Demo Script, Maker Faire Bay Area 2026

STALE — references the BMI160 IMU and a 90° “verticalize to dry the bed” pitch beat. The IMU is disabled in v3.2 and the 90° mode was retired (fails the wind calc). Update the script before the booth. See firmware/README.md for what the controller actually does today.

Three tiers, matched to the visitor. The 30-second version is for the casual walk-by, most of your visitors. The 5-minute version is for the maker who wants to see the firmware. Read the room.

30-Second Hook (the casual walk-by)

“Wattplot is a DIY solar planter that gives an old panel a second life. A twelve-year-old panel from a decommissioned rooftop is still good for shade and some power. You put it on a raised bed, add a hinge and a smart controller, and it tilts to follow the sun. This is the benchtop version, watch, “

[ press the button ]

”: the panel tilts to follow the sun. The whole thing runs off a twelve-volt battery that the panel charges through a real MPPT charge controller. The smart controller decides when to tilt based on rain, wind, soil moisture, and the time of day. Everything you see is open source, scan the QR, the build guide is free.”

[ hand them a cut-list card ]

“Have a good faire.”

Time: ~30 seconds. Goal: plant a seed. The QR is the ask.

2-Minute Pitch (the curious visitor)

“Wattplot is what happens when you stop choosing between a garden bed and a solar panel, and when you give an old panel a second life instead of sending it to the landfill. The whole thing is sized to fit a salvage panel from a decommissioned rooftop: a twelve-year-old 250 watt panel, still good for 235 watts, in a planter you build from 8-foot lumber. Up to 8 by 5 feet, fits any panel, the bed resizes when you call a preset.”

“What it does is grow food and make power from the same square footage. A full-size build in Phoenix makes about 850 kilowatt-hours a year with a salvage panel, or 2,200 with a new bifacial. A hundred and twenty-four kilos of tomatoes a year. The trade-off is engineered: the panel tilts so the bed gets morning and evening sun, but the harsh midday is shaded, which actually helps the plants.”

“What’s new is the control, a thirty-dollar ESP32 and a real hardware MPPT charge controller. The panel charges the battery through MPPT, the MPPT runs its own charge profile, the ESP32 only reads the battery voltage. The smart controller, “

[ press the button ]

”: is a PI loop on motor current. The IMU tells it the actual panel angle. The decision stack decides what angle to want: rain, wind, soil, time of day. The PI loop just keeps the motor from burning out under wind load.”

“The whole thing is parametric, change one number in the Python file, the 3D model and the sun sim and the wind sim all update. Open source, MIT, every drawing is in the repo. The mini is one ninety-three in parts. The full-size is fourteen hundred with a new panel, about eight hundred with a salvage.”

[ hand them a take-home card ]

“The repo’s on the card. The upcycling guide is in the docs. If you build one, send me photos.”

Time: ~2 minutes. Goal: the visitor understands the project and remembers it tomorrow.

5-Minute Deep Dive (the maker who wants the firmware)

After the 2-minute pitch, but only if the visitor leans in:

“OK, the firmware. It’s ESPHome. Single YAML file in firmware/wattplot.yaml. The PI loop runs at one hertz, deadband is plus or minus two degrees, the current limit trips at two and a half amps, the DRV8871 trips at three-point-six, so you’ve got a one-point-one amp safety margin. The wind-fold logic is in the decision stack: if motor current exceeds a threshold for more than five seconds, fold to zero. If NWS says more than fifty miles an hour in the next hour, preemptively fold to fifteen degrees.”

“The rain capture is a fun one. If the soil’s dry and NWS says rain, the controller holds the panel at fifteen degrees to maximize rain landing on the bed. If the soil’s already wet, the controller verticalizes to ninety degrees to dry the bed out and prevent root rot.”

“Sun position is computed with pvlib. The decision stack is priority-ordered, user override beats safety beats weather beats soil beats time-of-day. The whole thing is sixty-odd lines of decision logic on top of the PI loop.”

“Calibration: the IMU has a zero-tilt offset you have to measure on the bench. The actuator’s stroke gives you a tilt range empirically, the 100mm kickstand actuator hits its end-stop at about thirty-five degrees, which is fine because the full-size build’s twenty-four-inch actuator hits ninety.”

[ point to the screen ]

“The dashboard is just a stand-alone HTML page. Reads the pre-computed sim for the demo, the live version talks to the ESP over WiFi. You can see the panel kWh, the DLI, the soil moisture, the wind, the battery. The kWh number updates as the panel moves. The DLI is a rollup of the morning and afternoon sun on the bed.”

“Want to build one? The hardest part is the bed. The half-lap corners need a router or a chisel. Everything else is a drill and a screwdriver. Take a card.”

Time: ~5 minutes. Goal: the maker goes home and reads the repo.

Asks to plant (every pitch)

Pick the one that fits:

“Scan the QR.” (primary ask, repo is the outcome)
“If you build one, send me photos.” (community growth)
“I’m here all weekend, come back tomorrow.” (return visits)
“Tell a teacher, this is a great classroom project.” (multiplier)

Do not ask for an email unless they offer. Do not push the volunteer-to-help angle. People are at a faire to look, not to be recruited.

What NOT to say

“It’s not finished yet”, say “this is the validation prototype, the full-size build is on the poster.”
“The smart controller is a work in progress”, say “the controller has a PI loop on motor current and a priority-ordered decision stack, the firmware’s in the repo.”
“It’s complicated”, never. Every part of the build is a board and a screw.
“I don’t know, that’s a good question”, say “good question, let me look that up tonight and post the answer in the repo.” Then actually do it. This is how good projects get built.
Long apologies for the unfinished state. Acknowledge, move on.

Body language notes

Stand, don’t sit. Visitors are afraid to interrupt someone sitting. Standing says “I want to talk to you.”
Look at the visitor, not the project. The project is the prop. The visitor is the audience.
Smile at the kid before the parent. Kids drive the parent’s attention span. If the kid is engaged, the parent is too.
Move the panel while you talk. Constant micro-motion is what catches the eye from across the aisle.
Have a piece of paper in your hand. People trust someone who writes things down. Use the cut-list card.

Booth etiquette (3 rules)

One pitch per visitor. If they’re not into it, thank them and hand them a card. Don’t follow.
Don’t pitch the person who’s pitching you. If someone is selling their project to you, listen. Trade cards.
If the table is empty, look up. A booth with no eye contact is a booth with no visitors. When the table is empty, you’re allowed to look bored, but you’re not allowed to be bored.

Run-of-show (3 days)

Day	Open	Talk	Lunch	Talk	Close
Fri	10:00	10:00 – 12:30	12:30 – 1:30	1:30 – 5:00	5:00 – 5:30
Sat	10:00	10:00 – 12:30	12:30 – 1:30	1:30 – 5:00	5:00 – 5:30
Sun	10:00	10:00 – 12:30	12:30 – 1:30	1:30 – 4:00	4:00 – 4:30 (early close)

Two operators minimum. Stagger lunch. Sunday is family day, more kids, more questions, slower talk.