of inertia, and the MIDI interface preserves that character by driving the same motor through the same TRIAC circuit. Switch the relay on, and the horn spins up exactly the way it always has.

What comes after

Once the Leslie is proven, the build sequence adds channels incrementally. The output stage prototyping approach validates each new channel in isolation before connecting it to the organ: DAC output measured with a voltmeter, relay tested with a continuity meter, then connected to the organ one circuit at a time.

Expression volume is next (a single DAC channel replacing a potentiometer). Then the Orbit III CV targets (VCO pitch, Wah-Wah, LFO parameters) once the service manual arrives and the voltage ranges are confirmed. Then the binary targets (tab stops, tremolo, vibrato) as relay channels.

The investigation queue on the output inventory page tracks every unknown that needs bench measurement before a channel can be committed. Some targets are high confidence (expression volume is a potentiometer, and paralleling a potentiometer with a DAC is a textbook application). Some are low confidence (the Orbit III’s Delta Pitch control mechanism is unknown until the service manual is in hand). The build follows the confidence gradient: high-confidence targets first, low-confidence targets after investigation.

The reverse path is the part of this project that turns a MIDI encoder into a MIDI instrument. The organ already speaks voltage. MIDI-In just gives it a second language.