RaceEye Anchor Planner

Photo

Drag anchors to move them. Double-click to add or remove one.

Anchors needed for 98% of the lap

Computing…

Coverage by anchor count

Best pole layout the optimizer found for each count. Solid bars cover 98% of the lap; in TWR the tight-pack test must also pass. Click a bar to load it.

Allow poles on the infield tyre islands

Anchors needed for 98%, by tag mount and method (click to switch):

Anchor layout

On the map: drag an anchor to move it, double-click empty ground to add one, double-click an anchor to remove it. Click a kart to see its radio links.

Tag on the kart

Mount position

Tyre barrier height

Accuracy target (95%)

Radio

Positioning method

Karts one anchor can serve

Blink timing

Line-of-sight range

Range sets how many poles you need. See the sources under Model assumptions, and measure it on site with your own hardware.

Anchor pole height

Ranging noise (1σ)

Karts on track

Karts

Updates per kart

Top speed

Formation

Radio air time used

Live simulation · last 8 s

Fixes delivered

–

Per kart

–

Median error

–

95% error

–

Show radio links of the selected kart

Model assumptions

Scale 6.4 px/m, from the 100 m and 40 m measurements on the screenshot. Walls and the racing line are traced by hand, so treat positions as ±1 m.
Radio: UWB channel 5 (6.5 GHz), 6.8 Mb/s, 128-symbol preamble. A TDoA blink takes 0.25 ms of air; a TWR fix takes 3.0 ms (poll, 4 responses, final). Schedulers use at most 85% of air time.
Range default 40 m. At the −41.3 dBm/MHz emission limit (about −14 dBm EIRP over 500 MHz) and about −94 dBm sensitivity at 6.8 Mb/s, the free-space link budget with 0 dBi antennas gives about 35 m. Field reports for DW3000 modules: “40–50 m for typical settings”, about 50 m at best with 850 kb/s (Qorvo forum); 20–30 m for some users, with Qorvo staff noting that 850 kb/s barely helps under the power limit and recommending an external LNA (Qorvo forum); about 30 m for QM35825 at normal power (Qorvo forum). Pozyx quotes up to 50 m in real industrial sites (Pozyx). Zebra’s NFL system reads up to about 107 m (350 ft) with 22–24 receivers per stadium and high-gain antennas (RFID Journal).
Connection limits. In TWR each kart is a radio node that runs a ranging session with each of its anchors. Qorvo staff say a PANS anchor handles “15 tags at 10 Hz or 150 at 1 Hz” and a tag uses at most 4 anchors (Qorvo forum). FiRa one-to-many allows at most 8 responders per session (Qorvo forum), and the QM33 SDK's QANI demo runs one session at a time (Qorvo forum). The scheduler serves the karts with the fewest usable anchors first and gives everyone 3 anchors before anyone gets a 4th. The pack test parks 20 karts, two abreast and 6.5 m apart, every 12 m around the lap. TDoA anchors only listen, so they have no session limit; a Pozyx gateway, for example, handles 500–5000 positions per second (Pozyx).
TWR ranges to 4 anchors chosen for geometry. TDoA uses every synchronised anchor that hears the blink; anchors sync over the air, so each must be within range of another anchor.
Ranging noise grows to 3× at the edge of range. Beyond 85% of range, packets start to drop.
A ray that passes below the top of a tyre barrier loses 16 dB and picks up ~0.35 m of positive bias. Passing within 60% of the first Fresnel zone costs 3 dB. The driver's body costs 10 dB, and each kart in the way costs 8 dB. Every 6 dB halves the usable range.
Error is 2DRMS (≈95%) from the anchor geometry. The live karts solve every fix with Gauss-Newton on noisy, biased ranges.
Coverage is the share of lap samples (every 1 m, on the centre line and ±1.8 m to each side) with at least 3 anchors and error within the target.