We’re all familiar with Arduino and Raspberry Pi, single-board computers that are helping build the growing Internet of things. Built around ARM-based microcontrollers, they’re low-cost, high-volume items that are easy to craft into prototype hardware, and with their easily addressable IO ports and sensors and actuators that are easy to connect to.
But they’re not the only devices used by the growing Maker movement, where hobbyist hackers build hardware that scratches their itches, and where developers explore new scenarios and try out new ideas on the fly.
I spent last weekend at the 10th Maker Faire in San Mateo, wandering around a huge hall full of devices showcasing the latest maker boards and projects built using them. In a sign of the growing importance of this market, some of the largest exhibitors were silicon vendors and hardware IP companies. ARM, for example, showcased a range of projects built using its processor designs, from the M-series controllers to the more powerful A-series devices. ARM powers most of the maker boards, with Atmel, TI, Cypress, and many others using ARM cores in their processors.
It’s also fascinating to see how cheap this stuff is getting. Raspberry Pi is at the higher end, with a Linux-capable, single-board computer for $35, but at the lower end I was able to buy a Red Bear Labs Bluetooth Low Energy developer kit for $5 -- with everything I needed to design and build an ARM-based BLE (Bluetooth Low-Energy) beacon.
Devices based on the Arduino open source hardware are common and many others use its Wiring programming environment. TI’s Energia development environment for its ARM Launchpad system uses a real-time OS rather than the typical maker board firmware, allowing developers to build multithreaded applications, while still supporting Wiring code (and using a browser-based cloud IDE) reading multiple sensors and driving multiple outputs at the same time. You can use shared variables to pass information between threads or work with TI’s own multitasking libraries.
Where it comes to powering maker boards, ARM’s not alone. Intel’s Galileo and Edison boards based on its Quark devices are increasingly popular, and Intel used the event to showcase the range of sensors that could be connected to its hardware. Similarly, U.K.-based Imagination unveiled a new version of its Creator CI20 board, which is built around the MIPS architecture.
Imagination recently took a step further, releasing a version of its MIPS architecture as FPGA files for use in university silicon design courses. Using off-the-shelf FPGAs, colleges will be able to show students how a modern processor comes together, allowing them to design their own modules and build their own system-on-a-chip (SoC) hardware. Imagination’s educational license doesn’t allow the development of final silicon, but with FPGAs as an important part of the processor design process, the approach gives students the type of design experience needed to build the custom silicon that’s the foundation of the Internet of things.
If you want custom silicon, Cypress’ Programmable SoC (PSoC) is an intriguing device. If you’re designing hardware for a specific task you’re not going to need all the components on a modern ARM-based SOC. With a PSoC developer board, you can connect to a development PC’s USB port and run Cypress’s free tooling to configure the SoC modules you want to use -- as well as write and test your code and any surrounding hardware. Once you’ve built the PSoC you want, Cypress’ tools not only generate the appropriate APIs for the hardware, they also deliver the Verilog code that can be used to manufacture a matching SoC. It’s certainly an economical way of getting started with hardware development: A developer board costs only $4.
That’s where today’s new maker boards and their manufacturers are starting to differ from the first Arduinos and the like: They’re focusing on the process of going from idea to product. Work with TI and you get access to its library of components and sensors; work with Cypress and you get to design your own silicon. Intel and ARM will help you choose the right silicon to make your prototype a product, and a whole army of companies will help you find and work with manufacturers.
One interesting approach comes from Particle, with its Proton wireless module, designed to add wireless connectivity to IoT devices. You can start with a stand-alone Wi-Fi Proton module connected to Particle’s cloud service. Once you’ve built and tested your prototypes, you can buy the Proton’s radio module and build it into your own hardware, while still using the Particle cloud service to connect your devices to your services. Then, when you’re ready to move from initial runs of a 1,000 or even 10,000 devices to full mass production, Particle will connect you directly with silicon vendors -- while keeping you as a service customer so that you don’t have to change the software you’ve developed.
At heart Particle is a cloud service with a hardware entry ramp. It’s a model we’ll see more often, as the Internet of things becomes increasingly important. Building and programming hardware is relatively easy, but managing and maintaining the cloud services they need for connectivity is a major task, and they won’t be a core competency for many businesses. Particle is extending its model to cellular connectivity, with the recently announced Electron module. If you don’t want to negotiate a machine-to-machine cellular deal with an operator, Particle’s own MVNO will provide connectivity.
Maker boards like these are an ideal way to get started with Internet of things development. They’re cheap and easy to use, and they already support a range of cloud services and APIs. With sensor platforms and connected devices, you’re coding close to the hardware, so it pays to work with maker boards when experimenting with new technologies or designing prototypes. Once you’re familiar with the hardware and with the software you’re developing, you can start to build your own specialized hardware.