The Method

Access to a plant is two problems. The first is legal: a lawful path to nursery-propagated stock. The second is horticultural: actually getting the plant to root and grow reliably enough to hand to someone else. This page is about the second half, and about how we work when the answer is not yet known.

Our propagation bench is a small high-pressure aeroponics rig: a bare-root mist chamber where cuttings hang in humid air and get short, fine bursts of atomized water instead of sitting in wet media. High oxygen at the root zone and no waterlogged substrate is exactly the environment that fights the rot which kills most difficult cuttings. It is not magic, and it is not proven for every plant we care about. Where the science is settled we say so; where it is our own bet, we label it a hypothesis.

The chamber is watched, not just run. We drive it on a Pono OODA loop (observe, orient, decide, act): the sensors report the chamber state, we orient on what changed, decide the next move, and the controller acts, then the cycle repeats. The live-vitals strip elsewhere on this site is that loop's readout, wired to the same telemetry the loop reads. When the feed is stale or absent the strip says the chamber is offline rather than showing a stale number, because an honest blank beats a confident wrong reading.

The bill of materials, honestly. These are real parts with verified specs, and where we caught a common mistake we say what it was and why our pick is different. Prices marked confirmed were read at the source; the rest are ranges to verify at checkout.

Pump (12V DC, about 100 PSI): a SEAFLO 34-Series true 12V DC pump with a built-in pressure switch, about 6A, near $85.99 confirmed. The obvious-looking alternative, the Aquatec CDP-6800, runs on 24VAC, alternating current, so it fails the direct-current requirement outright; do not reach for it here.

Misting nozzles: 0.012 inch (0.3mm) brass anti-drip nozzles with a ball-and-spring check, 10/24 thread (AeroMist ClearFlow, about $6.25 each). Anti-drip matters so the chamber does not keep dripping after the pulse ends.

Accumulator: a 1-gallon pre-pressurized tank, 1/4 inch MNPT, 100 PSI (TankPAC TP-4, about $43.84 confirmed). It lets the pump rest between pulses instead of cycling constantly.

Solenoid: a 12V DC normally-closed valve, 1/4 inch quick-connect. Normally-closed is the fail-safe choice: lose power and the mist stops rather than running open.

The switch, done right: a logic-level MOSFET module (IRLZ44N class) with a flyback diode across the solenoid coil, about $5 to $10. Do not wire a bare SSR straight to a Raspberry Pi Pico GPIO pin; a common SSR-25DD draws roughly 7.5mA on its input LED, which is marginal-to-over-limit for a Pico pin. The MOSFET plus flyback is the correct, cheaper, safer switch, and it protects the pin from the coil's collapse spike. The 6A pump gets its own 12V supply; the Pico only switches the low-current solenoid.

Controller: a Raspberry Pi Pico 2 W (RP2350, WiFi and Bluetooth, 3.3V ADC), about $7.00 confirmed.

pH sensor: a PH4502C board with a BNC probe. Its analog output is 5V, above the Pico's 3.3V ADC ceiling, so it needs a voltage divider (see below).

EC/TDS sensor: a DFRobot Gravity TDS (SEN0244), about $11.90. Its output tops out at 0 to 2.3V, already under the Pico's 3.3V ADC ceiling, so it needs no voltage divider. Adding one would only waste ADC resolution.

Water temperature: a DS18B20 waterproof probe, 1-Wire digital, with a 4.7k ohm pull-up. Being digital, it needs no divider either.

Voltage divider (5V sensor to 3.3V ADC): R1 = 2.2k ohm on top, R2 = 3.3k ohm on the bottom. Vout = 5 x 3300 / 5500 = 3.00V, about 9 percent of headroom under the 3.3V ceiling. Tap between R1 and R2 into GP26/27/28, share a common ground with the sensor, and add about 100nF to ground for a stable read. This divider is for the 5V PH4502C output only, not for the DFRobot TDS.

Three corrections worth repeating, because they are the kind of thing a confident guess gets wrong: the pump must be true 12V DC (SEAFLO), not the 24VAC Aquatec; the DFRobot TDS at 0 to 2.3V needs no divider while the 5V pH board does; and the solenoid is switched by a MOSFET with a flyback diode, never a bare SSR off a Pico pin.

None of this makes a plant grow by itself. It makes a repeatable, watched environment where a cutting has its best odds, and it gives us numbers to reason from instead of guesses. When we say a plant is a good candidate for this rig and we have not yet proven it, that is a hypothesis on this page, and it stays labeled that way on the plant's own page until the cuttings say otherwise.