OneWire not working in CP 3.0.0 rc0

[caternuson]

REF: https://forums.adafruit.com/viewtopic.php?f=60&t=137405

D5 is a special pin on the Itsy, but even using D7 for testing indicates something is not working:

CP 2.3.1

Adafruit CircuitPython 2.3.1 on 2018-05-07; Adafruit Itsy Bitsy M0 Express with samd21g18
>>> import board, busio
>>> ow = busio.OneWire(board.D7)
>>> ow.write_bit(1)

CP 3.0.0 rc0

Adafruit CircuitPython 3.0.0-rc.0 on 2018-06-18; Adafruit ItsyBitsy M0 Express with samd21g18
>>> import board, busio
>>> ow = busio.OneWire(board.D7)
>>> ow.write_bit(1)

[sommersoft]

I think it is beyond OneWire. I ran this test for DigitalInOut, since that is all that shared-modules/bitbangio/OneWire is doing:

Adafruit CircuitPython 3.0.0-alpha.1-737-gd1b3c3c-dirty on 2018-06-22; Metro M4 Express with samd51j19
>>> import board, digitalio
>>> d = digitalio.DigitalInOut(board.D11)
>>> d.switch_to_output(True, digitalio.DriveMode.PUSH_PULL)
>>> d.switch_to_input()
>>>

Same flatline result:

The 3.x change in DigitalInOut that I have a hunch is causing this, is a switch to using hri_port_set_PINCFG_DRVSTR_bit to set the pin config in switch_to_output. Then for whatever reason, gpio_set_function in switch_to_input isn't overriding the pin config.

I'm going to dip into the ASF4/datasheet to see if any documentation supports this.

NOTE: while i'm on a dirty, oldish 3.x, I verified it includes the same digitalio code as RC0. also, for whatever reason, my tags never update anyway. this is post 3.alpha-6 firmware. 🤷‍♂️

[dhalbert]

The DRVSTR bit setting thing was added about eight months ago, but the code was wrong for PBxx pins initially. It worked OK for PAxx pins. That was fixed 2 months ago.

btw if you want to update your tags, you can do git pull --tags or git fetch --tags.

[dhalbert]

@sommersoft How did you decide to do digitalio.DriveMode.PUSH_PULL and give no args to switch_to_input()?

shared-module/bitbangio/OneWire.c` does this:

void shared_module_bitbangio_onewire_write_bit(bitbangio_onewire_obj_t* self,
        bool bit) {
    common_hal_mcu_disable_interrupts();
    common_hal_digitalio_digitalinout_switch_to_output(&self->pin, false, DRIVE_MODE_OPEN_DRAIN);
    common_hal_mcu_delay_us(bit? 6 : 60);
    common_hal_digitalio_digitalinout_switch_to_input(&self->pin, PULL_NONE);
    common_hal_mcu_delay_us(bit? 64 : 10);
    common_hal_mcu_enable_interrupts();
}

which is more like

d = digitalio.DigitalInOut(board.D11)
d.switch_to_output(False, digitalio.DriveMode.OPEN_DRAIN)
# sleep 6 or 60 usecs
d.switch_to_input(digitalio.Pull.NONE)
# sleep 64 or 10 usecs

[sommersoft]

My initial test with OPEN_DRAIN didn't pull the pin high, since I'm using the pin directly. If I were using a OW device, the slave would be pulling it high; at least that is my understanding. So, I forced it with PUSH_PULL.

The no arg switch_to_input is because it defaults to PULL_NONE anyway.

I was really just trying to force the behavior of switching between modes, and being able to capture it with the Saleae.

[sommersoft]

Yep! That was my bad. Using switch_to_input(pull=Pull.DOWN) fixes my test. Disregard my rambling... 😄

[dhalbert]

I spent time on this and have OneWire working, if a bit erratically, with a DS18B20 temp sensor. Two issues:

1. common_hal_digitalinout_set_pull() worked differently in 2.x vs. 3.0. In 2.x it always sets the port direction to input. In 3.0 it just set the pull-up/down/none bits, without forcing the direction to input. common_hal_digitalio_digitalinout_switch_to_input() was depending on the direction change, because it wasn't doing it itself explicitly. (Thanks to @sommersoft for looking around here, which caused me to look at it all carefully.)
2. The timing is flaky, because it depends on short microsecond waits. In shared_module_bitbangio_onewire_read_bit() there is a delay of 6us. There was already a comment that it should be 9 but 9 didn't work. I just commented out the delay completely and reads of the sensor work most of the time, though sometimes I get a CRC error.

Given the strict timing, perhaps using pulseio would be better, if possible.

[tannewt]

Thanks everyone!

[caternuson]

Yep. Looks good:

Adafruit CircuitPython 3.0.0-rc.0-13-gf1248dc on 2018-06-29; Adafruit ItsyBitsy M0 Express with samd21g18
>>> import board, busio
>>> ow = busio.OneWire(board.D7)
>>> ow.write_bit(1)

DS18B20 checks out:

Adafruit CircuitPython 3.0.0-rc.0-13-gf1248dc on 2018-06-29; Adafruit ItsyBitsy M0 Express with samd21g18
>>> import board
>>> from adafruit_onewire.bus import OneWireBus
>>> import adafruit_ds18x20
>>> ow_bus = OneWireBus(board.D7)
>>> devices = ow_bus.scan()
>>> ds18b20 = adafruit_ds18x20.DS18X20(ow_bus, devices[0])
>>> ds18b20.temperature
24.5

DS2413 checks out:

Adafruit CircuitPython 3.0.0-rc.0-13-gf1248dc on 2018-06-29; Adafruit ItsyBitsy M0 Express with samd21g18
>>> import board
>>> from adafruit_onewire.bus import OneWireBus
>>> import adafruit_ds2413
>>> ow_bus = OneWireBus(board.D7)
>>> ds = adafruit_ds2413.DS2413(ow_bus, ow_bus.scan()[0])
>>> ds.IOA.direction = adafruit_ds2413.INPUT
>>> ds.IOA.value
True
>>> ds.IOA.value
False