Make analogWrite(255) generate a PWM of 99.6% #4297
Conversation
Writing 255 on OCRxx generates a PWM of 99.6% (255/256). However, analogWrite(255) sets the pin permanently high, which is equivalent to a PWM of 100% (256/256), so there's a bigger step between analogWrite(254) and analogWrite(255) than any other analogWrite(x) and analogWrite(x+1). This patch changes the range of analogWrite() from 0..255 to 0..256. Additionally, it makes out of range values (<0, >256) be treated always as a digital write rather than "overflowing" as they do now.
|
FWIW, this is how Teensy 3.x implements analogWrite. It's been well received and hasn't ever been reported as causing a problem. |
|
I also think you are wrong. 0-255 are 256 steps, 8bit. So 255 is the maximum. @PaulStoffregen However you could consider to add the >= and <= which makes sense to me. |
|
@NicoHood I'm reading https://www.arduino.cc/en/Tutorial/SecretsOfArduinoPWM and it seems that you may be right IF the timers work on "phase correct mode". If they work in "fast PWM mode" then the range of duty cycles goes from 1/256 (for OCRxx=0) to 256/256 (for OCRxx=255); i.e. the "extra step" that is missing would be at 0, not 255 as I thought (unless that tutorial is wrong). |
|
Confirmed; pins 5 and 6 behave differently from 9,10,11,3. analogWrite(6,256) produces an intermediate value between analogWrite(6,0) (totally off) and analogWrite(6,1) (on for 2/256 of a cycle), because pins 5 and 6 work on fast PWM mode rather than phase corrected mode. However, analogWrite(9,256) just turns off the pin just as analogWrite(9,0) does. Is there a reason for this behavior? Personally I see no point on using phase corrected PWM for everyday applications, and although having a PWM value range from 0 to 255 makes more sense than 0 to 256 it's weird to have each pin have a different behavior. I'm closing this PR since it doesn't offer a correct solution for all the pins (actually, any of the pins; although I was close for 5 and 6). But it'd be nice if there could be a unified PWM behavior for all the timers. |
|
(Nevertheless I agree that adding the >= <= would be a good idea; it doesn't add more overhead than the current ==s do) |
|
Go ahead and open another PR. |
Writing 255 on OCRxx generates a PWM of 99.6% (255/256). However, analogWrite(255) sets the pin permanently high, which is equivalent to a PWM of 100% (256/256), so there's a bigger step between analogWrite(254) and analogWrite(255) than any other analogWrite(x) and analogWrite(x+1).
This patch changes the range of analogWrite() from 0..255 to 0..256. Additionally, it makes out of range values (<0, >256) be treated always as a digital write rather than "overflowing" as they do now.