Session 2 of the AFNOG tutorial
on IoT sensors

Uli Raich

• an rgb LED
• a color sensor
• a distance sensor

Different rgb LEDs are available on the market:

WS2812	SMD rgb LED	LilyPad LED	Standard LED

each LED is individually addressed.

A precisely timed digital communication protocol is used to accessed the LED.

Too complex for our tutorial

• r,g,b
• Gnd

be provided externally (which values if we want max. 10mA in one LED?)

The board uses a common cathode circuit, which means that it has a common ground

and the rgb signals must be on Vcc to light the LED

This means that the common pin is on Vcc and the rgb pins

must be brought to ground to light the LED

means of the cobbler (blue PCB) and a flat cable.

Like this we get access to interface pins on the Broadcom BCM2837 processor chip.

can be programmed to be input or output.

In addition internal pullup resistors can be enabled if desired.

In order to light our LED we therefore need 3 GPIO pins, one for each color.

These pins must be programmed to be output and the corresponding signal level

must be set (active high in case of common cathode, active low in case of common anode).

The intensity can be changed using pulse width modulation (pwm)

where a high frequency signal is sent to the LED.

The intensity is changed by changing the duty cycle of the signal

register access on the processor chip.

This however is clumsy and needs detailed knowledge of the chip.

A library is available giving the programmer easy access to GPIOs,

including intensity variation through pwm.

The library we will be using is pigpio which has a Python interface.

The library uses a daemon (pigpiod), which is usually started at system startup.

It is also possible to start it manually as super user.

The Python module connects to the daemon and sends commands to it.

The daemon is responsible for the hardware access.

Please have a look at the pigpio docs.

• Vcc to the red, green or blue pin and GN to ‘-’ in case of the common cathode PCB
• Vcc to ‘+’ and GND to red, green or blue in case of the common anode PCB

Then we connect the LED to the Raspberry Pi as follows:

• red to pin 18
• green to pin 17
• blue to pin 22
• ‘-’ to GND in case of the common cathode version or
  ‘+’ to Vcc in case of the common anode version

• First we import the pigpio library:
  
  import pigpio # imports the library
• Create a pi object from the pigpio class
  
  This will connect our program to the pigpiod daemon
  
  pi = pigpio.pi()# connects to the daemon
• We program the gpio pins to be output:
  
  pi.set_mode(_RED,pigpio.OUTPUT) # _RED must be set to 18
• And finally we write 0 or 1 to the pin:
  
  pi.write(_RED,pigpio.HIGH) # or pigpio.LOW
• In order to change the intensity of the color component we use pwm:
  
  pi.set_PWM_dutycycle(_RED, intensity) # intensity: 0..255

All pins off is black, all pins on is white.

pi = pigpio.pi(host,port,show_errors)

After this the program is identical to what we have seen. So…

What is the difference?

In the first case both, pigpiod and the Python program run on the Raspberry Pi

and only the keyboard, mouse and screen are used to interact via the ssh interface.

In the second case the Python program runs of the PC

and only the I/O to the sensors or LEDs happen through the daemon.

Of course you can do the same thing for green and blue shades

or for any combination of the colors

can only be driven by binary signals?

The answer is pwm: pulse width modulation.

The signal to the LEDs is sent a a high frequency where the switching is invisible to the human eye.

The duty-cycle determines the current flowing through the LED and thus the color.

It can easily be seen that with respect to a permanent “1”

a pwm signal with 50% duty-cycle produces on average only ½ of the current

is much more complex than driving the LEDs

However, we are lucky:

There are Python classes supporting these devices written for us by experienced Python programmers.

These classes come with example code showing how to use the class methods.

Of course we can just use the code. It is however much better to try and understand the code before using it.

Reading code written by expert programmers can help us to substantially improve our own programming skills.

On a trigger the sensor sends an 8 cycle burst of ultra-sound (40 kHz)

It then reads the echo and generates a pulse whose length is proportional to the signal travel time.

wait for 10 μs and setting it low again.

Then we read the time at which the Echo signal becomes high and when it becomes low again.

The time difference corresponds to the sound travel time.

Since we know that the

speed of sound in air is ~ 343 m/s

the distance the sound wave has traveled is distance = 34300/2 * signalLength [cm]

(the factor 2 comes from the fact that the wave travels from the loudspeaker

to the paper and back to the microphone, which is twice the distance)

the Raspberry Pi works with 3.3V logic levels.

This is ok for Raspberry Pi output signals. The correct logic level is seen by the sensors.

However, the input signals to the Raspberry Pi must be converter to 3.3V,

which can be done with a resistor voltage divider or with a level shifter:

On the B side we provide 5V signals which are converted to 3.3V on the A side

Both sides need Vcc (side B: 5V, side A: 3.3V) and GND

What can happen when an interrupt is treated while our program is waiting for a transition?

We may wait forever! The program gets blocked!

There are 2 ways to treat this:

• use timeouts: make sure that the time since we started waiting
  
  for the transition does not exceed a predefined max. value
• use callbacks