cpu/esp8266: fix pwm_set func

[gschorcht]

Contribution description

In pwm_set function, switch-on and switch-off times for PWM channels were only determined for the following phase, but not for the current phase. This could result in

• a missing duty cycle when function pwm_set was called before switch-on time of the current phase or
• an extended duty cycle function pwm_set was called before switch-off time of the current phase.

This PR fixes the problem.

Testing procedure

Use test/periph_pwm with following changes in main.c:

-#define OSC_STEP        (10)
+#define OSC_STEP        (1)
-#define OSC_STEPS       (1000U)
+#define OSC_STEPS       (100U)

Flash the application with and without the changes in this PR to compare the results.

make BOARD=esp8266-esp-12x -C tests/periph_pwm flash test

Execute the osci command. Without the changes of this PR, you should be able to observe one extended duty cycle every 9th cycles. That is, pwm_set function is executed in 10th cycle and does not switch off the output port.



With changes of this PR, the output port should be switched on and off also in 10th cycle.

[MrKevinWeiss]

make BOARD=esp8266-esp-12x -C tests/periph_pwm flash test

For your testing procedure you should maybe clarify use term not test (it has happened in a few PRs)

[MrKevinWeiss]

It seems like the pwm osci with those settings have some issues on master and this PR seems to resolve them. Tested with a scope.
It seems the code adds (80315 - 80223) 92 bytes or so, but I don't know how you can reduce so I think it is good.

[MrKevinWeiss]

Thanks, I will have it tested by tomorrow afternoon

[MrKevinWeiss]

I think the signs are messed up. It doesn't work anymore.

[MrKevinWeiss]

Tested again, all good. Please squash. ACK

[gschorcht]

@MrKevinWeiss Thanks for reviewing and testing.