GPIO Zero v1.5 is here!
GPIO Zero is a zero-boilerplate Python library that makes physical computing with Python more accessible and helps people progress from zero to hero.
Today, I’m pleased to announce the release of GPIO Zero v1.5.0. It’s packed full of updates, including new features, bug fixes, and lots of improvements to the documentation.
Guido, the creator of Python, happened across the library recently, and he seemed to like it:
GPIOzero I love you! https://t.co/w3CnUGx3yO
— Guido van Rossum (@gvanrossum) January 17, 2019
Pin factories – take your pick
GPIO Zero started out as a friendly API on top of the RPi.GPIO library, but later we extended it to allow other pin libraries to be used. The pigpio library is supported, and that includes the ability to remotely control GPIO pins over the network, or on a Pi Zero over USB.
This also gave us the opportunity to create a “mock” pin factory, so that we could emulate the effect of pin changes without using real Raspberry Pi hardware. This is useful for prototyping without hardware, and for testing. Try it yourself!
As well as the pin factories we provide with the library (RPi.GPIO, pigpio, RPIO, and native), it’s also possible to write your own. So far, I’m aware of only one custom pin factory, and that has been written by the AIY team at Google, who created their own pin factory for the pins on the AIY Vision Kit. This means that you can connect devices to these pins, and use GPIO Zero to program them, despite the fact they’re not connected to the Pi’s own pins.
If you have lots of experience with RPi.GPIO, you might find this guide on migrating from RPi.GPIO to GPIO Zero handy.
Ultrasonic distance sensor
We had identified some issues with the results from the
DistanceSensor class, and we dealt with them in two ways. Firstly, GPIO Zero co-author Dave Jones did some work under the hood of the pins API to use timing information provided by underlying drivers, so that timing events from pins will be considerably more accurate (see #655). Secondly, Dave found that RPi.GPIO would often miss edges during callbacks, which threw off the timing, so we now drop missed edges and get better accuracy as a result (see #719).
DistanceSensor results come when using pigpio as your pin factory, so we recommend changing to this if you want more accuracy, especially if you’re using (or deploying to) a Pi 1 or Pi Zero.
A really neat feature of GPIO Zero is the ability to connect devices together easily. One way to do this is to use callback functions:
button.when_pressed = led.on button.when_released = led.off
Another way is to set the source of one device to the values of another device:
led.source = button.values
In GPIO Zero v1.5, we’ve made connecting devices even easier. You can now use the following method to pair devices together:
led.source = button
Read more about this declarative style of programming in the source/values page in the docs. There are plenty of great examples of how you can create projects with these simple connections:
An important part of software development is automated testing. You write tests to check your code does what you want it to do, especially checking the edge cases. Then you write the code to implement the features you’ve written tests for. Then after every change you make, you run your old tests to make sure nothing got broken. We have tools for automating this (thanks pytest, tox, coverage, and Travis CI).
But how do you test a GPIO library? Well, most of the GPIO parts of our test suite use the mock pins interface, so we can test our API works as intended, abstracted from how the pins behave. And while Travis CI only runs tests with mock pins, we also do real testing on Raspberry Pi: there are additional tests that ensure the pins do what they’re supposed to. See the docs chapter on development to learn more about this process, and try it for yourself.
You may remember that the last major GPIO Zero release introduced the
pinout command line tool. We’ve added some new art for the Pi 3A+ and 3B+:
pinout also now supports the
--xyz) option, which opens the website pinout.xyz in your web browser.
Zero boilerplate for hardware
The goal of all this is to remove obstacles to physical computing, and Rachel Rayns has designed a wonderful board that makes a great companion to GPIO Zero for people who are learning. Available from The Pi Hut, the PLAY board provides croc-clip connectors for four GPIO pins, GND, and 3V3, along with a set of compatible components:
Since the board simply breaks out GPIO pins, there’s no special software required. You can use Scratch or Python (or anything else).
This release welcomed seven new contributors to the project, including Claire Pollard from PiBorg and ModMyPi, who provided implementations for
PumpkinPi, and the
JamHat. We also passed 1000 commits!
Watch your tone
As part of the work Claire did to add support for the Jam HAT, she created a new class for working with its buzzer, which works by setting the PWM frequency to emit a particular tone. I took what Claire provided and added some maths to it, then Dave created a whole Tones module to provide a musical API. You can play buzzy jingles, or you can build a theremin:
…or you can make a siren:
The Tones API is a really neat way of creating particular buzzer sounds and chaining them together to make tunes, using a variety of musical notations:
>>> from gpiozero.tones import Tone >>> Tone(440.0) >>> Tone(69) >>> Tone('A4')
We all make mistakes
One of the important things about writing a library to help beginners is knowing when to expect mistakes, and providing help when you can. For example, if a user mistypes an attribute or just gets it wrong – for example, if they type
button.pressed = foo instead of
button.when_pressed = foo – they wouldn’t usually get an error; it would just set a new attribute. In GPIO Zero, though, we prevent new attributes from being created, so you’d get an error if you tried doing this. We provide an FAQ about this, and explain how to get around it if you really need to.
Similarly, it’s common to see people type
button.when_pressed = foo() and actually call the function, which isn’t correct, and will usually have the effect of unsetting the callback (as the function returns
None). Because this is valid, the user won’t get an error to call their attention to the mistake.
In this release, we’ve added a warning that you’ll see if you set a callback to
None when it was previously
None. Hopefully that will be useful to people who make this mistake, helping them quickly notice and rectify it.
Update your Raspberry Pi now to get the latest and greatest GPIO Zero goodness in your (operating) system:
sudo apt update sudo apt install python3-gpiozero python-gpiozero
Note: it’s currently syncing with the Raspbian repo, so if it’s not available for you yet, it will be soon.
We have plenty more suggestions to be working on. This year we’ll be working on SPI and I2C interfaces, including I2C expander chips. If you’d like to make more suggestions, or contribute yourself, find us over on GitHub.
Great to see all these enhancements and improvements, and lovely to see the development of the library being democratised by allowing others to contribute. One of the biggest things, of course (which you didn’t mention), is to allow the use of Physical pin numbering using the BOARD prefix. This is brilliant, and I think will allow some people who were dragging their heels about using the library to come on board.
It’s a bit hidden on the ChangeLog, but it’s definitely there on the Basic Recipes page – https://gpiozero.readthedocs.io/en/stable/recipes.html#pin-numbering
Thanks everyone for all the work you’ve put in! :-)
GPIO Zero is great and very convenient. I have one query however which I think I mentioned to Ben before. One thing I do with computers is use them to control simple DC slotcar-type motors with PWM. This works very well at higher power but as you drop off the power, the motors will stall if the frequency is too high. It needs to be below 20Hz to give the kick required to turn the motor, the usual 100Hz just averages out as low power leading to a stall. I corresponded with CympleCy who wrote ScratchGPIO about this and he rewrote that system to reduce the frequency at lower power levels which worked very well. When I raised it with Ben my memory is that he said it was an interesting problem and that something could probably be done but I haven’t seen anything since. Is it possible or would it require somebody to write something to implement it? My Python wouldn’t be up to itso it would be great if someone could look into it. It is truly amazing to see these little motors turning at no more than 3 or 4 turns per second with good torque so would be a really useful addition.
Interesting, certainly shouldn’t be too hard for someone to write a custom Motor class that dynamically adjusts the PWM frequency as the duty-cycle changes. Feel free to create an https://github.com/RPi-Distro/python-gpiozero/issues (otherwise it’ll probably get forgotten about)
Presumably these slotcars are forwards-only and don’t support going backwards?
Ben Nuttall — post author
I remember when I first implemented the Motor class I had issues on low speeds (say 0.1), so had to set to, say 0.5 and then back down to 0.1 for it to register. It seemed it was just down to poor powering, so we took that bit out as it seemed to work ok generally. Maybe your problem is different though. As Andrew said, open an issue and we’ll discuss it there.
Fantastic post! Thanks for all the support for my favorite little SBC family!
Thanks Andrew and Ben.
I have opened an issue on GitHub as suggested.
(I am not actually driving slotcars at the moment but the little motors I use come from a bag of slotcar motors I bought about forty years ago in Proops, Tottenham Court Road. Anyone remember Proops?)
Excellent to see Test-Driven Development gets a mention: “You write tests…Then you write the code to implement the features…Then after every change you make, you run your old tests to make sure nothing got broken.”
A great article. One day somebody should write a small guide listing all the GPIO-Zero variables and how they are accessed and used, with a couple of programming examples. An article like this, to me, only gives a taste of what is here. (Or maybe there is a link to this information that I missed?)
Ben Nuttall — post author
Yeah it’s here: gpiozero.readthedocs.io – see the API docs and the recipes :)
Nice to see more and more added, still my go to way of LEDs, button and anything that can be used as one.
It’s just easy.
I got excited about the Google AIY pin factory, but just read this
These APIs are **not compatible** with the Voice HAT (V1 Voice Kit)
Nice to see Google have moved on and don’t bother too much with their old stuff :-(. Also a shame as that’s the one most of us have from the MagPi and only one outside the USA iirc.
Not GpioZero’s fault!
Now if you could tie the SenseHAT into gpiozero somehow, better (easier ‘button’ control of the joystick?)
Ben Nuttall — post author
The pins on the Google AIY Voice HAT are directly connected to the Pi pins, so they are compatible with GPIO Zero – you don’t need to do anything fancy with pin factories to use them. I think that’s all the warning means.
We added GPIO Zero style joystick control to the Sense HAT a couple of years ago: https://pythonhosted.org/sense-hat/api/#joystick
And there’s another joystick API in my pisense library (shameless plug ;) too:
I’ve been testing the ultrasonic distance sensor and found the improvement of stability and accuracy in v1.5.0 is awesome. I rewrote my code to show more digits because now it’s ‘boring’ to see too many same numbers in row. :)
In fact, the fluctuation is kind of sub-mm order in a good condition.
I knew GPIO Zero, but I realize it’s very nice to use it in my class of young learners along with Pygame, or Pygame Zero lately.
Thank you for the nice work!
GPIO “Zero” would seem to imply Raspberry Pi Zero, but I presume that it’s not limited to that model/series.
Ben Nuttall — post author
No, “GPIO Zero” means “zero boilerplate” and predates the Pi Zero. This is covered in the FAQs: https://gpiozero.readthedocs.io/en/stable/faq.html#why-is-it-called-gpio-zero-does-it-only-work-on-pi-zero
Love GPIO Zero and physical computing! Completed projects (so far): Dishwasher monitor that tracks who unloaded last time, and sends a text to the appropriate kid – called “MiTurn”; a water sensor and alarm for the upstairs laundry room, also with texting – called “iLEAK”. Current one, which is Almost done is a Laser smoke detector and security door alarm. The laser beam is bounced off a kinematic mirror on the far wall back to an detector – called “iSMOKE” of course. Physical computing is fun, and the projects can be inexpensive. The Pi Zero W is my favorite! Keep up the good work with GPIO Zero.
Does this work with MRAA?
Comments are closed