How much field a count is

What one count is worth

An Uno's analogRead has 1023 steps across 5 V, 4.9 mV each. At 3.25 mV per gauss that is about 1.5 G per count. A Pico in Arduino reads 1023 steps across 3.3 V, 3.2 mV, and at 2.1 mV per gauss that is also about 1.5 G. An ESP32 reads 4095 steps across about 3.1 V, 0.76 mV, so a third of a gauss on paper; its ADC jitters by several counts, which is why the sketches average.

Pick your board and a magnet, then slide it away. The field is the same model the TK18 book uses: a neodymium disc, face on, the chip on its axis. It is the best case; a real magnet at an angle does less. For a 10 × 3 mm disc it gives about 460 G at 7 mm, where the straight line ends, 40 G at 20 mm, where the TK18 would switch, and 10 G at 34 mm.

So a neodymium disc is readable from touching distance to a few centimetres, and the useful range for a number is roughly the middle of that: close enough to be well clear of the jitter, far enough to stay on the straight line. The distances depend on the magnet far more than on the chip. Try yours.

A wire is not a magnet

The page this book replaces said contactless current sensing starts here. It does not, with this block on its own.

A straight wire carrying a current makes a field that circles it and falls off with distance: 2 G for every amp, divided by the distance in millimetres. 5 A at 10 mm is 1 G, under one count on an Uno and about the size of the Earth's field, which changes as you turn the board. Pressed against the wire you might see a few counts, and any magnet or steel nearby moves the reading as much.

Current sensors that use a Hall element do it differently. The ACS712 runs the current through its own package, right against the Hall element; clamp meters wrap an iron ring round the wire to gather the field into a gap where the sensor sits. For a number in amps, use a sensor made for it, such as the TK119 INA219 block.