Does deepsleep work at all

[Thosch42]

Hi all!
I recognized a similar behaviour with my Lopy. The DS-shields arrived yesterdayand I start to work with them. My hardware setup is pretty simple: Expansionboard V2.1A, DS shield V1.0, LoPy, 2000mAh LiPo, attached to the Expansionboard and powered by USB.
I start with upgrading a OTAA LoRaWAN node. It work for a while, then it stopped working with:
Joined yet!!!
Traceback (most recent call last):
File "main.py", line 75, in <module>
File "deepsleep.py", line 156, in go_to_sleep
File "deepsleep.py", line 105, in calibrate
IndexError: list index out of range
MicroPython v1.8.6-724-g2b9ed601 on 2017-07-28; LoPy with ESP32

Cheers,
Thomas

[chumelnicu]

Lopy + DeepSleep Shield after 10-12 cicles of 10 minites deepsleep stop working to0

I use deepsleep.py library

I really think is one bug with deepsleep function

[bucknall]

Hi @Thosch42,

Could you please share which version on the deepsleep.py library you are using? I think myou might be on an older version of the library.

You can find the latest version, here - https://github.com/pycom/pycom-libraries/tree/master/deepsleep

Let me know if this helps,

Thanks!

[Thosch42]

@bucknall It's Version 1.0.0

[Thosch42]

@bucknall My last pull was yesterday, at app. 9.00 MEST, right after I held my Air Mail Bag from Pycom in my hand ;-)

[chumelnicu]

@bucknall that from your link library i use too.

The question is ..Why deepsleep is working 10-12 cycles ..and after that lopy program is stop.
I use Lopy+deepsleep sheeld, powered from USB

[this.wiederkehr]

The relevant lines are:

    pulses = pycom.pulses_get(COMM_PIN, 50000)
    self.uart = UART(1, baudrate=10000, pins=(COMM_PIN, ))
    self.clk_cal_factor = (pulses[3][1] - pulses[1][1]) / EXP_RTC_PERIOD

obviously pycom.pulses_get returned a list not compatible with the indexes in pulses[3][1] - pulses[1][1]

[bucknall]

@chumelnicu Ok thanks for letting me know, I'll investigate this!

[chumelnicu]

@bucknall Thank you !

[chumelnicu]

@this.wiederkehr And ? Solutions is .. ? :)

[livius]

@this.wiederkehr said in Does deepsleep work at all:

returned a list not compatible with the indexes

What do you mean in "not compatible"?

[this.wiederkehr]

I can only guess from the source as there is no documentation for pycom.pulses_get:

pulses[3][1] - pulses[1][1] should get the time difference from the second to the fourth pulse. If e.g. there is no fourth pulse there will be no list entry for pulses[3] which will result in the error 'IndexError: list index out of range'.

The hacky solution would be to wrap this in a try-except block and retry if there is the IndexError ...

[chumelnicu]

@this.wiederkehr Ok .. now speak for human beings now please :)

I dont know how to do that. Can you tell me the code line who is supose to fix that ?

Than you ! :)

[chumelnicu]

This post is deleted!

[this.wiederkehr]

@chumelnicu said in Does deepsleep work at all:

dont know how to do that. Can you tell me the code line who is supouse to fix that ?
Than you ! :)

Exchange the calibration function in deep_sleep.py with the following one, take care on the indentation (sorry I dont have a board at my hand to test right now, so use at your own risk :-)
That is not a fix, that is a hack to catch an error (which should not be there in the first place) and retry the calibration procedure ...

def calibrate(self):
    """ The microcontroller will send the value of CTRL_0 after setting the bit
        and then will send the following pattern through the data line:
           val | 1 | 0 | 1*| 0 | 1*| 0 | 1
           ms  | 1 | 1 | 1 | 1 | 8 | 1 | -
        The idea is to measure the real life duration of periods marked with *
        and substract them. That will remove any errors common to both measurements
        The result is 7 ms as generated by the PIC LF clock.
        It can be used to scale any future sleep value."""


    while True:
        try:
            # setbits, but limit the number of received bytes to avoid confusion with pattern
            self._magic(CTRL_0_ADDR, 0xFF, 1 << 2, 0, 7)
            self.uart.deinit()
            Pin(COMM_PIN, mode=Pin.IN)
            pulses = pycom.pulses_get(COMM_PIN, 50000)
            self.uart = UART(1, baudrate=10000, pins=(COMM_PIN,))
            self.clk_cal_factor = (pulses[3][1] - pulses[1][1]) / EXP_RTC_PERIOD
            break
        except IndexError:
            print('calibration failed')
    if self.clk_cal_factor > 1.25 or self.clk_cal_factor < 0.75:
        self.clk_cal_factor = 1

[chumelnicu]

@this.wiederkehr Ok .. maybe you right but this is beyond me , I think is beter to wait for @daniel and @bucknall opinion or fix solution.

Thank you very much @this-wiederkehr

[daniel]

Hello,

While we find out why sometimes less calibration pulses are received, please use the new library that has a try-except block around the calibration:

https://github.com/pycom/pycom-libraries/commit/074bc7ee0181451720bbeb30d627debafc617ad8

[chumelnicu]

@daniel I will try and feedback tonight

Thx @daniel

[athtest800]

This post is deleted!

[athtest800]

@Camot
Have you fixed the 0.4mA consumption during deep sleep?