Once every 22 ms

Brief glances

The CC6201 is a micropower part, built for battery equipment. To keep its average current down to about 5 µA it spends 99.8 % of its time asleep. Its datasheet gives the cycle: awake for about 50 µs, then asleep until the next scan, about 22 ms after the last. That is roughly 45 looks a second.

Each time it wakes it measures the field, compares it with its thresholds, sets its output, and goes back to sleep. Between looks, the output holds. The datasheet's word for that is a latch, which is where the old page's idea that this is a latching sensor came from.

What it costs

Run the figure with each of the three passes:

• 100 ms in range. Seen, but not at once: SIGNAL went LOW at the first look after the magnet arrived, 20 ms after it did here, and went back HIGH at the first look after it left. Any pass of a whole period or longer is always seen, up to one period late at each end.
• 30 ms in range. Seen the same way, and held LOW after the magnet had gone.
• 10 ms in range. Not seen at all. The magnet arrived just after one look and left before the next. With a different timing it might have been caught; the point is that nothing guarantees it.

The timing in the figure is illustrative. On a real chip the looks fall wherever they fall, and the 22 ms is a typical figure with no maximum given.

What that means for a project

A door, a lid, a drawer, a tray going back into a printer: the magnet stays put for seconds, and 22 ms of delay is nothing. That is the job this chip was made for.

A magnet flicking past on a fast wheel, a fan blade, or a motor shaft is a different job. Each pass may be shorter than one look apart, and the counter misses turns without any sign that it did. Counting turns works out where the line is. The page this book replaced called this block "the basis of every wheel-speed counter". For a slow wheel, yes. For a fast one, it needs a sensor that does not sleep.