Today at Spectra, Particle announced five new additions to our platform, all intended to help customers make more intelligent devices. They are:
- Particle on Linux: Particle can now be used to manage and deploy software to any Linux device
- M-HAT for Raspberry Pi: An LTE HAT for Raspberry Pi with on-board EtherSIM and support for low-power operations
- EtherSIM+: An upgraded EtherSIM with high-bandwidth data plans and support for more native cellular connections
- M1 Enclosure: A rugged IP67-ready customizable enclosure for Tachyon, Muon, or Raspberry Pi-based products
- Particle Blueprints: Deployable applications for intelligent devices
The technology landscape we operate in is changing quickly. When we launched 10 years ago, it was barely possible to reliably connect an IoT device to a Wi-Fi network. Now we have a breadth of connectivity options, much faster processors, and AI accelerators. Problems that were once solved with expensive and finicky sensors can now be solved with cameras and microphones. We’ve officially entered the era of “Edge AI.”
But regardless of the terms we use — IoT, edge computing, edge AI — and the particular technologies we select to build our product, we’re all trying to do the same thing; we’re trying to make “things” (devices, sensors, machines, etc.) that do their jobs better. Whether that’s an industrial compressor, a security camera, a robot, or a drone — we take the technologies that are developed for smartphones, PCs, and the internet, and we reassemble them into other products to make a better mousetrap.
Note: That is a metaphor, no one needs an AI-enabled mousetrap. Please, don’t.
While it’s amazing that we keep getting new bits and pieces of tech to use, the landscape is changing so quickly that it’s nearly impossible to keep up. According to IoT Analytics, the average time-to-market for an IoT product has gone from 24 months a few years ago to 41 months today. That’s three and a half years! I guarantee that if you start to build a product today and ship it in early 2028, all of the technology built into that product will be obsolete by the time you launch.
Our job at Particle is to try to wrangle all of the complexity of building an intelligent device so that it’s possible to bring a product to market in six months — after which you can continue to iterate and expand upon the software and models that make the product work. We give you the infrastructure and tooling you need to develop and deploy intelligent devices. And right now our most important goal is to stay ahead of the shifting technology landscape to make great tech accessible to our customers and users as soon as it becomes available.
So, onto our product announcements!
Introducing “Particle on Linux”
Particle has spent ten years providing an application infrastructure for embedded devices powered by microcontrollers. We provide a bare metal runtime to write and deploy software over-the-air to microcontrollers, and a Hardware Abstraction Layer (HAL) that makes that software portable across the different combinations of microcontroller and radio that we support.
A few months ago we launched our first single-board computer, Tachyon — a powerful 5G-connected, AI-accelerated SBC powered by a Qualcomm Snapdragon SoC. By taking the technology that powers modern smartphones and packaging it up into a single-board computer, we make it possible to add quite a bit of horsepower to power an intelligent device — with all of the built-in efficiency and connectivity of the phone in your pocket.
But there are a wide array of single-board computers and application processors on the market, each with a different sweet spot. Raspberry Pis are inexpensive and ubiquitous. NXP i.MX processors are extremely flexible and well-suited for industrial and automotive applications. NVIDIA Jetson scales up to 275 TOPS for the heaviest workloads. While Tachyon hits a great sweet spot, we recognize that customers need a wide array of options, and so we’re excited to bring our software platform to… well, anything that runs Linux.
Installing Particle on a Linux device is as easy as downloading a package and running a setup command — at which point your device can be managed through Particle. You can monitor the performance and connectivity of the device, open a secure remote shell, deploy OS and software updates — all for free for less than 100 devices.
Particle on Linux will be available in beta in Q4; you can register for the beta program here.
But what if you also want your Linux device to have a cellular connection? Well we’ve got that too.
Introducing M-HAT: an LTE HAT for Raspberry Pi with on-board EtherSIM
We’re big believers in the power of the global cellular networks. While many of our customers use Wi-Fi or other radios to connect their devices, every radio has trade-offs — and the biggest challenge with Wi-Fi devices when deployed out into the world is that they are often offline. There’s setup friction, fat-fingered passwords, and IT administrators with strict rules about what’s allowed on their networks — all of which can be avoided with a cellular connection.
Cellular connectivity comes with a cost and complexity burden that can scare folks away. We’ve been eliminating that burden since we first launched support for cellular-connected IoT devices in 2015, and we’re excited to bring that expertise to Raspberry Pi.
M-HAT is a LTE HAT for Raspberry Pi that adds cellular connectivity (specifically LTE Cat 1) to a Raspberry Pi (or any single-board computer with a 40-pin connector). But we don’t just stop at adding the modem; we’re bringing the whole tech stack of our LTE SoMs along for the ride. That means:
- Built-in EtherSIM: M-HAT comes with an embedded EtherSIM+ on-board with a free (up to 100 devices) low-bandwidth data plan for telemetry and larger data plans (see below)
- Low-power modes: M-HAT has a flexible power module on-board that lets you run the system at nano-amp low-power modes, so your device can go to sleep (to be turned on by the real-time clock or a wake pin) and consume nearly zero power
- Microcontroller for low-power operations: If you want your device to have some degree of intelligence while the Raspberry Pi is asleep, you can run an application on the microcontroller within the SoM to continue to operate in a limited fashion while the Raspberry Pi is powered down
M-HAT is available now for pre-order; grab yours now and if you’re hoping to launch a product with it, come talk to our sales team.
Introducing EtherSIM+ with larger data plans and more native profiles
Application processors and Linux-based systems require significantly more data than a microcontroller-based device; an OS update or a download of a new model can easily be hundreds of megabytes.
Tachyon and M-HAT will have on-board an upgraded version of EtherSIM, our embedded SIM that powers all of our cellular-connected devices. EtherSIM+ supports larger data plans (2GB/mo for prototyping, with custom plans available for enterprise customers), alongside some other upgrades.
EtherSIM+ is a true eSIM; not just an embedded SIM but a reprogrammable SIM (industry term: eUICC). This means we can expand our connectivity options beyond the “multi-IMSI” redundancy that we currently provide and also add “multi-profile” redundancy, where we can provide multiple connectivity profiles with intelligent connectivity management to switch between those profiles based on geography and availability.
This means that we can add more native profiles to EtherSIM+ — so that your device can connect in more countries via a native connection (as opposed to global roaming, which is the norm for IoT SIMs). This allows us to navigate countries like Brazil and Turkey that have limited cellular roaming support, as well as future-proofing EtherSIM for whatever changes might come down the road; EtherSIM+ can be updated over the air, just like your software.
EtherSIM+ will launch with Tachyon and M-HAT, and will be coming in 2025 to B-SoM and M-SoM; register here if you’re interested in more information.
Introducing M1 Enclosure: a rugged enclosure for Muons, Tachyons, Raspberry Pis, and their HATs
“Intelligence” in a device often comes from the sensors that the device has to gather data about the physical world. There is a nearly infinite list of sensors one can purchase and integrate within an intelligent device, and in many cases those sensors can be found on an accessory like a HAT or a Qwiic module, making it much faster to integrate that sensor. But those HATs and modules can become an issue when you go to deploy the product; the “stackable” and daisy-chained components make prototyping easy but don’t provide a simple path to deployment — especially for a pilot-scale deployment where you might not want to invest in a full custom PCB and enclosure.
The M1 Enclosure is a rugged but customizable enclosure that makes it possible to ship a product built with a Muon, Tachyon, or Raspberry Pi out into the field. You might recognize it from Monitor One, our industrial equipment monitoring gateway:
The M1 Enclosure is built for real-world deployments; it can be mounted to basically anything, it’s IP67-ready, and it has ports for external sensors, connectors, cameras, power supplies, and/or or antennas. It fits a single-board computer and a HAT inside, along with a LiPo battery. And it can be manufactured with custom colorways and custom branding for your device.
The M1 Enclosure is shipping in December; you can pre-order yours today or register for more information.
Introducing Particle Blueprints: deployable applications for intelligent devices
The more technology we have available to us, the more intelligent our devices can be. But that intelligence comes with complexity. And while Particle is abstracting that complexity at every turn, there is still the issue of combining all of this stuff together. To illustrate the point, here’s a block diagram for an intelligent barrier gate to provide entry into a parking lot:
That’s quite a few components that are necessary to make that barrier gate work! As the scope of our platform has increased, so has the need for tools to provide coordinated management of all of the code, models, infrastructure, and services that make up the “application” that is the brains of your device. Enter Particle Blueprints.
Blueprints are deployable applications for intelligent devices that include everything necessary to run that application. That includes:
- Code to be run on the device and in the cloud
- Files and assets (e.g. models, graphics, etc.) to be deployed to the device alongside the code
- Configuration of the edge services running on the device and cloud services running in the cloud that provide the building blocks of the system
These Blueprints are compiled into a git repo — in other words, a folder — that can be deployed with a script and shared on Github or Gitlab.
By making Blueprints shareable, we hope to go beyond just providing another tool; we will also be building out a gallery of ready-to-deploy applications for common use cases.
Our Blueprints Gallery will include both applications we create as well as anything that members of our developer community choose to share; the more we can make things that work out of the box, the easier it will be for all of us to make devices that are a bit smarter.
Particle Blueprints will be coming soon; register here for more information.