search

adafruit / circuitpython

CircuitPython - a Python implementation for teaching coding with microcontrollers
https://circuitpython.org
Other
4.1k stars 1.22k forks source link

PyPortal RTC/time stops advancing when I2C device connected #1912

Closed bbtinkerer closed 5 years ago

bbtinkerer commented 5 years ago

On the PyPortal if I connect any powered I2C device to side 4 pin connector, the RTC/time stops advancing and just pauses. The time continues again (does not restart, just picks up where it paused) after disconnecting the I2C device's power. The code I used to just check the time behavior as follows, then just power on/off the I2C device:

import time

i = 0
while True:
    i = i + 1
    print("time: {0}, counter: {1}".format(time.time(), i))

I know the devices are correctly connected because I am able to control/read their value. I tried externally powering the I2C devices also, but still had the same behavior.

bbtinkerer commented 5 years ago

I do not think the problem is a software problem. I changed the PyPortal voltage output on the connectors to the 3.3V by cutting the trace 5V trace on SJ1 and solder bridging to the 3.3V. The RTC works now when I2C devices are connected.

dhalbert commented 5 years ago

This is still worth looking at, because the I2C implementation should not hang up the RTC if there's something wrong. https://forums.adafruit.com/viewtopic.php?f=60&t=151839

jerryneedell commented 5 years ago

FWIW - I have reproduced this with a I2C soil sensor on a pyportal -- time.time() stops incrementing. time.monotonic() continues incrementing. Just connecting the I2C device stops time.time() from incrementing - it does not have to be accessed.

dhalbert commented 5 years ago

@jerryneedell Which sensor? And are you applying 5v or 3.3v via the connector?

jerryneedell commented 5 years ago

https://www.adafruit.com/product/4026 with he jumper in the default 5V position Also not that the sensor does not work properly. the temperature values are wrong. I don't want to repeat this test! at 3V it works fine as reported

jerryneedell commented 5 years ago

I hav e the jumper set to 3V for portals I use with I2C -- I have one at 5V still so I used it to check.

jerryneedell commented 5 years ago

I may have damaged the sensor

jerryneedell commented 5 years ago

hmm -- looking at the schematic for the pyportal -- it looks like with the jumper at 5V, it puts 5V on the I2C VCC -- SDA/SCL are pulled up to 3.3 V on the pyporta for its adt7410 temperature sensor, but the I2C device (the soil sensor) will pull them to 5V -- not sure what this does. In any case, my soil sensor temperature values are not working properly any more. I think I'll stop testing this.

jerryneedell commented 5 years ago

the soil sensor that I was giving me bad temperature readings after using 5V I2C on the pyportal still will not work with Circuitpython, but it does work under Arduino!! I'm confused, but relieved.

This sensor was working recently with CP - so it appears something has recently changed with in CP or the seesaw library -- I'll see If I can make any sense of it.

example with bad temperature readings

Press any key to enter the REPL. Use CTRL-D to reload.
Adafruit CircuitPython 4.1.0-beta.0-31-g3863a6ab8 on 2019-06-20; Adafruit PyPortal with samd51j20
>>> 
>>> 
>>> import soil_simpletest
temp: 16383.0  moisture: 338
temp: 16383.0  moisture: 351
temp: 16383.0  moisture: 351
temp: 16383.0  moisture: 351

It is completely unclear how any of this can cause the RTC to stop....

jerryneedell commented 5 years ago

After convincing myself that the soil sensor was not damaged by the 5V I2C I have resumed testing this.

I am now finding that the issue is intermittent. Here is an example where time.time() does increment wit the I2C sensor connected

I am unable to reproduce the issue now but I have reproduced it on other occasions.

It does appear that the time increment is slowed but not always stopped. - even though incrementing it sometimes reports the same time twice before incrementing.


Press any key to enter the REPL. Use CTRL-D to reload.
Adafruit CircuitPython 4.1.0-beta.0-31-g3863a6ab8 on 2019-06-20; Adafruit PyPortal with samd51j20
>>> 
>>> import timetest
temp: 16383.0  moisture: 322
time: 946684816, counter: 1
temp: 16383.0  moisture: 331
time: 946684816, counter: 2
temp: 16383.0  moisture: 332
time: 946684817, counter: 3
temp: 16383.0  moisture: 354
time: 946684818, counter: 4
temp: 16383.0  moisture: 331
time: 946684819, counter: 5
temp: 16383.0  moisture: 334
time: 946684820, counter: 6
temp: 16383.0  moisture: 337
time: 946684821, counter: 7
temp: 16383.0  moisture: 331
time: 946684821, counter: 8
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "timetest.py", line 24, in <module>
KeyboardInterrupt: 
>>>

this is the code I am running


import time

from board import SCL, SDA
import busio

from adafruit_seesaw.seesaw import Seesaw

i2c_bus = busio.I2C(SCL, SDA)

ss = Seesaw(i2c_bus, addr=0x36)
i = 0

while True:
    # read moisture level through capacitive touch pad
    touch = ss.moisture_read()

    # read temperature from the temperature sensor
    temp = ss.get_temp()

    print("temp: " + str(temp) + "  moisture: " + str(touch))
    i = i + 1
    print("time: {0}, counter: {1}".format(time.time(), i))
    time.sleep(1)
jerryneedell commented 5 years ago

I created a dummy I2C device with just 10K resistors pulled to VCC. Runnig this code


import time
i = 0
while True:
    i = i + 1
    print("time: {0}, counter: {1}".format(time.time(), i))
    time.sleep(1)

connecting the "device" to the Pyportal with the 5V I2C you can see that the rime increment stops


time: 946685538, counter: 391
time: 946685539, counter: 392
time: 946685540, counter: 393
time: 946685541, counter: 394

time: 946685542, counter: 395
time: 946685542, counter: 396
 connected "device"
time: 946685542, counter: 397
time: 946685542, counter: 398
time: 946685542, counter: 399
time: 946685542, counter: 400
time: 946685542, counter: 401
disconnected  "device"

time: 946685543, counter: 402
time: 946685544, counter: 403

This does not happen on a PyPortal with the 3,3V jumper in place for I2C

jerryneedell commented 5 years ago

I was not able to reproduce this on a feather_m4_express. I tried both with just the Dummy device with its pull-ups connected to VUSB and with th dummy device and an ADT7410 connected (like it is on the Py Portal. The time.time() kept incrementing and the adt7410 works normally.

jerryneedell commented 5 years ago

hmmm -- if I connect my "dummy" device to the PyPortal before power up -- it takes it along time to boot -- much longer than normal ~10 sec or more

I this can I am seeing time.tme() increment, but it is somewhat delayed -- and slowing down

time: 946684887, counter: 56
time: 946684887, counter: 57
time: 946684888, counter: 58
time: 946684889, counter: 59
time: 946684890, counter: 60
time: 946684890, counter: 61
time: 946684891, counter: 62
time: 946684891, counter: 63
time: 946684891, counter: 64
time: 946684892, counter: 65
time: 946684892, counter: 66
time: 946684892, counter: 67
time: 946684892, counter: 68
time: 946684892, counter: 69
time: 946684892, counter: 70
time: 946684893, counter: 71
time: 946684893, counter: 72
time: 946684893, counter: 73
time: 946684893, counter: 74
time: 946684893, counter: 75
time: 946684893, counter: 76
time: 946684893, counter: 77
time: 946684893, counter: 78
time: 946684893, counter: 79
time: 946684893, counter: 80
time: 946684893, counter: 81
time: 946684893, counter: 82
time: 946684893, counter: 83
time: 946684893, counter: 84
time: 946684893, counter: 85
time: 946684893, counter: 86
time: 946684893, counter: 87
time: 946684893, counter: 88
time: 946684894, counter: 89

so this may be a clue -- the increment starts off working - then slows down over time so it appears to not be incrementing at all.

Note: the long delay only occurs at Power on , not hard RESET or Control-D

jerryneedell commented 5 years ago

and another observation: releasing either SDA or SCL starts the time.time() incrementing again. It only stops if both are pulled up.

Also with just SCL or SDA pulled up - the boot time is normal and it interments normally. pulling up the other signal causes it to stop.

I tried pulling up D3 while SDA was pulled up and it had no impact. Pulling up SCL stopped it.

ladyada commented 5 years ago

make sure the i2c pins are pulled up to 3V not 5V

jerryneedell commented 5 years ago

Right -- this only happens if you connect an I2C device to a "default" PyPortal where I2C VCC is set to 5V on the JST connector

ladyada commented 5 years ago

yeah thats a known issue with the samd51 - it does not like pullups to 5V

jerryneedell commented 5 years ago

A more prominent warning may be needed for PyPortal -- this comes up a lot.

ladyada commented 5 years ago

sure, where do you think it should go?

jerryneedell commented 5 years ago

I guess here https://learn.adafruit.com/adafruit-pyportal/pinouts#i2c-connector-2-18 it says you can set the jumper -- may want say you "should" set the jumper for I2C devices

dhalbert commented 5 years ago

@ladyada Maybe next board rev should change the jumper to default 3.3V.

ladyada commented 5 years ago

the next board rev will have a level shifter (its in the mini version already)

ladyada commented 5 years ago

added!

jerryneedell commented 5 years ago

Should we close this? Just say "don't do that" ;-)

dhalbert commented 5 years ago

Yes, it's "solved". "If it hurts, don't do that." Since the guide has been updated, this is enough. But thanks for testing.

ladyada commented 5 years ago

yep future boards don't have this issue - e.g. the pygamer/pybadge have proper level shifting. things get learned! :D