Reading analog-in on BeagleBone, avoiding "segmentation fault" error

Question

I am new to the Beaglebone, and was moving along at a pretty good click until this strange problem popped-up. I spent all weekend trying to figure it out before posting here.

I am trying to read analog input using multiple sensors from a beaglebone with Angstrom 3.2.5+ and I have a problem that I think might be a hardware issue.

Using this Python script and Alexander Hiam's PyBBIO I have been able to test a photoresistor with success (the value responds to more and less incidental light):

# Import PyBBIO library:
from bbio import *

# Create a setup function:
def setup(): pass # Nothing to do here

# Create a main function:
def loop():

  print "-----"

  print "AIN0 [P9.39] - " + str(analogRead(A0))
  print "AIN1 [P9.40] - " + str(analogRead(A1))
  print "AIN2 [P9.37] - " + str(analogRead(A2))
  print "AIN3 [P9.38] - " + str(analogRead(A3))
  print "AIN4 [P9.35] - " + str(analogRead(A4))
  print "AIN5 [P9.36] - " + str(analogRead(A5))
  print "AIN6 [P9.33] - " + str(analogRead(A6))
  print "AIN7 [P9.39] - " + str(analogRead(A7))

  delay(500)

# Start the loop:
run(setup, loop)

But, any other analog-in (AINx(2-7)) I read with the script above always shows the same value of the photo resistor I have plugged-into AIN1

And, possibly related, I am unable to read any of the AINx(1-7) using cat with an error

cat /sys/bus/platform/devices/tsc/ain1
Segmentation fault

Any ideas? I am really stuck. Thanks in advance.

PS: More info...

My circuit consists of a 10k resistor and a photoresistor. My pin connections are:

• Header P9 Pin 32 (VDD_ADC(1.8V)) to one end of photocell
• Header P9 Pin 34 (GNDA_ADC) to the other end of the resistor
• Header P9 Pin 39 (AIN0) to the other side of the photocell along with one end of pull-down resistor

Here is an image which matches my own circuit:

The results from all of the AINx(1-7)'s and then some...

# cat /sys/bus/platform/devices/tsc/ain0 
cat: /sys/bus/platform/devices/tsc/ain0: No such file or directory 
# cat /sys/bus/platform/devices/tsc/ain1 
Segmentation fault 
# cat /sys/bus/platform/devices/tsc/ain2 
Segmentation fault 
# cat /sys/bus/platform/devices/tsc/ain3 
Segmentation fault 
# cat /sys/bus/platform/devices/tsc/ain4 
Segmentation fault 
# cat /sys/bus/platform/devices/tsc/ain5 
Segmentation fault 
# cat /sys/bus/platform/devices/tsc/ain6 
Segmentation fault 
# cat /sys/bus/platform/devices/tsc/ain7 
Segmentation fault 
# cat /sys/bus/platform/devices/tsc/ain8 
Segmentation fault 
# cat /sys/bus/platform/devices/tsc/ain9 
cat: /sys/bus/platform/devices/tsc/ain9: No such file or directory

UPDATE 1

I edited the analog_test_simple.py file to show all ADC ports on the BeagleBone. This is the result from the shell with no sensors plugged-in.

-----
AIN0 [P9.39] - 3948
AIN1 [P9.40] - 4016
AIN2 [P9.37] - 4002
AIN3 [P9.38] - 3934
AIN4 [P9.35] - 4001
AIN5 [P9.36] - 3935
AIN6 [P9.33] - 3968
AIN7 [P9.39] - 4005
-----
AIN0 [P9.39] - 3946
AIN1 [P9.40] - 4005
AIN2 [P9.37] - 3945
AIN3 [P9.38] - 3957
AIN4 [P9.35] - 4018
AIN5 [P9.36] - 3945
AIN6 [P9.33] - 3967
AIN7 [P9.39] - 3955
...

And if I plug the sensor back in I get a change in value, but again, they are all the same:

-----
AIN0 [P9.39] - 2836
AIN1 [P9.40] - 2836
AIN2 [P9.37] - 2837
AIN3 [P9.38] - 2831
AIN4 [P9.35] - 2840
AIN5 [P9.36] - 2837
AIN6 [P9.33] - 2837
AIN7 [P9.39] - 2837
-----
AIN0 [P9.39] - 2834
AIN1 [P9.40] - 2834
AIN2 [P9.37] - 2829
AIN3 [P9.38] - 2825
AIN4 [P9.35] - 2826
AIN5 [P9.36] - 2817
AIN6 [P9.33] - 2815
AIN7 [P9.39] - 2815
...

Answer 1

OK, the answer is that there are two separate problems, both of which are caused by using the PyBBIO library. Read on...

PROBLEM 1: Inconsistent/incorrect sensor values

A suggestion in the comments on my question prompted me to add a delay to my original PyBBIO-based script. Here it is without the sensor:

# python PyBBIO/examples/analog_test_simple.py
-----
AIN0 [P9.39] - 3955
AIN1 [P9.40] - 4015
AIN2 [P9.37] - 4012
AIN3 [P9.38] - 3950
AIN4 [P9.33] - 3998
AIN5 [P9.36] - 4007
AIN6 [P9.35] - 3984
-----
AIN0 [P9.39] - 4007
AIN1 [P9.40] - 3953
AIN2 [P9.37] - 4005
AIN3 [P9.38] - 3941
AIN4 [P9.33] - 3956
AIN5 [P9.36] - 3934
AIN6 [P9.35] - 3946
...

And here are the results after I plug-in the sensor:

# python PyBBIO/examples/analog_test_simple.py
-----
AIN0 [P9.39] - 2888
AIN1 [P9.40] - 2894
AIN2 [P9.37] - 2899
AIN3 [P9.38] - 2907
AIN4 [P9.33] - 2914
AIN5 [P9.36] - 2912
AIN6 [P9.35] - 2914
-----
AIN0 [P9.39] - 2912
AIN1 [P9.40] - 2905
AIN2 [P9.37] - 2901
AIN3 [P9.38] - 2891
AIN4 [P9.33] - 2896
AIN5 [P9.36] - 2890
AIN6 [P9.35] - 2896
...

You can see that plugging-in the sensor still affects all the values, regardless of the delay.

After thinking about this a bit I decided to try to repeat the results with a script without using PyBBIO. I wrote a new Python file to test and print the reading from each analog-in pin on the BeagleBone using the Python os module. The file uses os.system('cat ...') to print the values, and includes a short delay between each. Here is the code:

import os,time
i = 1

while(True):
    os.system("cat /sys/bus/platform/devices/tsc/ain" + str(i))
    print " - AIN" + str(i-1)
    if i==7: 
        i=1
        time.sleep(.5)
        print "------"
    else:
        i+=1    
    time.sleep(.1)

Here are the results without any components plugged-into the BeagleBone:

# python analog-test-all-pins.py
------
3943 - AIN0
3819 - AIN1
3955 - AIN2
2018 - AIN3
2093 - AIN4
3583 - AIN5
3658 - AIN6
------
3947 - AIN0
3746 - AIN1
3959 - AIN2
2034 - AIN3
2123 - AIN4
3547 - AIN5
3644 - AIN6
...

And here are the results with the photoresistor circuit above plugged-into AIN0:

# python analog-test-all-pins.py
------
2915 - AIN0
3347 - AIN1
3556 - AIN2
1478 - AIN3
1602 - AIN4
2393 - AIN5
2402 - AIN6
------
2913 - AIN0
3337 - AIN1
3560 - AIN2
1487 - AIN3
1606 - AIN4
2350 - AIN5
2489 - AIN6

And here are the results with the photo resistor circuit data lead plugged-into AIN2:

# python analog-test-all-pins.py
------
3939 - AIN0
3792 - AIN1
2881 - AIN2
1986 - AIN3
2089 - AIN4
3462 - AIN5
3543 - AIN6
------
3923 - AIN0
3791 - AIN1
2866 - AIN2
1960 - AIN3
2055 - AIN4
3528 - AIN5
3615 - AIN6
...

AIN3:

# python analog-test-all-pins.py
------
3951 - AIN0
3764 - AIN1
3933 - AIN2
2899 - AIN3
2134 - AIN4
3606 - AIN5
3737 - AIN6
------
3960 - AIN0
3823 - AIN1
3957 - AIN2
2910 - AIN3
2118 - AIN4
3635 - AIN5
3645 - AIN6
...

And here are the results if I plug the photoresistor data lead back into AIN0 and plug a TMP36 temperature sensor into AIN3.

# python analog-test-all-pins.py
------
2866 - AIN0
3340 - AIN1
3540 - AIN2
885 - AIN3
1591 - AIN4
2348 - AIN5
2389 - AIN6
------
2872 - AIN0
3339 - AIN1
3551 - AIN2
884 - AIN3
1560 - AIN4
2383 - AIN5
2434 - AIN6
...

And just to confirm that the sensors are being read without affecting the other pins, here is the TMP36 on AIN5

# python analog-test-all-pins.py
------
2897 - AIN0
3338 - AIN1
3557 - AIN2
1464 - AIN3
1578 - AIN4
888 - AIN5
2459 - AIN6
------
2901 - AIN0
3344 - AIN1
3561 - AIN2
1460 - AIN3
1579 - AIN4
889 - AIN5
2395 - AIN6
...

This leads me to conclude that there is a issue with the PyBBIO library that is causing this. It is also causing the second problem, though I have no idea how...

PROBLEM 2: Segmentation fault error

The problem I had with the Segmentation fault error is due to something in the PyBBIO library. For example, if I restart the BeagleBoard and run:

# cat /sys/bus/platform/devices/tsc/ain1
3953
# cat /sys/bus/platform/devices/tsc/ain2
3818

I get values which match the above output from the Python script using the os module.

But if I run any PyBBIO script which accesses AIN pins (examples of which are above) and then run the cat again I get the error:

# cat /sys/bus/platform/devices/tsc/ain1
Segmentation fault
# cat /sys/bus/platform/devices/tsc/ain2
Segmentation fault

If I restart the board (unplug and plug-in the power) I can use cat again:

# cat /sys/bus/platform/devices/tsc/ain1
2890
# cat /sys/bus/platform/devices/tsc/ain2
3366

I also tried using Matt Richardson's mrBBIO library instead, with this file:

from mrbbio import *

def setup():
    #pinMode("P9.36", INPUT)
    pass

def loop():

    print "-----"

    for key, value in analogPinDef.items():
        # There is currently an error in mrBBIO, as the analogPinDef dict points to:
        # ain[0-6] in /sys/devices/platform/tsc/ while it should point at: ain[1-7]
        if key=="P9.39": 
            print analogRead(key) + " - " + value
            delay(100)

    delay(1000)

run (setup,loop)

to get this:

# python mrbbio/analog-test-all-pins.py
-----
2005 - ain4
3636 - ain6
3812 - ain2
2114 - ain5
3872 - ain3
3950 - ain1
-----
2002 - ain4
3530 - ain6
3787 - ain2
2059 - ain5
3895 - ain3
3952 - ain1
...
Cleaning up. Unexporting the following pins:

And I found it does not mess with my ability to get the results with cat -- possibly because it exports the pins at the end.

# cat /sys/bus/platform/devices/tsc/ain1
3960
# cat /sys/bus/platform/devices/tsc/ain2
3830