Fix bug #169 - analogWrite by micooke · Pull Request #195 · sandeepmistry/arduino-nRF5

micooke · 2017-09-29T04:37:03Z

wiring_analog_* : fallback to digitalWrite if no available PWM channel (copied from AVR core)
wiring_analog_nRF52.c : convert pwmChannelPins & pwmChannelSequence -> pwmContext to semi-standardise pwm pin allocation and pwm status between nRF51 and nRF52
wiring_private.h : Move pwm structures defines out of wiring_analog_* into wiring_private and make the instantiation of these structure externs instead of statics
wiring_digital.c : disable the appropriate pwm timer when a digitalWrite is sent to an allocated (from analogWrite) pwm pin, and free up the pwm channel for re-allocation

wiring_analog_* : fallback to digitalWrite if no available PWM channel (copied from AVR core) wiring_analog_nRF52.c : convert pwmChannelPins & pwmChannelSequence -> pwmContext to semi-standardise pwm pin allocation and pwm status between nRF51 and nRF52 wiring_private.h : Move pwm structures defines out of wiring_analog_* into wiring_private and make the instantiation of these structure externs instead of statics wiring_digital.c : disable the appropriate pwm timer when a digitalWrite is sent to an allocated (from analogWrite) pwm pin, and free up the pwm channel for re-allocation

micooke · 2017-09-29T04:37:41Z

Tidy-up of #193

sandeepmistry

@micooke Thanks for submitting!

sandeepmistry · 2017-10-02T00:53:15Z

  }
+
+  // fallback to digitalWrite if no available PWM channel
+  if (ulValue < 128)


Can the write resolution be higher than 8-bits? If so, we need to adjust this.

Yes, and on it!

One thing i noticed is that the read and write resolution isnt bounded and for the NRF52 they are stored as signed integers so you could set the read or write resolution to +/-2e9!

If its alright ill add a static int with the half max value and a switch statement to set it in writeResolution so that valid values are used (and so they arent re-calculated on each analogWrite)

For reference;

nRF51 TIMER1 (the only timer currently used for PWM) is either 8b or 16b write resolution

nRF52 is up to 32b write resolution for all 12 PWM channels

sandeepmistry · 2017-10-02T00:55:19Z


  ulPin = g_ADigitalPinMap[ulPin];

+  for (uint8_t i = 0; i < PWM_COUNT; i++)


What do you think about moving this to a private analogWriteOff (or something with a better name) function in cores/nRF5/wiring_analog_nRF5x.c?

Rather than creating another function I could change analogWrite so that it turns off PWM when a duty cycle of zero is passed?

Its up to what you prefer as you are the owner and this affects the API. Either implementation is easy to implement.

micooke · 2017-10-06T02:09:41Z

Ok, can of worms officially opened 💥
nrf51 is done, but i noticed something i should have noticed earlier - the nRF52 only uses 3 out of 12 PWM channels. It has 3 PWM/TIMERS with 4 compare channels each, and in the core each compare channel output pin points to the same pin - so 3 outputs max (currently). This will need some work to fix it up, and luckily now i have the hardware to test it on!

micooke · 2017-10-06T02:10:43Z

Sorry i should have asked, @sandeepmistry , @dlabun - do you want me to fix the nRF52 PWM channel issue here or in a separate PR?

The two issues are somewhat linked in that to enable the extra pwm i need to add another parameter in the PWMContext structure, which will then require me to update how the TIMER/PWM modules are turned off.

dlabun · 2017-10-06T02:42:04Z

Sure, go for it.

micooke · 2017-10-06T02:53:13Z

+  }
+  else
+  {
+    digitalWrite(ulPin, LOW);


Should be HIGH - copy/paste error

micooke · 2017-10-06T02:53:49Z

+  }
+  else
+  {
+    digitalWrite(ulPin, LOW);


The same copy/paste error again

…s with M channels each) * replace static 8b resolution for nRF51 and nRF52 with user-defined up to 16b resolution * enable 6x PWM signals for nRF51 (up from 3) * enable 12x PWM signals for nRF52 (up from 3) * disable / free-up PWM signals when ZERO is written as the analog value analogWrite(PIN, 0)

micooke · 2017-10-09T03:54:21Z

Update :

6x analog/PWM signals enabled for nRF51 (up from 3)
12x analog/PWM signals enabled for nRF52 (up from 3)

The nRF51 now uses TIMER2 as well as TIMER1. As i have said before, the softdevice uses TIMER0 so i didn't touch that.

The nRF5x documentation is lacking pretty severely compared to the AVR chips I have dealt with. Im very surprised but the implementation for the nRF51 uses software to set/clear the hardware pins. It also uses the first channel as the period interrupt, rather than having a separate overflow interrupt (which can be seen in the RBL core and several Nordic source code).

It basically looks like PWM is implemented rather poorly in the nRF51, and i guess that is the reason that the nRF52 has dedicated PWM channels.

dlabun · 2018-02-27T22:58:33Z

@micooke Any issues with merging this fix?

sandeepmistry · 2018-02-27T23:01:11Z

Let’s hold off on this one for a bit ...

micooke · 2018-05-09T22:56:18Z

Hi @sandeepmistry, is there an issue with this PR I haven't addressed or does it just need more testing?

sandeepmistry · 2018-08-19T13:01:28Z

@micooke @dlabun this one needs some more testing.

I'm concerned of the performance implications of calling analogWrite(pin, 0) for every digitalWrite(pin, value) call.

micooke · 2018-08-19T22:11:18Z

The idea for that was actually copied from the Arduino AVR core
https://github.com/arduino/ArduinoCore-avr/blob/master/cores/arduino/wiring_analog.c#L112-L119

Which uses digitalWrite() if 0 or 255 are passed to analogWrite()

micooke changed the title ~~Fix bug #169~~ Fix bug #169 - analogWrite Sep 29, 2017

sandeepmistry reviewed Oct 2, 2017

View reviewed changes

micooke commented Oct 6, 2017

View reviewed changes

sandeepmistry mentioned this pull request Sep 6, 2018

PWM signal from analogWrite does not stop when pin is set to different mode #169

Open

deadprogram mentioned this pull request Oct 4, 2018

NRF52: implement ADC and PWM interfaces tinygo-org/tinygo#32

Closed


		ulPin = g_ADigitalPinMap[ulPin];

		for (uint8_t i = 0; i < PWM_COUNT; i++)

Uh oh!

Conversation

micooke commented Sep 29, 2017

Uh oh!

micooke commented Sep 29, 2017

Uh oh!

sandeepmistry left a comment

Choose a reason for hiding this comment

Uh oh!

sandeepmistry Oct 2, 2017

Choose a reason for hiding this comment

Uh oh!

micooke Oct 6, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

micooke Oct 6, 2017

Choose a reason for hiding this comment

Uh oh!

sandeepmistry Oct 2, 2017

Choose a reason for hiding this comment

Uh oh!

micooke Oct 6, 2017

Choose a reason for hiding this comment

Uh oh!

micooke commented Oct 6, 2017

Uh oh!

micooke commented Oct 6, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

dlabun commented Oct 6, 2017

Uh oh!

micooke Oct 6, 2017

Choose a reason for hiding this comment

Uh oh!

micooke Oct 6, 2017

Choose a reason for hiding this comment

Uh oh!

micooke commented Oct 9, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

dlabun commented Feb 27, 2018

Uh oh!

sandeepmistry commented Feb 27, 2018

Uh oh!

micooke commented May 9, 2018

Uh oh!

sandeepmistry commented Aug 19, 2018

Uh oh!

micooke commented Aug 19, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

micooke Oct 6, 2017 •

edited

Loading

micooke commented Oct 6, 2017 •

edited

Loading

micooke commented Oct 9, 2017 •

edited

Loading

micooke commented Aug 19, 2018 •

edited

Loading