Added support for Input Capture timer mode for frequency and pulse duration measurement

[faunel]

Added support for the Input Capture timer mode.

Why this is needed:

The Input Capture mode allows measuring signal frequency or pulse duration. This is especially useful for tasks that require precise frequency measurement, such as monitoring fan RPM, working with sensors, or analyzing signals. Additionally, a single timer can use up to four channels, enabling the measurement of multiple signals simultaneously.

[burrbull]

I do not see ant sense in additional copy of WithPwmCommon trait.
It is better to just rename WithPwmCommon to something more common.

[burrbull]

This macro can also be joined with previous as it is implemented for same timers.

[faunel]

Yes, I agree. That was code duplication.
I refactored it and added comments.

[burrbull]

This was done to allow connecting several pins on same timer channel (although as for me I've never used it). Is it allowed in capture mode?

[faunel]

I don't understand how this is supposed to work.
I took a lot of code from pwm.rs as a basis. So if it worked there, it should also work in input capture mode.
If you tell me how I can test it, I'll try to do so.

[faunel]

If this is what you mean, then yes, it works:

let (mut tim5, (ch1, ..)) = Timer::new(dp.TIM3, &clocks).capture_hz(48.MHz());
let mut ch1 = ch1.with(gpioa.pa6).with(gpiob.pb4.into_alternate());

But in this case, PB4 will be the active input. That means it will register signals and measure the frequency.
PA6 won't do anything.
In other words, the last pin in the chain is active.
I don't know if this is the correct behavior.

[faunel]

I tested further and got unexpected results.

The pins are as follows:
TIM3 channel1 = PA6 original
TIM3 channel1 = PB4 remap

TIM3 channel2 = PA7 original
TIM3 channel2 = PB5 remap

Using channel 1

In this case, only PB4 is active:

let (mut tim5, (ch1, _ch2, ..)) = Timer::new(dp.TIM3, &clocks).capture_hz(48.kHz());
ch1.with(gpiob.pb4.into_alternate()).with(gpioa.pa6.into_alternate());

If you swap them, the same happens: only PB4 is active:

let (mut tim5, (ch1, _ch2, ..)) = Timer::new(dp.TIM3, &clocks).capture_hz(48.kHz());
let mut ch1 = ch1.with(gpioa.pa6.into_alternate()).with(gpiob.pb4.into_alternate());

If you specify only one pin PA6 for channel 1, it works:

ch1.with(gpioa.pa6.into_alternate());

But when using channel 2, the result is different:

Both pins PA7 and PB5 are active:

let (mut tim5, (_ch1, ch2, ..)) = Timer::new(dp.TIM3, &clocks).capture_hz(48.kHz());
ch2.with(gpioa.pa7.into_alternate()).with(gpiob.pb5.into_alternate());

If you swap them, both pins PA7 and PB5 are still active:

let (mut tim5, (_ch1, ch2, ..)) = Timer::new(dp.TIM3, &clocks).capture_hz(48.kHz());
ch2.with(gpiob.pb5.into_alternate()).with(gpioa.pa7.into_alternate());

[burrbull]

Interesting results.

Both pins PA7 and PB5 are active:

Is this useful or should we prohibit this?

[faunel]

Personally, I don't see any benefit in this and can't come up with use cases for it.
I would remove this entirely:

pub enum CaptureLines<P> {
    One(P),
    Two(P, P),
    Three(P, P, P),
    Four(P, P, P, P),
}

Usually, it should be used like this:

let timer = Timer::new(dp.TIM4, &clocks);
let (_, (ch1, ch2, ch3, ch4)) = timer.pwm_hz(25.kHz());
let mut ch1 = ch1.with(pwm4_ch1);
let mut ch2 = ch2.with(pwm4_ch2);
let mut ch3 = ch3.with(pwm4_ch3);
let mut ch4 = ch4.with(pwm4_ch4);
ch1.enable();
ch2.enable();
ch3.enable();
ch4.enable();
let tim_4 = (ch1, ch2, ch3, ch4);

[burrbull]

What about "complementary pins"?

[faunel]

In the input capture mode of the timer, complementary inputs do not exist.
In this mode, both CC1P and CC1NP bits in the TIMx_CCER register
are used to configure the input:

• on the rising edge,
• on the falling edge,
• on both.