Tingslab — Building a Cellular GPS Tracker from Circuit Board to Cloud
Tingslab was a Toronto hardware startup I co-founded in 2018 to build Tings, a 23-gram GPS tracker that connected over LTE-M and NB-IoT. I led engineering from the circuit board to the cloud, and we reached working prototypes before COVID-19 ended the company in 2020.
At a glance
- My role: Co-founder. I led engineering end to end with a small team: device hardware and firmware, the cloud backend, and the web dashboard.
- Years: 2018–2020
- Founded: 2018, Toronto. Co-founded with Milo Rastgoo.
- Product: Tings, a 23-gram GPS tracker that worked anywhere with cellular coverage. It connected over LTE-M and NB-IoT on a built-in global SIM.
- Stage reached: working prototypes in 3D-printed enclosures that reported location from the device, through uBeac, to a web dashboard. The launch price was set at $79, with a $3-a-month data and cloud plan, and a Kickstarter campaign was planned.
- Ended: 2020. COVID-19 hit our funding, our manufacturing plans and travel-related demand all at once.
The challenge
In 2018, tracking a personal item meant accepting one of two compromises. Bluetooth tags were small and cheap, but they could only find things near your phone or near someone else’s. Cellular GPS trackers worked anywhere, but they tended to be bulky and power-hungry, and they needed their own data plan.
LTE-M and NB-IoT are low-power cellular standards designed for devices that send small amounts of data and sleep most of the time, and carriers were rolling them out around then. Tingslab’s bet was that these networks could close the gap. The goal was a tracker small enough to slip into a bag or tuck under a bike seat, that could report its location from almost anywhere in the world and run for weeks on one charge.
Getting there meant solving problems at every layer at once:
- RF and power design inside a 45 mm enclosure
- firmware that keeps the modem and GPS asleep as much as possible
- a cloud backend that could take in location data from a fleet of devices and serve it back to their owners
Who it was for
Tings grew out of fleet tracking. Our early customer interviews were about tracking the temperature and location of refrigerated trucks and trailers in cold-chain transportation, and customers showed significant interest in seeing their fleets in real time. We then tested the same problem with used-car dealerships, which run lots of twenty or more vehicles with a few employees. For those interviews we built a proof-of-concept device from electronics evaluation boards.
The product we designed, and the business plan we wrote in 2019, were aimed at general consumers with valuable things that move. The plan named seven customer groups: parents keeping track of their children, bicycle owners, vehicle owners, travelers with luggage and handbags, business owners running fleets, drone operators and pet owners.
Our use-case notes went further on paper, with geofences for ski resorts and national parks, elder care, warehouses that alert workers when a truck arrives, and construction sites where machines or materials leave the site.
What we built
The device. Tings measured 45 × 45 × 15 mm and weighed 23 g. It was built around Nordic Semiconductor’s nRF9160, which packs an LTE-M/NB-IoT modem and a GPS receiver into a single module. Around it:
- a built-in global SIM, so owners had no carrier plan to set up
- a 9-axis motion sensor, plus temperature and barometric pressure sensors
- a large programmable button for SOS or check-in, ringed by an RGB LED the owner could set to any color
- a rechargeable lithium-polymer battery that charged over USB-C, with a target battery life of weeks to a month depending on how often the device reported
- a splash-proof enclosure
The platform. Tings reported over MQTT into uBeac, the IoT platform I founded in 2017. uBeac was built on .NET / ASP.NET Core, with MongoDB storing location and telemetry data.
The dashboard. A web dashboard on uBeac showed each device’s location on a map.
The roadmap. The consumer app was designed but never built. The plan was iOS and Android apps with:
- several trackers per account, or several profiles on one tracker
- custom geofences for places like home, school or work
- proximity alerts based on a moving “safe radius” centered on the owner’s phone
- an augmented-reality view for finding a tracker by looking through the phone’s camera
We also planned a developer program to open the platform to third-party apps built around the device.
Under the hood
Parts chosen for size and current. The nRF9160 did the heavy lifting, so most of the board design went into the parts around it: the antenna, a low-noise amplifier (LNA) and the voltage regulator (LDO). Each one was chosen for its footprint and for how little current it drew.
The power budget. On paper, the nRF9160 sleeps at 4 µA. Our board measured 35 µA. For a tracker that spends almost all its life asleep, that gap decides the battery life, so most of the power work went into sleeping longer and waking less often. We tuned the modem’s two power-saving modes, PSM and eDRX, and the network timers that control them: how often the device checks in with the network, and how long it stays reachable afterwards. Carriers don’t all grant the same timers, so we tested with SIMs from three providers: Hologram, iBasis and Twilio. The result was an estimated 42 days on a 750 mAh battery.
Firmware. The firmware ran on Zephyr, the real-time operating system behind Nordic’s nRF Connect SDK, and was written in C. We built and debugged it with SEGGER’s tools.
From device to dashboard. To uBeac, a Tings tracker was one more device type. It published over MQTT to uBeac’s ingestion hub. A decoder in uBeac’s catalog turned each message into location and sensor readings, which were stored as time series in MongoDB. The dashboard’s map then updated live over SignalR. The uBeac page describes that pipeline in full.
Prototypes. The working boards ran in 3D-printed enclosures built to the final 45 × 45 × 15 mm industrial design. We stopped before any production tooling was made.
The business model
Tings was to have two revenue streams: the device, at $79, and a data and cloud plan at $3 a month. Owners would buy the plan from Tingslab, so they never had to deal with a carrier. We planned to collect pre-orders and fund the first production run on Kickstarter, then Indiegogo, and afterwards to sell online through tingslab.com and Amazon.
My role
I co-founded Tingslab with Milo Rastgoo in 2018 and led engineering end to end, with a small team.
- Device: I owned the electronics and firmware, from the nRF9160-based design to the power budget.
- Cloud: I owned the backend on uBeac, the IoT platform I founded. That covered MQTT ingestion from the devices, the .NET services and the MongoDB data layer.
- Dashboard: I built the web dashboard used to track the devices.
Owning every layer meant owning every trade-off between them. For example, a reporting interval picked to save battery also limits how current the dashboard’s location can be. Getting those decisions right across the whole system was the core of the job.
Why it ended
We stopped Tingslab in 2020. COVID-19 didn’t just slow the company down. It hit all three things a pre-revenue hardware startup depends on, at the same time:
- Funding. We had planned a Kickstarter campaign for mid-2019. We held it back while we lined up grant and other funding first, and while certification and manufacturing prerequisites were still open, so it never launched. When money for early-stage hardware dried up in 2020, so did our runway.
- Manufacturing. Component supply and factory timelines for a first production run became impossible to plan around.
- Market. Travel stopped, and luggage tracking, one of the uses we had planned for, stopped with it.
A young company can sometimes absorb one of these shocks. We couldn’t absorb all three at once.
What I learned
- Hardware concentrates risk. A software company can usually keep going through a bad year. A pre-revenue hardware company needs capital, a supply chain and paying customers all at once, and losing any one of them can stop it. Next time I would line up committed revenue or a distribution partner before committing to manufacturing.
- Follow the demand you’ve already found. Our early customer interviews showed real interest in real-time fleet tracking. We went on to design Tings for consumers, and fleets became one of seven customer groups in our plan. An early paid fleet pilot would have given us revenue, and a second market that didn’t depend on a consumer launch.
- Use proven silicon, own the platform. Building on an integrated modem-plus-GPS module got a small team to working prototypes quickly. Owning the cloud side meant we controlled the whole data path from device to dashboard, instead of depending on someone else’s platform for the part customers would actually use.
Outcome
- Working prototypes, end to end: device → LTE-M → MQTT → uBeac → web dashboard, in 3D-printed enclosures
- A measured power budget: an estimated 42 days on a 750 mAh battery, after tuning the modem’s power-saving modes across three SIM providers
- A fully defined product and plan: industrial design, specifications, a $79 price with a $3-a-month plan, a written business plan, an app roadmap and a pre-launch website
- Stopped in 2020, before a production run or the Kickstarter launch
Stack
Nordic nRF9160 (LTE-M / NB-IoT + GPS) · Zephyr RTOS · nRF Connect SDK · C · SEGGER tools · MQTT · uBeac · .NET / ASP.NET Core · RabbitMQ · MongoDB · SignalR