Extend PWM resolution from 8 to 10 bits for low brightness lights.

[s-hadinger]

Description:

The internal ESP8266 PWM implementation is 10 bits resolution but we use only 8 for LED control. This makes a significant difference when activating Gamma Correction (Ledtable 1) since the Gamma curve is very flat at low light levels. Going to 10 bits resolution makes a smoother low light display and better color accuracy at low brightness.

Due to current implementation of PWM, it seems that you cannot go beyond 9 bits resolution with core_2_3_0 or core_2_4_2. Full range is available when switching to core_2_5_0 and when PWMFrequency is below 440 Hz.

To see the difference:

Activate Gamma Correction: LedTable 1
Set color to ligth white: Color #FFFFFF
Set Dimmer to 1: Dimmer 1
Set PMW Frequency to 440 Hz: PwmFrequency 440

With previous version (8 bits), you won't see an increase in brightness from Dimmer 1 until Dimmer 13.

With current version with core_2_5_0, you'll see a first light increase at Dimmer 8, then at Dimmer 11 and so on.

**Related issue (if applicable):

Implementation:

• I internally use a new adaptative resolution Gamma table that uses the same memory footprint as the original table (256 bytes) - hence all table values are below 255. To ease calculation, the table goes from 11 bits resolution in low light up to 8 bits. Gamma correction works no more by looking the Gamma table but calling ledGamma() instead.
• Ability to use full power LEDs. Previously, max PWM value was 1020 (255*4), and was limited to 1008 because of issue Command "Dimmer 100" breaks H801 device #1146. Although it does not make a visual difference, PMW sometimes creates a small buzzing noise in LEDs because of internal vibration. Currently, a LED value of 255 actually drives the PMW to 1023 (max). I patched the 2_3_0 and 2_4_2 to detect a PMW==1023 value and deactivate PWM in this case (this is native behavior in 2_5_0). This means that when LED is off or LED is at full power, there are no more vibrations or noises.
• Added a formal light_subtype LST_MAX value (currently 5) to size all arrays. Also added a check to ensure that light_subtype is never above LST_MAX. If this would accidentally happen, the code would crash because of arrays overflows.
• Added detailed documentation about the flow of calculations from changing a color value to driving the PWM pins.
• Code increase: 352 bytes (core_2_5_0)

Checklist:

• The pull request is done against the latest dev branch
• Only relevant files were touched (Also remember to update changelog.ino file)
• Only one feature/fix was added per PR.
• The code change is tested and works.
• The code change pass travis tests. Your PR cannot be merged unless tests pass
• I accept the CLA.

[s-hadinger]

Yes, the patch is to retrofit a native behavior in 2.5.0. There is no change required in 2.5.0.

The patch is to make sure that a PMW value of 1023 just forces the PIN to HIGH value and deactivates PWM. It was already done for a value of zero.

[s-hadinger]

I forgot to mention in the PR description that it only changes behavior for direct PMW pins.

It has no impact on devices with their own PMW like: SM16716, Tuya, Armtronix, PS 16 DZ.

[s-hadinger]

Thanks.
@arendst Sorry, I forgot to add the description in _changelog.ino.

Could you please add the following or should I do a PR for it?
* Extend PWM resolution from 8 to 10 bits for low brightness lights

[arendst]

I'm mostly offline. Just add it with your next PR 😀