Pixel() and Dot() classes are now complete. Tested DotStar RGB, NeoPixel RGB, and NeoPixel RGBW, all running at same time on a single ESP32.
Next up: Must update Pixel Example as well as Holiday Lights Project
To Do: Add documentation page for Pixel() and Dot()
Pixel.h now contains Pixel() and Dot() classes.
Dot() class uses more streamlined methods; must next update Pixel() to use similar methods, which will allow Pixel and Dot to be more "interchangeable".
This time using customized interrupts to fill RMT memory on-the-fly.
* Added getChannel() to RFControl
* Add 3rd, optional, boolean argument to RFControl(int pin, bool refTick, bool defaultDrive) to that RMT can be initialized but without the default driver (allows for use of custom interrupt code instead)
Pixels now lets you reserve memory for pixels so that the call to start the RF transmission can be done for multiple pixels at once. However, gain is not as much as expected. May need to revisit if driving a large matrix of pixels is needed.
Constructor now allows you to specify high/low timings for 1-bit and 0-bit, as well as timing for reset delay. Default parameters are included if none are specified.
Also add getPin() to RFControl (which is used by SK68XX Class) as well as boolean operator overrides for both RFControl and PwmPin/ServoPin so that instances can be checked for validity.
Allows you to overlay a carrier wave on the RF Signal - in practice this is only used for IR signals (not RF). Automatically scales frequency to account for 80x difference between APB Clock and Ref Tick Clock depending on which is used. Checks to ensure resulting parameters (high period and low period) are all in bounds (0,65536) and reports an error if they are not.
Added second argument to RFControl(uint8_t pin, boolean refClock=true) to allow choice of Ref Tick (1 MHz) clock or APB (80 MHz) clock. Default is to use 1 MHz Ref Tick.
Also fixed bug in logic that divides clock for ESP32-C3. Factor should be 79, not 80, since divider is apparently configured to divide by factor+1.
Instead of limiting number of ticks to 15-bits (32767), RFControl allows for tick size to be any 32-bit number. If ticks > 32767, RFControl adds repeated LOW or HIGH phases as needed to match full duration. This provides for much more flexibility in creating pulse trains that include very long-duration "spaces" between repeats.
Testing completed for RFControl and PWM on all three ESP32 chip types.
To Do: update RFControl documentation to include total number of usable channels per chip, as well as the alternate version of start();
PWM has now been fully tested and verified with an ESP32 device under Arduino-ESP32 versions 1.0.6 and 2.0.0, and with an ESP32-S2 device under Arduino-ESP32 version 2.0.0. Tests confirmed using both high (5000 Hz) and low (1 Hz or 5Hz) frequencies to ensure timers are correctly configured.
Next Task: Update RFControl routines for 2.0.0 and ESP32-S2 compatibility.
Accounts for new *required* elements of the LEDC channel and timer structures to be set in later versions of the IDF. These elements are not available in earlier versions of the IDF and the program can't be compiled unless ignored.
New IDF parameter in 2.0.0 for LEDC allows PWM signal to be inverted! Need to ensure flags.output_inverted is set to 0!
Also: Deleted old PWM class, which was saved for backwards compatibility. This "breaks" HomeSpan for those using the old PWM class (instead of LedPin).
Also: Added checks to ensure that frequency is achievable (for S2 and C3, the 14-bit duty resolution is insufficient to allow frequencies slower than 5 Hz - this is not a practical limit when using LedPin to drive actual LEDs and lights).
Gregg