pimoroni TMAG5273 3D Hall Effect Magnetometer Breakout - STEMMA QT

AI context

Dense context for pasting into a model prompt — addresses, pinout, voltages, tolerances and the vendor's reference circuit.

BOARD: Adafruit 6489, a small breakout carrying a Texas Instruments TMAG5273 3D linear Hall-effect sensor (SOT-23-6, three independent Hall channels on X, Y and Z plus a 12-bit ADC, a temperature sensor and a CORDIC angle engine). Adafruit sells two orderable variants on the same board: the A1 with selectable plus or minus 40 mT and 80 mT ranges, and the A2 with plus or minus 133 mT and 266 mT. The pinout and firmware are identical between them and Adafruit's library detects which is fitted. Around the sensor the board adds an AP2127K-3.3 regulator, a BSS138 two-channel level shifter and two 4-pin STEMMA QT / Qwiic sockets. Header strip is supplied loose. Resold by Pimoroni; the maker is Adafruit.
ADDRESSES: 0x35 is the address Adafruit publishes as the default for this breakout. The TMAG5273 data sheet states the part is offered in four different factory-programmed addresses and that a user-configurable register can move it, so a bus scan is the way to be sure which one is fitted rather than trusting the default.
PINOUT: 14 pads on three connectors, each connector numbering its own pads from 1, so the two QT sockets are offset and the offsets are stated here. JP1, the 1x6 0.1 inch header, pads 1-6: 1 = VIN (VCC, the raw supply the user feeds in), 2 = 3V3 (the AP2127K's regulated output), 3 = GND, 4 = SCL, 5 = SDA, 6 = INT. CONN4, the STEMMA QT JST SH 4-pin socket on the left, pads 7-10: 7 = GND, 8 = VCC, 9 = SDA, 10 = SCL. CONN3, the socket on the right, pads 11-14: 11 = GND, 12 = VCC, 13 = SDA, 14 = SCL. Both QT sockets are the same four nets in parallel, not two buses. INT is the sensor's interrupt line, configurable as an input or an output and used to trigger a conversion as well as to report a threshold cross.
SUPPLY: VIN must be given the same voltage as the microcontroller's logic - 3.3 V for a Feather or a Pi, 5 V for an Arduino-class board - because it feeds the regulator and the high side of the level shifter. The sensor itself runs from the regulated 3.3 V rail, and its own VCC range is 1.7 V to 3.6 V per the data sheet. The 3.3 V output on pad 2 can supply up to 100 mA to external circuitry, and the sensor's share comes out of that.
LOGIC LEVEL: SCL and SDA are level shifted, so the board accepts 3.3 V or 5 V logic on the QT sockets and on JP1 pads 4 and 5. Each of those two lines carries a 10 kOhm pull-up to VCC on the host side and a second 10 kOhm pull-up to 3.3 V on the sensor side - the four elements of the R3 10K pack - so a chain of these on one bus already has pull-ups fitted. Adafruit publishes no absolute maximum for the header pins; the sensor's own 1.7-3.6 V VCC is the limit that matters, and it is fed by the regulator, not by VIN.
REFERENCE CIRCUIT: nothing to add. U2 (AP2127K-3.3, SOT-23-5) regulates VIN down for the sensor, C2 10 uF and C3 10 uF are its input and output bulk capacitors, C1 0.1 uF is the sensor's local decoupling, R3 is a 4-element 10 kOhm pack giving both sides of the shifter their pull-ups, Q2 (BSS138, SOT-363) is the shifter itself, and R1 10 kOhm with the green LED D1 is the power indicator.
TRAPS: this is a magnet-field sensor for a magnet embedded in something that moves or rotates - Adafruit states plainly that it is not meant for measuring the earth's field, so a compass built on it will read junk. Magnetic gain and offset correction exist on the chip but the host must manage the values and write them in at boot. The STEMMA QT sockets are wired in parallel, so daisy-chaining does not isolate anything. The two variants are the same PCB: tell them apart by the orderable part number, not by this board's silk. VIN is not the sensor's supply - feeding 5 V to VIN does not put 5 V on the sensor, and tapping the 3.3 V out on pad 2 to power other hardware draws from the same 100 mA budget the sensor uses.