CircuitPython Basics I2C and SPI
2017-11-14 | By Adafruit Industries
License: See Original Project Arduino Adafruit Feather
Courtesy of Adafruit
Guide by Tony DiCola
Overview
Talking to hardware from your development board is when the real fun starts with a project. You can connect sensors, actuators, and more to make your project come alive with motion, sensation, sound, etc. However to communicate with other hardware typically requires knowledge of a serial protocol like I2C or SPI. These protocols are the common language that chips and add-on boards talk so they can be connected to a development board. The board knows how to ‘speak’ these protocols and control the connected hardware. This guide explores two very common serial protocols, I2C and SPI.
I2C Devices
I2C Protocol
The I2C, or inter-integrated circuit, protocol is one example of a serial protocol for devices to communicate with one another. I2C is a serial protocol because it has a clock line and single data line which is used for both sending and receiving data. Compared to other serial protocols I2C has some interesting properties:
- The I2C protocol only uses 2 wires to send and receive data. One line is a clock, called SCL, which pulses high and low to drive the sending and receiving of bits. The other line is the data line, called SDA, which contains the value of a sent or received bit during clock line transitions.
- Multiple I2C devices can be connected to the same clock and data lines. This means you can have many different sensors and devices all connected to the same couple pins from your development board. The I2C protocol uses a 7-bit address assigned to each device as a way for the development board to talk to a specific device. As a result of using 7-bit addresses the I2C protocol is limited to 127 unique devices connected to one bus (or pair of data and clock lines).
- The speed of the I2C bus is fixed, typically to 100kHz, 400kHz, or 1MHz. This means I2C is a good protocol for talking to devices that don’t send a lot of data or need very fast responses. A TFT display which receives hundreds of kilobytes and even megabytes of image data wouldn’t make sense as an I2C device because sending so much data over a 100kHz bus would be quite slow. However a small sensor like a temperature or light sensor that sends a small 16 or 32-bit value typically doesn’t need a fast bus.
- The I2C clock and data lines need pull-up resistors to prevent from floating to random values. Since many different devices can share these lines the I2C protocol requires that each device ‘give up’ or stop driving the lines when not in use. If no device is driving the lines then the pull-up resistors ensure they go up to a high logic level instead of floating at random values. Most I2C device boards (in particular the boards Adafruit creates) have these pull-up resistors built-in, but if you’re talking to a chip or building a board yourself you might need to add ~2.2-10 kilo-ohm resistors connected from both data and clock lines up to high logic level.
- The I2C protocol includes a simple guarantee that data has been transferred between devices. When one I2C device receives data from another device it uses a special acknowledgement bit to tell the sending device that data has been received. There’s no error correction, parity checks, etc.–just a simple yes/no that data has been successfully sent and received.
- Typically one device on an I2C bus is the ‘master’ which controls the clock line and sends requests to other connected devices. In most cases your development board is the master device that drives the I2C bus clock. Sensors and other I2C devices connected to the bus listen for requests from the master and respond appropriately. This guide covers this most common scenario where your development board is the I2C master and is talking to connected devices.
- Many I2C devices expose data through a simple register table. This means to query data from the device you need to know both the address of the device and the address of the register you wish to query. Check your device’s datasheet for the exact device and register address values. Typically registers are 8-bit values (0 to 255) but devices are free to use larger or smaller sizes–always check your device’s datasheet to be sure how it exposes data!
These properties make I2C an attractive protocol for sensors and other simple devices that don’t need to send or receive data quickly. Many different sensors can all be connected to the same I2C clock and data lines. By giving each sensor a unique 7-bit address your development board and code can query each one for their current value.
MCP9808 I2C Temperature Sensor
To demonstrate interacting with an I2C device this guide will show you how to query the temperature from a MCP9808 high precision temperature sensor. This sensor is a good example of an I2C device because it has a very simple register structure. To read the temperature you simply read one register from the device–there’s no complex logic or need to read other registers like on some other sensors. You’ll need these parts to follow this section:
- MCP9808 high precision temperature sensor
- A breadboard and wires to connect the components and board together.
Connect the components together as follows:
- Board 5V or 3.3V output to MCP9808 VDD or VIN
- Board ground/GND to MCP9808 GND
- Board SCL (I2C clock line) to MCP9808 SCL
- Board SDA (I2C data line) to MCP9808 SDA
Remember the I2C protocol requires pull-up resistors to be on the clock and data lines. If you’re using an Adafruit breakout board like the MCP9808 sensor linked above then these pull-ups are built-in and nothing else is necessary. However if you’re wiring a chip directly to your board or using a different breakout you might need to add pull-up resistors. Typically these are 2.2k - 10k ohm resistors that connect both clock and data up to high logic / 3.3V.
Once the device is wired up you’re ready to start interacting with it from CircuitPython. The easiest way to demonstrate this control is from the serial REPL and an interactive Python session. Connect to your board’s serial REPL, then import the board and busio module:
>>> import board
>>> import busio
The busio module contains an interface for using hardware-driven I2C communication from your board. Note that on some boards, like the ESP8266, they might not support hardware-driven I2C and must fall back to a slower software driven approach. You’ll learn more about software driven, or bit-banged, access later in this guide. If you’re not sure check your board’s documentation for its support of I2C–most boards like the Metro M0, Trinket M0, Gemma M0, Circuit Playground Express support a hardware-driven I2C interface.
Within the busio module you’ll use the busio.I2C class to create an interface to access the I2C bus:
>>> i2c = busio.I2C(board.SCL, board.SDA)
When creating the I2C class you must specify the clock line and data line pins. Typically these are the board.SCL and board. SDA objects but check your board’s documentation in case there are other hardware I2C buses with different clock and data line names.
Once you have access to the I2C bus it’s easy to scan the bus to find the address of all devices connected to it. Call the busio.I2C.scan() function.
However before you make calls against the I2C bus you first need to take control, or ‘lock’, it to ensure your code has exclusive access to I2C. There are a few ways to lock the bus like waiting on the busio.I2C.try_lock() function and then calling the busio.I2C.unlock() function when finished (typically in a Python try-finally block). Locking the bus tells CircuitPython that your code needs to use I2C and that any other code using I2C should wait for your code to finish. This helps different bits of code use the same hardware peripherals by ensuring they don’t try to use it at the same time or interrupt each other.
To lock the I2C bus you want to use a special loop syntax that waits for the busio.I2C.try_lockfunction to succeed:
>>> while not i2c.try_lock():
... pass
...
>>>
This loop will continually call the try_lock function until it returns true that the I2C bus was locked. Remember other code might be using the bus so the loop will keep trying to lock the bus and ensure it’s available to use. Once the bus is locked you can start to call functions to access it, like scanning for any available I2C devices. Try calling the busio.I2C.scan() function to list the addresses of all I2C devices found on the bus:
>>> i2c.scan()
[24]
Notice when you use the with statement all the code inside of it is indented and doesn’t run until you end the with statement (but removing the indentation or pressing enter three times). The great thing about the with statement and context manager is that it’s automatically locking and unlocking the I2C interface so calls like scan can be made.
Notice the busio.I2C.scan() function returns a list of 7-bit I2C device addresses. Be careful as Python treats these numbers as normal base 10 values when printing them, whereas most I2C addresses from datasheets are in hex. You can use a special list comprehension syntax to convert the list of numbers into hex strings:
>>> [hex(x) for x in i2c.scan()]
['0x18']
Now you’ll see a list of 2 digit hex strings which are easier to double check with your device’s datasheet. In this case a device with address 0x18 is visible on the I2C bus and if you check the MCP9808 datasheet you’ll see by default its I2C address is 0x18. Perfect! This means the sensor is properly connected, powered, and responding to requests.
If for some reason you don’t see anything returned by scan, or completely different addresses then double check your wiring, power, and if pull-up resistors are necessary to add. If any one of those things isn’t setup correctly the device will not be visible to the I2C bus and scan.
Next you can read bytes from registers using a combination of writing and reading functions. With the I2C protocol all requests are actually transactions where the master devices writes to and then reads from a connected device. First the master writes the address of the register it wants to read, and then it reads a number of bytes from the device.
For example with the MCP9808 its temperature value is stored in a 16-bit register at address 0x05. You can read the value of this register by running:
>>> i2c.writeto(0x18, bytes([0x05]), stop=False)
>>> result = bytearray(2)
>>> i2c.readfrom_into(0x18, result)
>>> result
bytearray(b'\xc1s')
Let’s break down step by step what’s happening here:
- First the busio.I2C.writeto() function is called to start an I2C transaction by writing bytes of data from the board to the MCP9808. The first parameter is the address of the MCP9808, 0x18, and the second parameter is a list of bytes to be written. In this case only one byte, the value 0x05, is written. If you check the MCP9808 datasheet this 0x05 value is the temperature reading register. Finally the stop=False keyword argument tells the write function that there are more calls to come in this I2C transaction. Each I2C device can be different about how they expect transactions to be structured–some devices want an explicit stop after any master writes, whereas others expect the master to not send a stop and continue waiting for data from the device. You’ll need to check your device’s datasheet to be sure how it expects I2C calls to work, but for the vast majority of I2C devices this write with stop=False is what they expect.
- After writing data from the board to the MCP9808 we need to receive two bytes of temperature sensor register data. To do this we’ll call the busio.I2C.readfrom_into()function. But before you can call the function you need a place to store the returned bytes and to do this a bytearray of size 2 is created. This result bytearray will be passed to the read function and then filled with the results read from the MCP9808.
- The busio.I2C.readfrom_into() function is finally called to read two bytes of data from the MCP9808. Remember the I2C transaction is still ‘open’ from the previous write with stop=False so the MCP9808 knows to send back the previously requested temperature sensor register values. Again the first parameter to the read function is the address of the device (0x18) and the second parameter is a bytearray that will be filled with the bytes that are read. How does the function know how many bytes to read? The size of the passed in bytearray by default will determine how many bytes to read, so if you need to read more or less bytes the easiest way is to change the size of the bytearray passed in.
The last statement prints the two bytes that were read, 0xC173. This is the response from the MCP9808 after it was asked to send the temperature sensor register. If you check the datasheet you can see the format for this response actually encodes the sensed temperature. Luckily with Python it’s easy to make a function that decodes the temperature:
>>> def temp_c(data):
... value = data[0] << 8 | data[1]
... temp = (value & 0xFFF) / 16.0
... if value & 0x1000:
... temp -= 256.0
... return temp
...
>>> temp_c(result)
23.1875
Notice the temperature of the MCP9808 is printed in Celsius!
Once you’re finished using I2C devices be sure to call the busio.I2C.unlock() function to give back control of the I2C bus to other code. You can explicitly call it like:
>>> i2c.unlock()
Or you can put your code in a try-finally block in Python which ensures the unlock function is called no matter what, even if your code fails. For example a complete scan and read with try-finally might look like:
>>> while not i2c.try_lock():
... pass
...
>>> try:
... [hex(x) for x in i2c.scan()]
... i2c.writeto(0x18, bytes([0x05]), stop=False)
... result = bytearray(2)
... i2c.readfrom_into(0x18, result)
... finally:
... i2c.unlock()
...
['0x18']
>>> result
bytearray(b'\xc1s')
That’s all there is to interacting with I2C devices from CircuitPython. With the busio.I2C class you have direct access to crafting I2C transactions of almost unlimited complexity. Most devices will use the basic write register, read bytes flow you saw here, but be sure to check your device’s datasheet in case it has different I2C protocol requirements.
I2CDevice Library
You saw above how to interact with an I2C device using the API built-in to CircuitPython. Remember using the built-in API requires careful management of the lock and unlock functions to access the I2C bus. If you’re writing code to talk to an I2C device you might find using the CircuitPython bus device library a bit easier to manage as it controls locking and unlocking automatically (using Python’s context manager with statement).
To use the bus device library you’ll first need to install the library on your board.
First make sure you are running the latest version of Adafruit CircuitPython for your board.
Next you'll need to install the necessary libraries to use the hardware--carefully follow the steps to find and install these libraries from Adafruit's CircuitPython library bundle. Our introduction guide has a great page on how to install the library bundle for both express and non-express boards.
For Express boards, install the entire bundle. For non-express boards, with limited space, you'll need to grab just the adafruit_bus_device folder from inside lib to the lib folder of your board’s CIRCUITPY drive.
Once you have the bus device library installed you can use the I2CDevice class to simplify access to a device on the I2C device bus. First setup the I2C bus exactly as you did before:
>>> import board
>>> import busio
>>> i2c = busio.I2C(board.SCL, board.SDA)
Now import the bus device module and create an instance of the I2CDevice class. Notice the I2CDevice class needs to be told both the I2C bus object, and the address of the I2C device to talk to (0x18 for the MCP9808 sensor here):
>>> from adafruit_bus_device.i2c_device import I2CDevice
>>> device = I2CDevice(i2c, 0x18)
Now you can use similar functions to read and write data on the I2C bus to interact with the I2CDevice. The important difference here is that the read and write functions on the I2CDevice object will remember and automatically send the right device address. In addition you can use Python’s with statement as a context manager to automatically lock and unlock the I2C bus.
Here’s how to read the temperature register using the I2C Device:
>>> with device:
... device.write(bytes([0x05]), stop=False)
... result = bytearray(2)
... device.read_into(result)
...
>>> result
bytearray(b'\xc1s')
>>> temp_c(result)
23.1875
Notice you no longer need to specify the address of the device (0x18) when reading and writing. The with statement is also automatically locking and unlocking the I2C bus so you don’t need to manage the locking yourself either. The only limitation of the I2CDevice class is that it needs to talk to a single device and can’t scan the entire bus or interact with multiple devices (instead create multiple I2CDevice instances!). If you’re writing code to interact with an I2C device it’s highly recommended to use the I2CDevice class!
Also for interacting with most sensors and devices you typically don’t need to write these low-level direct I2C bus manipulation requests (with either the built-in APIs or I2CDevice class). Instead look for a higher level library to interact with the device, like the CircuitPython MCP9808 library. Using a library saves you the work of writing this low-level I2C code and instead you can interact with simple temperature and other device properties. However it is handy to know how to write low-level I2C transactions in case you’re dealing with devices that don’t yet have a driver available!
Scan All Registers
An interesting property of most I2C devices is that they expose data with simple registers. Like you saw above with the MCP9808 sensor the register 0x05 held the temperature as 2 bytes of data. It’s sometimes handy to scan all of the registers of an I2C device and print out their values. Here’s an example of scanning a set of registers from the first I2C device found and printing their contents as hex (you might want to save this as a main.py file that runs at boot instead of typing it all into the REPL!):
import board
import busio
REGISTERS = (0, 256) # Range of registers to read, from the first up to (but
# not including!) the second value.
REGISTER_SIZE = 2 # Number of bytes to read from each register.
# Initialize and lock the I2C bus.
i2c = busio.I2C(board.SCL, board.SDA)
while not i2c.try_lock():
pass
# Find the first I2C device available.
devices = i2c.scan()
while len(devices) < 1:
devices = i2c.scan()
device = devices[0]
print('Found device with address: {}'.format(hex(device)))
# Scan all the registers and read their byte values.
result = bytearray(REGISTER_SIZE)
for register in range(*REGISTERS):
try:
i2c.writeto(device, bytes([register]))
i2c.readfrom_into(device, result)
except OSError:
continue # Ignore registers that don't exist!
print('Address {0}: {1}'.format(hex(register), ' '.join([hex(x) for x in result])))
# Unlock the I2C bus when finished. Ideally put this in a try-finally!
i2c.unlock()
For example with the MCP9808 you might see output like:
Found device with address: 0x18
Address 0x0: 0x0 0x1d
Address 0x1: 0x0 0x0
Address 0x2: 0x0 0x0
Address 0x3: 0x0 0x0
Address 0x4: 0x0 0x0
Address 0x5: 0xc1 0x83
Address 0x6: 0x0 0x54
Address 0x7: 0x4 0x0
Address 0x8: 0x3 0x1
Address 0x9: 0x60 0x1
Address 0xa: 0xa2 0x1
Address 0xb: 0x25 0x88
Address 0xc: 0x0 0x1
SPI Device
SPI Protocol
The SPI protocol, or serial peripheral interface, is another example of a serial protocol for two devices to send and receive data. The big difference between SPI and I2C is that SPI uses a few more wires, in particular an explicit data input and data output wire instead of sharing a single data wire like with I2C. There’s also a clock wire like in I2C, but with SPI it has the freedom to use almost any speed it desires from a few kilohertz up to hundreds of megahertz (if the hardware supports it!). This makes the SPI protocol great for devices like TFT displays that need to be sent very large amounts of data–with control over the clock speed it’s possible to very quickly send entire screen images to the display.
Compared to I2C the SPI protocol has some interesting properties:
- SPI uses 3 to 4 wires for sending and receiving data. One wire is a clock line that toggles up and down to drive bits being sent and received. Like with I2C only the master device can drive the clock. Another wire is MOSI, or ‘master output, slave input’ which is the data output from your board and sent to a connected device. Likewise a MISO wire, or ‘master input, slave output’, is for sending data from the device to the board receiving it. Finally most chips have a CS, or chip select, wire which is toggled to tell the chip that it should listen and respond to requests on the SPI bus.
- Like I2C multiple devices can share the same SPI bus, however a big difference is that each device typically requires its own unique CS line. Remember the CS/chip select line is what tells a chip that it should listen for SPI traffic. As a result for each SPI device you connect to your board it can share the clock, MOSI, MISO lines but must have its own CS line (typically connected to any free digital I/O pin).
- SPI devices have different requirements for speed (sometimes called baudrate), polarity, and phase. The SPI page on Wikipedia has a good description of what polarity and phase mean–they control how the data is sent and received over the MISO and MOSI lines. Different polarity values control if a digital high or low logic level means a bit is a one or zero. Similarly different phase values control when data is read and written by devices–either with the rising or falling edge of the clock line. The important thing to know about phase and polarity is that each device has its own requirement for setting them so be sure to check your device’s datasheet. Many devices are ‘mode 0’ which means a polarity and phase of 0 but watch out because some devices use different modes.
- Like with I2C the basic operations are reading and writing bits and bytes of data over the data lines. However unlike SPI there is no guarantee or check that a connected device received or sent data successfully. Sometimes chips have extra lines to watch for an acknowledgment, but sometimes they don’t and the SPI requests are ‘fire and forget’ with no guarantee they were received.
MAX31855 SPI Thermocouple Temperature Sensor
To demonstrate interacting with a SPI device this guide will show you how to query the temperature from a MAX31855 thermocouple temperature sensor. This sensor is a good example of an SPI device because it has a very simple interface, you just connect a MISO line and read bytes of temperature data. There are no registers or other complex structures to configure and process on the chip. You’ll need these parts to follow this section:
- MAX31855 thermocouple temperature sensor
- If you don’t have one a simple K-type thermocouple is also required to connect to the MAX31855
- A breadboard and wires to connect the components and board together
Connect the components together as follows:
Fritzing Source
- Board 5V or 3.3V output to MAX31855 VIN
- Board ground/GND to MAX31855 GND.- Board SCK (SPI clock line) to MAX31855 CLK/clock. Note this is on the small 2x3 header on a Metro M0 Express or other Arduino form-factor boards
- Board MISO to MAX31855 DO (data output, AKA MISO). Note this is also on the small 2x3 header on a Metro M0 Express or other Arduino form-factor board
- Board D2 (or any free digital I/O pin) to MAX31855 CS/chip select
The wiring above will configure hardware-based SPI communication. Like with I2C you can choose to use your microprocessor’s built-in SPI communication hardware, or you might use software ‘bit banging’ to talk SPI much more slowly over any digital I/O lines. You’ll see how to switch to software SPI further in this guide.
Once the board is wired up connect to the REPL. You’ll need to import the board, busio, and digitalio modules:
>>> import board
>>> import busio
>>> import digitalio
Remember the CS line is just a simple digital I/O line so we need to use the digitalio module to control it. Let’s setup a digital output to drive this line:
>>> cs = digitalio.DigitalInOut(board.D2)
>>> cs.direction = digitalio.Direction.OUTPUT
>>> cs.value = True
For most chips they expect the CS line to be held high when they aren’t in use and then pulled low when the processor is talking to them. However check your device’s datasheet as the polarity and phase (or mode) can change how the chip expects CS to work! In this case the MAX31855 expects CS to be high when not in use and pulled low when talking to it. We’ll start the CS line in a high or true value so that it isn’t yet listening.
Now we need to create an interface to the SPI hardware bus. Do so with this line to create an instance of the busio.SPI class:
>>> spi = busio.SPI(board.SCK, MISO=board.MISO)
To create the SPI class you must pass at least a clock pin and then optionally the MISO and MOSI pins. In this case the MAX31855 doesn’t use the MOSI pin so we only provide MISO.
Now we’re almost ready to read data from the sensor. However just like with I2C you must lock the SPI bus before you send and receive data. The busio.SPI.try_lock() and busio.SPI.unlock()functions can do this like with I2C. Let’s read 4 bytes of data from the chip:
>>> while not spi.try_lock():
... pass
...
>>> spi.configure(baudrate=5000000, phase=0, polarity=0)
>>> cs.value = False
>>> result = bytearray(4)
>>> spi.readinto(result)
>>> cs.value = True
>>> result
bytearray(b'\x01\xa8\x1a\xf0')
Before digging into the results let’s break down what just happened:
- The while loop at the start will attempt to lock the SPI bus so your code can access SPI devices. Just like with I2C you need to call try_lock (and later unlock) to ensure you are the only user of the SPI bus.
- The busio.SPI.configure() function is called to configure the speed, phase, and polarity of the SPI bus. It’s important to always call configure after locking the bus and before talking to your device as communication with other devices might have changed the speed, polarity, etc. You’ll need to look up the exact speed and other values from your device’s datasheet. For the MAX31855 we’ll use a speed of 5MHz and a polarity and phase of 0 (sometimes called mode 0).
- Next we toggle the CS line down to a low logic level. Remember with SPI each device needs a chip select line to tell it when it’s ready to send and receive data.
- A 4 byte buffer is created to hold the result of the SPI read. Just like with I2C reads you need to pass a buffer that will be filled with response data, and the size of the buffer determines how many bytes are read.
- The busio.SPI.readinto() function is called to read 4 bytes of data from the MAX31855. Remember the size of the passed in buffer determines how many bytes of data are read.
- Finally the CS line is toggled back to a high digital logic level. This tells the MAX31855 we’re done talking to it and it can stop listening or sending data.
Notice the returned data has the hex value 0x01A81AF0. Just like with the MCP9808 you’ll need to check your device’s datasheet to see how to interpret the data. In this case you can again make a little Python function to convert the raw bytes into temperature data:
>>> def temp_c(data):
... temp = data[0] << 8 | data[1]
... if temp & 0x0001:
... return float('NaN') # Fault reading data.
... temp >>= 2
... if temp & 0x2000:
... temp -= 16384 # Sign bit set, take 2's compliment.
... return temp * 0.25
...
>>> temp_c(result)
26.5
Awesome, a value of 26.5 degrees Celsius was read from the sensor! Try touching the thermocouple with your finger and running the SPI read code again to get another temperature reading to compare. Remember to toggle the CS pin low and then back high in between reading SPI data!
Although the MAX31855 doesn’t require it and it’s not shown above, you can also use the busio.SPI.write() function to send data over the MOSI line. For example to send the bytes 0x01, 0xFF you would run:
>>> spi.configure(baudrate=5000000, phase=0, polarity=0)
>>> cs.value = False
>>> spi.write(bytes([0x01, 0xFF]))
>>> cs.value = True
Just like with reading data you want to make sure the bus is configured for the right speed, phase, and polarity. Then you toggle the CS line low to tell the device you’re about to talk to it, send data with the busio.SPI.write() function, and toggle the CS line back high again.
Finally, don’t forget to call busio.SPI.unlock() to unlock the SPI bus and let other code use it:
>>> spi.unlock()
Again you might want to put this all in a try-finally block to make sure unlock is always called, even if something fails and throws an exception. Here’s an example of a complete sensor read with the try-finally syntax:
>>> while not spi.try_lock():
... pass
...
>>> try:
... spi.configure(baudrate=5000000, phase=0, polarity=0)
... cs.value = False
... result = bytearray(4)
... spi.readinto(result)
... cs.value = True
... finally:
... spi.unlock()
...
>>> result
bytearray(b'\x01\xa8\x1a\xf0')
That’s all there is to the basics of reading and writing data with the SPI protocol and the built-in SPI APIs of CircuitPython. However just like with I2C there’s a handy SPIDevice library that can simplify talking to SPI devices.
SPIDevice Library
You saw above how to interact with a SPI device using the API built-in to CircuitPython. Remember using the built-in API requires careful management of the lock and unlock functions to access the SPI bus, and explicit manipulation of the chip select line for a device. If you’re writing code to talk to a SPI device you might find using the CircuitPython bus device library a bit easier to manage as it controls locking & unlocking, and the chip select line automatically (using Python’s context manager with statement).
To use the bus device library you’ll first need to install the library on your board.
First make sure you are running the latest version of Adafruit CircuitPython for your board.
Next you'll need to install the necessary libraries to use the hardware--carefully follow the steps to find and install these libraries from Adafruit's CircuitPython library bundle. Our introduction guide has a great page on how to install the library bundle for both express and non-express boards.
For Express boards, install the entire bundle. For non-express boards, with limited space, you'll need to grab just the adafruit_bus_device folder from inside lib to the lib folder of your board’s CIRCUITPY drive.
Once you have the bus device library installed you can use the SPIDevice class to simplify access to a device on the SPI bus. First setup the SPI bus and CS line exactly as you did before:
>>> import board
>>> import busio
>>> import digitalio
>>> spi = busio.SPI(board.SCK, MISO=board.MISO)
>>> cs = digitalio.DigitalInOut(board.D2)
Now import the bus device module and create an instance of the SPIDevice class. Notice the SPIDevice class needs to be told the SPI bus, chip select line, baudrate, polarity, and phase of the SPI connection. These details will be remembered by the SPIDevice class so it can automatically lock and configure the bus appropriately (again using Python’s with statement and a context manager):
>>> from adafruit_bus_device.spi_device import SPIDevice
>>> device = SPIDevice(spi, cs, baudrate=5000000, polarity=0, phase=0)
Now you’re ready to interact with the SPI device instance using the same read and write functions as before. However this time you’ll put your code in a with statement context manager and it will automatically lock the bus, assert the CS line, configure the SPI bus, and unlock the bus when done:
>>> from adafruit_bus_device.spi_device import SPIDevice
>>> device = SPIDevice(spi, cs, baudrate=5000000, polarity=0, phase=0)
Notice you didn’t need to call configure or even change the CS line from high to low and back. The SPIDevice class takes care of all these details for you automatically!
You can even call the write function just like on the SPI bus directly and data will be written out the MOSI line. One important thing to note is that the CS line is asserted for the entire with statement block, so if you need to make two different transactions be sure to put them in their own with statement blocks. Another thing to note with the SPI device class is that it currently only supports devices with a chip select (it is not optional) and whose chip select is asserted with a low logic signal. Devices asserted with a high logic level are rare and uncommon so the SPI device class should cover most needs.
Just like with I2C you typically don’t need to go straight to these low-level SPI protocol requests (using built-in APIs or the SPIDevice class), instead look for a library to interface with your hardware. In this case the CircuitPython MAX31855 library is exactly what you want to use to talk to this thermocouple sensor. Using a library simplifies access to the sensor data and saves you from writing all the complex SPI transaction code. However if your device doesn’t have a library you might need to interface with it directly using code like the above!
Software SPI & I2C
As mentioned above there are some cases where using the hardware’s SPI (or even I2C) support isn’t possible. Perhaps you have so many devices you’ve exceeded the available pins or resources, or maybe the hardware bus pins aren’t accessible. In these cases you can fall back to a software-driven, or sometimes called ‘bit-banged’, approach to driving the SPI protocol. This approach uses simple digital I/O lines to read and write SPI protocol data. The big difference between hardware and software SPI is speed–with software SPI it will run much slower than hardware SPI because toggling digital I/O is slower than dedicated hardware SPI. However in many cases like reading this temperature sensor the speed of the bus doesn’t matter and you can use software SPI.
To try software SPI re-wire the MAX31855 as follows:
Fritzing Source
- Board 5V or 3.3V output to MAX31855 VIN
- Board ground/GND to MAX31855 GND.- Board D4 to MAX31855 CLK/clock
- Board D3 to MAX31855 DO (data output, AKA MISO)
- Board D2 to MAX31855 CS/chip select
Notice all of the SPI lines are connected to digital I/O lines. You can actually change these to any other digital I/O lines (but you’ll need to modify the code to match!).
Now import and configure the CS line exactly as before:
>>> import board
>>> import digitalio
>>> cs = digitalio.DigitalInOut(board.D2)
>>> cs.direction = digitalio.Direction.OUTPUT
>>> cs.value = True
At this point you’re ready to configure the software SPI bus by using the bitbangio module. The bitbangio modules provides all of the software-based protocol support, like SPI and I2C. Luckily the interface to the bitbangio classes is exactly the same as for the busio hardware-based interfaces so your code doesn’t change much beyond what library it imports and how it creates the SPI class:
>>> import bitbangio
>>> spi = bitbangio.SPI(board.D4, MISO=board.D3)
Just like with the busio SPI class the bitbangio.SPI class is created and told the clock line and MISO line (it can also optionally be told the MOSI line). Notice all these lines are just simple digital I/O pins that you wired above.
Now using the software-based SPI bus is exactly the same as with hardware like above! Try the same code:
>>> while not spi.try_lock():
... pass
...
>>> spi.configure(baudrate=5000000, phase=0, polarity=0)
>>> cs.value = False
>>> result = bytearray(4)
>>> spi.readinto(result)
>>> cs.value = True
>>> result
bytearray(b'\x01\xa8\x1a\xf0')
Awesome you received 4 bytes of temperature data just like with hardware SPI! The big difference here is that code in CircuitPython is driving the digital pins to run the SPI protocol instead of hardware built-in to the microprocessor. This means the call to read data is a little slower (it’s certainly not running at 5MHz like requested) but for most devices they don’t care about the slower speed.
Again remember to call bitbangio.SPI.unlock() to unlock the software SPI bus!
>>> spi.unlock()
Don’t forget you can even use the SPIDevice library with the bit-bang I2C bus!
You can do the exact same software ‘bit-bang’ trick with the I2C protocol too (even using the I2CDevice class). Use the bitbangio.I2C class from the bitbangio module in place of the busio.I2C class. The interface between the two classes is the same so you just change how you import and create the I2C interface, for example:
>>> import board
>>> import bitbangio
>>> i2c = bitbangio.I2C(board.D3, board.D2)
The above would create a software I2C interface using D3 as the clock and D2 as the data line. This is handy for adding more I2C peripherals or using pins other than SCL and SDA. Again the speed is slower, but most devices don’t care about speed. Also note on some boards like the ESP8266 software I2C is required!
Have questions or comments? Continue the conversation on TechForum, DigiKey's online community and technical resource.
Visit TechForum