RESEARCH · NOT A PRODUCT
The golf ball that finds itself
In 2023 we built a circuit board small enough to live inside a golf ball, then worked out what it would take to put one in the middle of a real one. It is not a product of ours and never will be — so the whole thing is published: the bench footage, the drawings, and three findings that surprised us.
WHAT GOES INSIDE
Seven parts, 3.45 grams
A golf ball is a spring with a skin on it. Putting electronics in the middle means taking the same weight of rubber out — and putting it back in a way the ball cannot tell apart.
THREE THINGS WE LEARNED
The physics is the hard part
You do not add electronics to a golf ball
You remove a sphere of rubber and put back something that weighs exactly the same. A legal ball is 45.93 g in 42.67 mm, which works out at 1.13 g/cm³ — almost precisely the density of the core it is mostly made of. Take out an 18 mm sphere and you have 3.45 g to spend. Not a target. A ceiling.
Nothing small enough can measure the hit
A driver takes the ball from nought to 70 m/s in 450 microseconds — a peak near 28,500 g. Every accelerometer that fits is saturated by a factor of a thousand or more. So you do not measure the impact. You treat it as a starting gun and reconstruct the shot from the silence afterwards, when the ball is weightless and the only thing it can feel is drag.
Spin hides in a compass, not a gyroscope
A wedge can spin at 10,000 rpm — 60,000 degrees per second, fifteen times what any gyroscope will read. An accelerometer at the centre feels nothing, because centripetal force scales with radius and the radius is zero. But a spinning ball watches the Earth’s magnetic field sweep past it once per turn. Rate falls out of the frequency, so there is nothing to saturate, and it works from the exact middle.
CHARGING A SPHERE
A ball has no this-way-up
There is no room for a socket, so power has to arrive through the cover. But a single charging coil only works when it happens to line up — and a ball dropped in a cup lands however it likes. The answer was to stop asking the ball to be clever and put three coils in the charger instead, at right angles, taking turns.
BENCH FOOTAGE · APRIL 2023
It actually worked
These are recordings from the workbench, not reconstructions. The board talked, paired, took a firmware update over the air and ran on microamps — which is the boring half of the problem, and the half that has to be finished before the interesting half is worth starting.
BLE 5 · 2 min
Sending data packets
The board streaming sensor packets over Bluetooth to a host. This is the plumbing everything else depends on — if a ball cannot get its numbers out, none of the rest matters.
Bonding · 3 min
Pairing and authorisation
A ball proving it is the ball it claims to be before a phone will trust its data. Unglamorous, and the difference between a scorecard you can believe and one anybody can forge.
Secure DFU · S132 · 2 min
Firmware over the air
Updating the firmware wirelessly. Once a ball is sealed inside urethane there is no port, no button and no way back in — so this had to work before anything else was worth building.
Li-ion · 1 min
Charging
Charge and discharge behaviour on the bench. A golf ball has nowhere to put a connector, so this is the only way power ever gets in.
µA-level · 1 min
Power draw
Measuring current in each state. The number that decides whether a ball lasts a season or a weekend is the one it draws while doing nothing at all.
WHY THIS IS PUBLIC
Not a product — not from us
The board works. The design behind it is worked through properly — the mass budget, the materials, the assembly sequence, the charging, all of it drawn and costed.
We are not building this ball. It is published so it is useful to somebody who might: a materials engineer, a ball manufacturer, a student after a genuinely hard problem. Everything above is real, reproducible, and yours.
There is hardware being worked on at The Caddy, and it is not this. It will be announced when there is something to announce.
Take it, argue with it, build on it. All of it can be shared — not published to a repository, but sent to anyone with a reason to want it:
- Firmware — the beacon application running in the videos above. C, nRF5 SDK, with the secure bootloader and the over-the-air update path.
- Schematic and bill of materials — the 2023 board as built, and the revision that came out of this study.
- CAD — parametric, so changing the pod diameter rebuilds the whole assembly.
- The working — every figure on this page comes from a script that reproduces it. Mass budget, impact, spin, materials, charging, assembly.
If you want any of that, or you think one of the drawings is wrong, get in touch. Telling me which part you care about is enough — there is no form to fill in.
MEANWHILE
The app is real and shipping
Everything the ball would have fed into already exists: scoring, AI swing analysis, a coaching plan built from your own rounds, strokes gained, tournaments and an Apple Watch companion. Range Coach already measures your practice — it just uses a phone and a watch instead of a ball.