01-10-2020, 10:22 PM
(01-07-2020, 01:22 PM)caligari Wrote: Finally, I have tested the button in a PineTime: it is a SPST connected to P0.13 and P0.15.
I can not understand the resistors and VCC drawn in schematics, but I have It working with interrupts activating P0.15 output-high and P0.13 input-pulled-low.
One way of using this configuration would be as follows:
Usually both pins are actively driven low, which will cause zero current draw whether the button is pushed or not.
On a timer interrupt the chip wakes up and briefly pulses one of them (WLOG) P0.15 high, while configuring the other as an input. The button is pressed if (and only if) P0.13 goes high. This causes current draw through P0.15 during the query pulse (because of the pull down R13) and, if the button happens to be pressed, current draw in P0.13's pull down R15 as well. So after briefly waking to query, both are dropped low again.
The series resistors R12 and R14 are pretty typical ESD protection for anything that a user could touch. And I think the pull-down resistors R13 and R15 are typical/necessary for ESD as well (although I know very little about ESD protection circuits!), so that might put limits on what games you could play by making one of the GPIOs pull-up instead of pull-down or letting it float.
Note that the benefit of connecting both sides to a GPIO is that you can prevent current draw through a pull-up or pull-down *while the button is pressed*. If one or the other side was passively pulled to a rail, you'd typically get current draw through a pull-up/pull-down for the duration of the button press. (You could add a mosfet to the pull-down to turn it off and let the input float while the button is pressed to save power---some microcontrollers have software-configurable pull-up/pull-down that would do the same thing---but I think the input can't be left floating for ESD reasons.)
On the other hand, the pinetime design means you get current drawn whenever you query the button, and you need to wake up reasonably frequently (100-250ms?) to do so. But each query can be super-fast. Presumably the engineers did the math comparing the integrated current draw of the queries vs the current draw from the amount of button-depressed time in typical usage. (Or they just copied the circuit from some previous engineer who'd done that math for an entirely different product/use case.)
It's an okay design, but not a clever/great one. Depends a lot on how quickly/power-efficiently you can do the periodic poll.