Skip to content
IEMI Labs
CourtOne

Our own circuit boards, in the backboard

Boards we design for the smart backboard, with every chip soldered on them: the RV1126B processor and its memory, the 8 MP top camera chip, two 5 MP corner camera chips, the rim motion chip, the call lights, the shot clock digits, the Wi-Fi module, the security chip and the mains supply. One rigid-flex piece lies in the top rail with the top camera at its window; a corner board stands in each side rail, a rim board sits in the ring box, light sections line the frame’s slots and the clock box holds the digit boards and the power board. The same set goes in both backboards. No bought modules, LED panels, ribbons or adapters.

Stage 1 of 6 done: shape and placementNot yet ready to manufactureUpdated 4 November 2026 · rev A-0

The board in 3D

Every part at its real size, in its real place. Turn it, zoom in, switch between the boards in the backboard and laid flat, and click any part to see what it is. The chip bodies are simple blocks until each part's own 3D model is added.

Loading the 3D board…

Drag to turn · scroll to zoom · right-drag to move · click a partDrag to turn · tap a part

Where everything is in the backboard

Top cameratop rail, middle

A bare Sony IMX415 8 MP camera chip on a small head off the main board, under an M8 lens whose front sits just inside the front face, tilted 51° down: the ball in and through the ring, the ring’s front edge, goaltending, the free-throw line and the shooter’s feet on the arc.

Corner camerasside rails, 700 mm up

In each side rail a corner board with an ESP32-S3 and a bare OmniVision OV5640 5 MP chip on a head, looking out of the side face 30° to the court and 30° down: the shooter’s feet in the corner and at the wing for three points. Frames go to the main board only round a shot.

Processor, radio and securitytop rail

The RV1126B with its memory on the main board, a heat block into the aluminium top rail, the Wi-Fi module with its antenna behind a radio window in the back face, the security chip that signs every call, the latched microSD card with our software, and a tamper sensor held by a magnet in the service cover.

Rim sensorring box

A small board on the ring plate with a motion chip that feels every rim and net hit through the steel: the shot clock reset on a touch of the ring, a lodged ball and a player hanging on the ring.

Call lights and shot clockframe and clock box

Twenty-nine light sections over the frame’s slots: red all round at the end of a period, yellow along the top at the shot clock signal. In the clock box, two digit boards show the seconds, and tenths under 5 s, on both faces.

Mains powerclock box

A power board between the digit boards takes the gym’s shot-clock feed into a push-in terminal, through a fuse to a soldered Mean Well supply, with a milled slot between the mains and 5 V sides. Only 5 V comes down to the backboard.

The backboard boards laid flat, both sides, every part at its real size

Laid flat: the main board with the top camera head; the corner board with its head; the rim board; one light section; one digit board; the power board. Tap the drawing to open it full size.

Design review: checked before we order

Every board-level number CourtOne's jobs depend on, checked against its target with the source of each number, using the rules every one of our boards shares (products/_shared/board-rules.mjs). Fixed means a lesson changed this board; to prove means it passes on paper and is measured on rev A before boards are ordered.

No check fails24 pass · 4 fixed by the review · 13 to prove before we order · 0 failDesign review of 4 November 2026
AreaCheckOur boardTargetStatus
CameraTop camera: a shoe (0.1 m) on the arc 30° off centre at 8.15 m
M8 lens, 80° × 50.6°, tilted 51° down: about 48 px per degree across 3840 px. Three points from the top of the arc, made baskets, the shot clock reset on the ring and goaltending all come from this camera; the shooter's foot on the arc is the smallest thing at the farthest place. The ball at the ring is 1 m away and 24 cm across, so 30 fps (23 cm of ball travel at 7 m/s) is enough to see it pass through the ring.
33.7 px≥ 20 pxpasses
CameraTop camera: smear at 11 km/h, 2 ms exposure
A rolling shutter skews a fast target: check the readout time.
6 mm≤ 1/3 of the shoe (33 mm)passes
CameraTop camera: frame rate the calls need
A standard mode of the IMX415.
30 fps at 3840 × 2160chip: 30 fps at 3840 × 2160passes
CameraTop camera: data rate vs its 4-lane link
30 % spare on the link.
2.49 Gbps (10-bit)≤ 80 % of 3.564 Gbpspasses
CameraRight corner camera: a shoe (0.1 m) at the wing, 45° on the arc at 7.39 m
M8 lens, 60° × 46.8°, rows cropped to the floor band: about 43.2 px per degree across 2592 px. Feet on or behind the three-point line in the corner and at the wing. The sensor crops to the 800 rows where the floor is, and the ESP32-S3 keeps the last 2 s of JPEG frames; when the top camera sees a shot leave the hand, the main board asks for the four frames round the release (about 0.5 MB, 1 s over the link), before the ball reaches the ring.
33.5 px≥ 20 pxpasses
CameraRight corner camera: smear at 7 km/h, 4 ms exposure
A rolling shutter skews a fast target: check the readout time.
8 mm≤ 1/3 of the shoe (33 mm)passes
CameraRight corner camera: frame rate the calls need
A standard mode of the OV5640.
10 fps at 2592 × 800chip: 15 fps at 2592 × 1944, 30 fps at 1920 × 1080, 60 fps at 1280 × 720passes
CameraRight corner camera: data rate vs its 1-lane link
57 % spare on the link.
0.17 Gbps (8-bit)≤ 80 % of 0.384 Gbpspasses
CameraLeft corner camera: a shoe (0.1 m) at the wing, 45° on the arc at 7.39 m
M8 lens, 60° × 46.8°, rows cropped to the floor band: about 43.2 px per degree across 2592 px. Feet on or behind the three-point line in the corner and at the wing. The sensor crops to the 800 rows where the floor is, and the ESP32-S3 keeps the last 2 s of JPEG frames; when the top camera sees a shot leave the hand, the main board asks for the four frames round the release (about 0.5 MB, 1 s over the link), before the ball reaches the ring.
33.5 px≥ 20 pxpasses
CameraLeft corner camera: smear at 7 km/h, 4 ms exposure
A rolling shutter skews a fast target: check the readout time.
8 mm≤ 1/3 of the shoe (33 mm)passes
CameraLeft corner camera: frame rate the calls need
A standard mode of the OV5640.
10 fps at 2592 × 800chip: 15 fps at 2592 × 1944, 30 fps at 1920 × 1080, 60 fps at 1280 × 720passes
CameraLeft corner camera: data rate vs its 1-lane link
57 % spare on the link.
0.17 Gbps (8-bit)≤ 80 % of 0.384 Gbpspasses
CameraImage processor load, all cameras
3 cameras into one image processor.
290 Mpx/s≤ 360 Mpx/s (RV1126B ISP)passes
CameraCameras in step for timed calls
A three-point call joins the shot release seen by the top camera to the feet seen by a corner camera; both sides of the court use the same pulse. A real-time (PREEMPT_RT) kernel is asked for in the quote (lesson: sync).
the corner boards timestamp each frame on a frame-sync pair from the main board, through the corner leadthe corner frames and the top camera on one clockpasses
CameraAI engine time per frame
The ball model runs on a crop round the ring and the net; the arc and the corner frames share the engine. Measured on the dev kit.
3 TOPS for the top camera and the corner framesball and ring model at 30 fps on a crop round the ring; person and foot model at 10 fps on the arc and on the corner framesto prove on the dev kit
PowerSupply
About 12 W in play (RV1126B board 2.7 W, two corner boards 1.6 W, rim board 0.3 W, both clock faces showing red digits 7.2 W) and 34.5 W at the end of a period, when the red rows run all round at the 0.9 A chain cap with the yellow row on (22.5 W, est.): 69 % of the supply.
the gym's shot-clock mains feed into a push-in terminal on the power board in the clock box, a PCB-mount Mean Well IRM-60-5 (5 V 10 A, 50 W)≤ 70 % of the supply (35 W)passes
PowerPower cut
The scorer's table keeps the official game clock; cutting the shot-clock feed mid-play is part of the rev A tests.
no cells: a power cut stops the board; every call is written to eMMC as it happens and signedno lost call in a power cutto prove on the dev kit
HeatRV1126B temperature, a full gym in summer, 32 °C at the top of the backboard
The top rail sits 3.9 m up, where the gym is warmest, out of the sun; counted at 36 °C on the aluminium. 2.2 W through a copper heat block and gap pad from the processor to the top rail's top wall (4 K/W, est.).
45 °C≤ 85 °C (rated)passes
HeatTop camera chip away from the processor
A warm chip adds noise in a dim gym.
45 mm from the RV1126B≥ 30 mmpasses
HeatMains supply in the closed clock box
The clock box is closed aluminium; the module's temperature in the box is logged on rev A.
IRM-60-5 at about 12 W in play (24 % load) and 35 W for 3 s at a period endinside the module's rating at 50 °C in the box, no deratingto prove on the dev kit
ShockShock and vibration
Balls hitting the board and the ring all game, dunks and players hanging on the breakaway ring, the backstop swung up for storage (est.). The rim board feels the ring through the plate; the other boards sit in the frame. MEMS clocks are fitted anyway and every card socket is latched.
200 g (est.)standard partspasses
SoftwareStarts our software from the microSD card, no setup
Owner requirement: power it from its own supply and our software runs from the card. Every board starts from its own card: the main board loads Linux and our program (the eMMC holds no system copy (it ships empty, so the card always starts first: Rockchip tries the eMMC before the card); the quote asks for U-Boot and a device tree from Rockchip's RV1126B SDK); the maker flashes our small loader into the corner and rim boards' ESP32-S3, which start our program from their cards. Updates are signed and never run during a game.
microSD socket with a hinged, latched lid (Amphenol 101-00660 class), fitted on every board: our software starts from the card; microSD socket with a hinged, latched lid (Amphenol 101-00660 class): the ESP32-S3 loads our program from the card; microSD socket with a hinged, latched lid (Amphenol 101-00660 class): the ESP32-S3 loads our program from the cardcard socket fitted on every board, latchedpasses
SoftwareCard power switch and protection
A card that hangs must be restarted by cutting its power, not the whole board; the card slot is the one place a person touches the electronics (lesson: sdstart). Change: U90, U92, U94, U91, U93, U95 added (4 Nov 2026, CricketOne card-start review).
3 power switches and 3 ESD arrays for 3 card socketsevery socket: switched card supply, ESD on all 6 linesfixed by the review
SoftwareThe card wins the start
Rockchip's start order is SPI flash, eMMC, card, USB: a written eMMC starts before the card. Proved on the dev kit: power on with a card in and the eMMC empty (lesson: sdstart). Change: eMMC ships empty and the loader reads the card first (4 Nov 2026; it used to say the eMMC kept a copy, which would have started instead of the card).
eMMC shipped empty; our loader reads the card firstthe processor starts from the card every timefixed by the review
SoftwareCard grade
Consumer cards wear out and fail hot; the image is read-only with a separate data partition. Proved: 1,000 power cuts during writes on rev A (lesson: sdstart).
industrial pSLC microSD (Kingston Industrial or Swissbit S-56u), −40 to 85 °Crated for the heat, cold and writesto prove on the dev kit
SoftwareClip buffer for an unsure call
The rest of the memory holds Linux, the program and the AI model (lesson: clip).
3.3 s of 8-bit 30 fps video in 40 % of the 2 GB memory≥ 3 spasses
RadioWi-Fi/BT module clock
The AP6256 runs from its own 37.4 MHz reference clock; a wrong or missing clock means no Wi-Fi or Bluetooth at all (lesson: wificlock).
37.4 MHz MEMS oscillator (Y2)37.4 MHz next to the AP6256passes
RadioWi-Fi and BT from inside the aluminium frame
The frame is a closed aluminium tube; the bench Pi Zero 2 W had no opening at all. The window faces the gym behind the board; the range to the scorer's table and the other end is measured on rev A. Antenna outside the metal: the Ø20 PC radio window in the top rail's back face (rev T) (lesson: enclosure). Measured on rev A (lesson: antenna).
AP6256 module, its chip antenna at the main board's back edge, 2 mm behind the Ø20 radio window in the top rail's back facecertified module; the scorer's tablet at the side of the court and the other backboard 26 m awayto prove on the dev kit
LayoutFast lines on a solid ground plane
The layout rules go to the maker in the quote; the routed board's design rule check (0 copper errors, every keep-out kept) and the maker's impedance coupon prove them (lesson: planes).
RV1126B memory; 5 camera MIPI links (100 Ω pairs); USB (90 Ω); antenna feed (50 Ω, keep-out on every layer); card lines (length-matched, no vias under the socket)ground on inner layer 1, supply on inner layer 2, no vias in pads, impedance lines routed firstto prove on the dev kit
MoistureConformal coating
Gyms are cleaned and mopped, condense when a cold gym warms up in winter, and the clock box carries mains: all boards get the same coat. Asked for in the quote (lesson: moisture).
coatedcoated, sensors, sockets and antennas maskedpasses
FitBoard inside the product CAD, every part checked in 3D
The board is checked on its own; the rev T CAD run places it in the product and checks every part in 3D (lesson: fit3d).
not yet in the product CADnothing touchesto prove on the dev kit
CameraEvery call in a camera's view, from where each lens sits
The ball in and through the ring, the ring's front edge, goaltending above the ring, the shooter's feet at the top of the arc and 30° off centre and the free-throw line from the top camera; the wing and both corner spots from the corner cameras (the rev O inView targets in cad/courtone-cad.mjs).
top camera worst 25.8° across, 22.7° up-down; corner cameras worst 20.2° / 4.5°inside 80 × 50.6° and 60 × 46.8°passes
CameraTop camera window (lesson: window)
The rev O Ø16 disc was never checked against the calls: the net is 74° below level, so a ray from the lens leaves the face far below the lens. An M8 lens (rim Ø8) with its front 0.5 mm inside the face, low in the rail, and a slot down to 1.5 mm above the rail's bottom edge passes every call with 1 mm to spare; the head still fits under the top wall. Change: CAD rev T: top camera window Ø16 → 18 × 39 slot, z 1,018.5 to 1,057.5; M12 → M8 lens (4 Nov 2026).
the ball through the net is 77° from the front face with the 3° margin; the view uses -6.7 to 6.7 across and -28.3 to 8.2 mm up-down of the window centrethe rev T window, 18 × 39 mm, inside the top rail's front facefixed by the review
CameraCorner camera windows (lesson: window)
In rev O the corner cameras looked through Ø12 windows in the front face, turned 55° out: the corner 0.6 m in front of the board is 84° from the front face, so no window in that face could pass it. The side face looks 30° away from the corner calls instead, and the steepest call is 61° from it. Change: CAD rev T: corner windows moved from the front face to the side faces, 24 × 22 at y 22, z 700 (4 Nov 2026).
in each side rail's outer face, 24 × 22 mm; steepest line of sight 61° from the face; margin 0.3 mmevery corner call through the window, ≤ 70° from the facefixed by the review
FitCamera heads in 3D
Each head is posed on its camera's axis and every corner checked against the rail walls and its board (lesson: fit3d); the printed brackets that hold the heads go in the rev T CAD with the boards.
top head 0.3 mm behind the wall, barrel 5.7 mm, hinge 29.2 mm; corner heads 0.3 mm, barrel 6 mm, hinge 26 mmholders ≥ 0.3 mm off the wall, heads inside the rail, hinges within their flexto prove on the dev kit
CameraCorner frames over the wire
The corner boards do not stream: they keep 2 s in PSRAM and send only the release frames, so the link stays a simple RS-485 pair. The UART rate and the call time are measured on the dev kit.
4 frames of a 2592 × 800 JPEG crop round each shot release, about 0.5 MB, about 1 s at 4 Mbpsthe three-point call before the ball reaches the ring (about 1 s)to prove on the dev kit
LightCall lights round the frame
The bench used NeoPixel strips; our sections sit where the strips were, in five chains with the brightness capped at 0.9 A a chain. Their brightness from the far end of the court is checked on rev A.
29 light sections of 14 RGB LEDs (406 per backboard) over the 12 mm slots: red all round, yellow along the topFIBA 1.1.6 and 1.1.7: red at the end of a period, yellow at the shot clock signal, seen from the court; RGB call colours only in club playto prove on the dev kit
LightShot clock digits
The bench used two bought 64 × 32 LED panels on ribbons; our digit boards carry only the LEDs the digits need, on a latched lead from the power board. Readability across the court is checked on rev A.
two digit boards, 2 digits of 130 mm and a point in 100 red LEDs each, 4 mm behind the windowsFIBA 5.6: seen from the court and the scorer's table; seconds and tenths under 5 sto prove on the dev kit
PowerMains on the power board
The bench used an enclosed LRS-50-5; on our power board a PCB-mount module is soldered with a slot under its input. The creepage is measured from the routed board, and UL 62368-1 is still to be done.
push-in 3-pole terminal, 2 A slow-blow fuse and varistor, PCB-mount IRM-60-5 (Class II), 2 mm milled slot and 6 mm keep-out between the mains and 5 V sides; the feed in through a gland in the box floor (rev T)creepage and clearance per IEC 62368-1 (est. 6 mm reinforced at 250 V); the gym's electrician wires the shot-clock feedto prove on the dev kit
FitLeads along the frame
Every lead is latched and runs inside the frame tube; the rim lead leaves the ring box top and enters at the bottom rail's back face.
clock box to the main board, 5 V: 0.6 m of 18 AWG, 0.08 V drop at 6.5 A; main board to each corner board: 1.16 m of 24 AWG, 0.07 V drop at 0.35 A; main board to the rim board: 2.05 m of 24 AWG, 0.05 V drop at 0.15 A; main board to the farthest light chain (bottom), 5 V and data: 1 m of 22 AWG, 0.1 V drop at 0.9 Aclock lead ≤ 0.15 V; corner and rim ≤ 0.25 V; light chains ≤ 0.3 Vpasses
SecurityService cover tamper sensor
The bench used a plunger switch; the Hall switch needs nothing to press. The switch point is measured on rev A.
DRV5032 Hall switch on the main board, a Ø10 × 5 magnet on a printed boss under the service cover 13 mm away (about 22 mT, est.)cover opened = logged and signedto prove on the dev kit

Before we order the first boards, we prove these on the RV1126B development kit or in the rev A tests:

  1. AI engine time per frame
  2. Power cut
  3. Mains supply in the closed clock box
  4. Card grade
  5. Wi-Fi and BT from inside the aluminium frame
  6. Fast lines on a solid ground plane
  7. Board inside the product CAD, every part checked in 3D
  8. Camera heads in 3D
  9. Corner frames over the wire
  10. Call lights round the frame
  11. Shot clock digits
  12. Mains on the power board
  13. Service cover tamper sensor

Where the design stands

  1. 1Shape and placementdone

    Main board with the top camera head, two corner boards with their heads, the rim board, the light sections, the digit boards and the power board in the backboard’s shape, all 182 parts placed at their real size, every call checked from where each lens sits, and KiCad design rule check (0 violations).

  2. 2Schematicnext

    Every wire between the chips, from the V1, S1 and C1 circuits.

  3. 3Footprintsnext

    The exact solder pads under each chip.

  4. 4Routinglater

    Copper tracks on all layers and the flexes, with the mains spacing on the power board.

  5. 5Files for the makerlater

    Gerbers with copper, drill, mask, paste, the rigid-flex drawings, the lead drawings and the assembly parts list. These go to the board maker.

  6. 6Rev A boardsMar 2027

    Made, assembled and tested on our fixture, then on one court.

Design files (rev A-0)

These are the stage 1 files: the board's shape and where every part goes. A board maker needs the stage 5 files to build it.

Who can make it

Each backboard set is two rigid-flex designs (the 8-layer main board with its camera head, and the 4-layer corner board with its head, twice), a rim board, 29 light sections, two digit boards and a power board, with the leads made to length; a court takes two sets. It needs a maker who builds rigid-flex boards and fine-pitch ball-grid chips, and who can test a mains board. We plan three quotes once the stage 5 files are ready: two from China for price, and one from the US for speed.

MakerWhereRigid-flexHDI / layersAssemblyFit for us
PCBWayfirst quoteShenzhen, Chinayes, 6 to 16 layersHDI up to 32 layers, 0.35 mm BGA pitchyes, turnkey, incl. BGA and X-rayBest fit for rev A: the rigid-flex board and its assembly in one order
NextPCBsecond quoteShenzhen, ChinayesHDI, 6 to 32 layersyes (PCBA)Second quote, to compare price and lead time
ALLPCBChina (a network of partner factories)yes, through partner factories1 to 30 layersyesThird quote; check which factory builds it
JLCPCBShenzhen, Chinano (flex only: 1, 2 and 4 layers, up to 490 mm)HDI added Sep 2026yes, incl. BGANot for the one-piece board today. Good for the test fixture, simple boards, and the fallback of separate rigid boards joined by soldered flex
Sierra Circuits (ProtoExpress)US quoteSunnyvale, California, USAyesyesyes, in-house, incl. BGA; turnkey in about 5 daysUS option: fastest turn and engineering help, at a higher price
CirexxSanta Clara, California, USAyesyesyes (turnkey or consigned)Second US option
Advanced CircuitsAurora, Colorado, USAyescheckyesUS quick-turn; check its rigid-flex assembly
EpecNew Bedford, Massachusetts, USAyescheckyes; also builds battery packsUS option that can also make a certified LiFePO4 pack

From each maker's own capability pages, checked 3 November 2026. Limits change, so each quote confirms them for our exact stack-up.

Full design notes

All 182 parts, what is on each board, every connection, heat, power, the factory test, cost and timeline.