GY-87 micropython library

robert-hh

@tttadam I justz received a sample of the board that you have linked. Three observations:

the HMC5883 does not show up in the i2c.scan(), even if the address is ping'ed.
the bmp180 returns nonsense-values.
the pull-up resistor values on the board are pretty low,
effectively 1.1kOhm. I will look tonight into the signals with an oscilloscope to verify the signal quality.

So all-in-all, the board might just be crap.

tttadam

@robert-hh ohhh, I didn't even thought about that. But could be. Can you recommend some other sensor?
pytrack has built-in gyro. what i need pressure, and compass.
(The weather ballon will have a camera, and when it took a picture, I would like to know the direction)

robert-hh

@tttadam So, looking at the device and the wiring it seems clear why the HMC5883 is not visible in the scan: it is connected to the MBU6050 slave interface. Still, according to the MPU6050 data sheet, it should be visible, because at start-up the MPU6050 connects the main I2C interface to the aux interface.
The I2C signals are fine.

However, on the board I have the BMP180 is broken. The temperature sensor always returns 229.3°C. At least that helped me to find a signed/unsigned bug in my driver.

tttadam

@robert-hh said in GY-87 micropython library:

BMP180

I just checked the bmp18. It's seems something off with the altitude...
(1028 is the current QNH in Budapest)

>>> bmp.sealevel = 1028
>>> bmp.altitude
51.46274
>>> bmp.pressure
1021.34
>>> bmp.temperature
24.2
>>> si7021.readTemp()
22.92749
>>> my_gps.altitude
120.3

So you have any guess how Can I get the compass data?

robert-hh

@tttadam The altitude is calculated from the actual pressure and what you enter as sea level pressure, roughly 8.5m/hPa. You can see it in the python script.
The compass chip should be available when the I2C-bus is set to pass-through. I'm trying it at the moment.

robert-hh

@tttadam So I found a way to talk to the compass, with the following sequence:

from machine import I2C
i2c=I2C()
i2c.writeto_mem(104, 107, 128) # reset MPU6050
i2c.writeto_mem(104, 107, 0) # disable sleep
i2c.writeto_mem(104, 55, 2)   # enable I2C passthru
i2c.scan()

The scan should show you another I2C address. In my case it is 13, which deviates from the HMV5883 data sheet.

tttadam

@robert-hh It's working for me too! :)
Awsome! :)
Now I just need to find a correct library for it.
I will start with this

Can you give me some explonation what happend?
What your overall expression of the board?

robert-hh

@tttadam When you do the i2c.scan(), which address is reported for the HMC5883? It puzzles me that my device reports 13 instead of 30, which is the address in all examples.
Besides that, I have the impression that in normal operation the HMC5883 is controlled by the MCU6050 for its own purpose. So you might get the information you need through the MCU6050.
If you ask me about what I did to find that way: I looked at the circuit, read the data sheets and looked at some code examples for the MCU6050.

tttadam

@robert-hh

This is what I see. So same for me.

>>> from machine import I2C
>>> i2c = I2C(1)
>>> i2c.scan()
[64, 104, 119]
>>> i2c.writeto_mem(104, 107, 128) # reset MPU6050
>>> i2c.writeto_mem(104, 107, 0) # disable sleep
>>> i2c.writeto_mem(104, 55, 2)   # enable I2C passthru
>>> i2c.scan()
[13, 64, 104, 119]
>>>

robert-hh

@tttadam So let's see if that is a bug or an intended address trabslation (but why?)

robert-hh

@tttadam A short google search gave an answer:
The chip in the board is not a HMC5883, but a QMC5883. See https://github.com/nodemcu/nodemcu-firmware/issues/2187
The QMC5883 server for the same purpose, but has different programming interface.

robert-hh

@tttadam I found a RasPi driver for the QMC5883 and adapted it to pycom MP:

#
# ported from http://nobytes.blogspot.com/2018/03/qmc5883-magnetic-field-sensor.html
#

import time
import struct
from machine import idle

class QMC5883:
    #Default I2C address
    ADDR = 0x0D

    #QMC5883 = Register = numbers
    X_LSB = 0
    X_MSB = 1
    Y_LSB = 2
    Y_MSB = 3
    Z_LSB = 4
    Z_MSB = 5
    STATUS = 6
    T_LSB = 7
    T_MSB = 8
    CONFIG = 9
    CONFIG2 = 10
    RESET = 11
    CHIP_ID = 13

    #Bit values for the STATUS register
    STATUS_DRDY = 1
    STATUS_OVL = 2
    STATUS_DOR = 4

    #Oversampling values for the CONFIG register
    MC5883L_CONFIG_OS512 = 0b00000000
    CONFIG_OS256 = 0b01000000
    CONFIG_OS128 = 0b10000000
    CONFIG_OS64  = 0b11000000

    #Range values for the CONFIG register
    CONFIG_2GAUSS = 0b00000000
    CONFIG_8GAUSS = 0b00010000

    #Rate values for the CONFIG register
    CONFIG_10HZ = 0b00000000
    CONFIG_50HZ = 0b00000100
    CONFIG_100HZ = 0b00001000
    CONFIG_200HZ = 0b00001100

    #Mode values for the CONFIG register
    CONFIG_STANDBY = 0b00000000
    CONFIG_CONT = 0b00000001


    def __init__(self, i2c):
        self.i2c = i2c
        self.oversampling = QMC5883.CONFIG_OS64
        self.range = QMC5883.CONFIG_2GAUSS
        self.rate = QMC5883.CONFIG_100HZ
        self.mode = QMC5883.CONFIG_CONT
        self.reset()

    def reconfig(self):
        print("{0:b}".format(self.oversampling | self.range | self.rate | self.mode))
        self.i2c.writeto_mem(QMC5883.ADDR, QMC5883.CONFIG, self.oversampling | self.range | self.rate | self.mode);

    def reset(self):
        self.i2c.writeto_mem(QMC5883.ADDR, QMC5883.RESET, 0x01)
        time.sleep(0.1)
        self.reconfig()
        time.sleep(0.01)

    def setOversampling(self, x):
        self.oversampling = x
        self.reconfig()

    def setRange(x):
        self.range = x
        self.reconfig()

    def setSamplingRate(self, x):
        self.rate = x
        self.reconfig()

    def ready(self):
        status = self.i2c.readfrom_mem(QMC5883.ADDR, QMC5883.STATUS, 1)[0]

        # prevent hanging up here.
        # Happens when reading less bytes then then all 3 axis and will end up in a loop.
        # So, return any data but avoid the loop.
        if (status == QMC5883.STATUS_DOR):
            print("fail")
            return QMC5883.STATUS_DRDY

        return status & QMC5883.STATUS_DRDY

    def readRaw(self):
        while not self.ready():
            idle()
            pass

        # Python performs a wrong casting at read_i2c_block_data.
        # The filled buffer has to be onverted afterwards by mpdule Struct
        register = self.i2c.readfrom_mem(QMC5883.ADDR, QMC5883.X_LSB, 9);

        # Convert the axis values to signed Short before returning
        x, y, z, status, t = struct.unpack('<hhhBh', register)

        return (x, y, z, t)

and a short test code:

import time

from qmc5883 import QMC5883

from machine import I2C
i2c=I2C()
i2c.writeto_mem(104, 107, 128) # reset MPU6050
i2c.writeto_mem(104, 107, 0) # disable sleep
i2c.writeto_mem(104, 55, 2)   # enable I2C passthru

print(i2c.scan())
time.sleep(1)

qmc = QMC5883(i2c)

while True:
    print(qmc.readRaw())
    time.sleep(1)

tttadam

@robert-hh thank you, it's looks useful, but after some time it's produce an OSError. Line 17 in my code is this : print(qmc.readRaw())
and the line 99 is this. register = self.i2c.readfrom_mem(QMC5883.ADDR, QMC5883.X_LSB, 9)
I think there were an ; which is not needed.
But something else is still not alright. I didn't find it yet.

(6967, -4882, -7620, -1943)
(4742, -7337, -7950, -1923)
Traceback (most recent call last):
  File "main2.py", line 17, in <module>
  File "/flash/lib/qmc5883.py", line 99, in readRaw
OSError: I2C bus error

tttadam

@robert-hh

This is the error what i get now. Maybe tomorrow with fresh eye I will see something :)

[13, 64, 104, 119]
11001001
Traceback (most recent call last):
  File "main2.py", line 17, in <module>
  File "/flash/lib/qmc5883.py", line 95, in readRaw
  File "/flash/lib/qmc5883.py", line 81, in ready
OSError: I2C bus error

kuqizz

This post is deleted!

robert-hh

@tttadam No, the ; is not needed (probably a left-over from an initial C-Code file), but it also does not hurt. There is another one in reconfig(). The OSError is probably caused by a I2C communication problem to the device. I did not let it run for longer than about 20 minutes or 1000 measurement. If you enclose that into a try/except clause, does the device recover?
The temperature reading has an offset. That is also told in the data sheet. If I add 4900 to the value, it is almost OK. The value seems to be scaled as °C*100.

tttadam

@robert-hh Okay, So I try-catch block for OSError for

status = self.i2c.readfrom_mem(QMC5883.ADDR, QMC5883.STATUS, 1)[0]

and

return QMC5883.STATUS_DRDY

lines. Looks like it's not recovering.

Something went wrongI2C bus error
Traceback (most recent call last):
  File "main2.py", line 17, in <module>
  File "/flash/lib/qmc5883.py", line 110, in readRaw
  File "/flash/lib/qmc5883.py", line 97, in ready
NameError: local variable referenced before assignment.

Did you used this library for porting?
Because it's mention something about readraw blocking.
Where did you find the desription about these four values? What are does mean?

The whole code looks like this, but it is the same, only thing I added, that I moved a i2c magic into the init method from main:

#
# ported from http://nobytes.blogspot.com/2018/03/qmc5883-magnetic-field-sensor.html
#

import time
import struct
from machine import idle

class QMC5883:
    #Default I2C address
    ADDR = 0x0D

    #QMC5883 = Register = numbers
    X_LSB = 0
    X_MSB = 1
    Y_LSB = 2
    Y_MSB = 3
    Z_LSB = 4
    Z_MSB = 5
    STATUS = 6
    T_LSB = 7
    T_MSB = 8
    CONFIG = 9
    CONFIG2 = 10
    RESET = 11
    CHIP_ID = 13

    #Bit values for the STATUS register
    STATUS_DRDY = 1
    STATUS_OVL = 2
    STATUS_DOR = 4

    #Oversampling values for the CONFIG register
    MC5883L_CONFIG_OS512 = 0b00000000
    CONFIG_OS256 = 0b01000000
    CONFIG_OS128 = 0b10000000
    CONFIG_OS64  = 0b11000000

    #Range values for the CONFIG register
    CONFIG_2GAUSS = 0b00000000
    CONFIG_8GAUSS = 0b00010000

    #Rate values for the CONFIG register
    CONFIG_10HZ = 0b00000000
    CONFIG_50HZ = 0b00000100
    CONFIG_100HZ = 0b00001000
    CONFIG_200HZ = 0b00001100

    #Mode values for the CONFIG register
    CONFIG_STANDBY = 0b00000000
    CONFIG_CONT = 0b00000001


    def __init__(self, i2c):
        self.i2c = i2c
        self.i2c.writeto_mem(104, 107, 128) # reset MPU6050
        self.i2c.writeto_mem(104, 107, 0) # disable sleep
        self.i2c.writeto_mem(104, 55, 2)   # enable I2C passthru
        self.oversampling = QMC5883.CONFIG_OS64
        self.range = QMC5883.CONFIG_2GAUSS
        self.rate = QMC5883.CONFIG_100HZ
        self.mode = QMC5883.CONFIG_CONT
        self.reset()

    def reconfig(self):
        print("{0:b}".format(self.oversampling | self.range | self.rate | self.mode))
        self.i2c.writeto_mem(QMC5883.ADDR, QMC5883.CONFIG, self.oversampling | self.range | self.rate | self.mode)

    def reset(self):
        self.i2c.writeto_mem(QMC5883.ADDR, QMC5883.RESET, 0x01)
        time.sleep(0.1)
        self.reconfig()
        time.sleep(0.01)

    def setOversampling(self, x):
        self.oversampling = x
        self.reconfig()

    def setRange(x):
        self.range = x
        self.reconfig()

    def setSamplingRate(self, x):
        self.rate = x
        self.reconfig()

    def ready(self):
        try:
            status = self.i2c.readfrom_mem(QMC5883.ADDR, QMC5883.STATUS, 1)[0]
        except OSError as e:
            print("Something went wrong"+str(e))
            #TODO: Tune it a little   

        # prevent hanging up here.
        # Happens when reading less bytes then then all 3 axis and will end up in a loop.
        # So, return any data but avoid the loop.
        if (status == QMC5883.STATUS_DOR):
            print("fail")
            try:
                return QMC5883.STATUS_DRDY
            except OSError as e:
                print("Something went wrong"+str(e))
                #TODO: Tune it a little   

        return status & QMC5883.STATUS_DRDY

    def readRaw(self):
        while not self.ready():
            idle()
            pass

        # Python performs a wrong casting at read_i2c_block_data.
        # The filled buffer has to be onverted afterwards by mpdule Struct
        register = self.i2c.readfrom_mem(QMC5883.ADDR, QMC5883.X_LSB, 9)

        # Convert the axis values to signed Short before returning
        x, y, z, status, t = struct.unpack('<hhhBh', register)

        return (x, y, z, t)

robert-hh

@tttadam
Thanks for the error log. But that only happens if you catch the error, print a message and then continue. The the symbol status is not defined, raising a new exception. So you would either add:
status = 0
or
return 0
after the print statements.

Did you used this library for porting?

No. I found Python code, which in turn might have been ported from that web side. See http://nobytes.blogspot.com/2018/03/qmc5883-magnetic-field-sensor.html with a link to another interesting port, of which I might carry over the calibration part.

I did intentionally not include that I2C magic, because that is only relevant for this GY87 board.

tttadam

@robert-hh I found the issue.
GY-87 is on breadboard and and from the breadboard connected to the pytrack via dupont cables.
It was a contact issue. Sorry for the false alarm.
Can you help me out what are those 4 numbers mean, or where do I find a some info about that? My guess is that the last one is related to the temperature somehow but I am not sure.

(-925, -2192, -8705, -2071)

robert-hh

@tttadam The first three are the x,y,z values for the magnetic field, the last one is the temperature. All numbers a raw values. I have prepared a version which returns proper scaled gauss values for the magnetic field and °C for temperature. I did not test it yet, so I will send it later.
The temperature value has an offset. According to the data sheet, it is scaled properly, so differences are fine. But the offset for the absolute value has to be determined. For my device it is about 5000. So if you local temperature is about 30°C, then it's about the same. The scaling is 100LSB/°C. So if you divide the value by 100, you get °C (with offset).

Explore Pybytes | Official Documentation | Report a Firmware Bug/Issue | GitHub

GY-87 micropython library

Pycom on Twitter