Latest / Internet of Things with Fexingo: Connected Devices, Sensors, and Industrial IoT / How IoT Sensors Are Preventing Gym Equipment Failures
Transcript
- Lucas: You know that moment on a treadmill when the belt suddenly jerks, or the motor starts making this grinding noise — and you're just waiting for the thing to die mid-stride? Luna: Oh, I've been there. It's usually right when you hit your stride in a HIIT session, and then — clunk. Machine down. Lucas: Exactly. And for gym owners, each failure means an angry member, a repair call, and lost revenue. But here's the thing — most equipment failures don't happen without warning. The motor bearings vibrate differently a week before they seize. The belt tension changes. The current draw spikes. We've just never listened. Luna: Until now, with IoT sensors. This feels like the same predictive maintenance logic we covered for pipelines and warehouse fans, but applied to something we all use. Lucas: Right. And before we dig into the numbers, I want to say something quickly. If today's conversation gives you a usable idea — whether you run a gym, manage a facility, or just appreciate clever sensor applications — and it felt worth a coffee to you, there's a link at buy me a coffee dot com slash fexingo. We keep this show ad-free, and listener support is what makes that possible. No pressure, just a genuine if it was useful thing. Luna: Yeah, and we mean that. A coffee's worth of support genuinely helps us cover hosting costs and research time. So, thank you if you do. Now — back to those failing treadmills. Lucas: Let's talk about a real deployment. FitCore — a national chain with about 200 gyms — started a pilot in late 2024. They fitted accelerometers and current sensors on the drive motors of treadmills, ellipticals, and some selectorized weight machines. The accelerometers are basically the same industrial-grade units you'd find monitoring a conveyor belt at a factory. Luna: So they're using off-the-shelf hardware, not custom fitness sensors? Lucas: Exactly. They're using something like the ADXL357, a three-axis accelerometer that costs maybe $15 in volume. They mounted it on the motor housing and on the frame near the drive pulley. The current sensors clamp onto the motor power wires. All that data streams to a local gateway, and then to the cloud for analysis. Luna: And what were they looking for? Specific vibration signatures? Lucas: Yeah. The key metric is something called motor harmonic distortion. When a motor is healthy, the vibration spectrum shows clean peaks at the fundamental frequency and its harmonics. As bearings wear, you start seeing sidebands — small extra peaks around the main frequency. Those sidebands appear weeks before any audible noise. FitCore's system flagged that pattern automatically. Luna: Weeks? That's a huge lead time. Lucas: Their pilot data showed an average of 18 days between the first anomaly detection and actual failure. That gave their maintenance team time to order parts, schedule a replacement during off-peak hours, and post an out-of-order sign before anyone got stranded mid-workout. Luna: I can see how that saves on emergency repair calls. But what were the actual cost savings? Lucas: They published results at a facility management conference last quarter. Across the pilot — 500 machines in 20 gyms — unplanned downtime dropped 62% year over year. Repair costs fell 35%, mostly because they stopped paying premium rates for emergency service calls. The sensors and gateways cost about $12,000 per gym to install. They recouped that in under eight months. Luna: So the ROI is clear for a big chain. But what about a smaller gym — a local CrossFit box or a community rec centre? Can they get similar benefits? Lucas: It's harder for the smallest operators, because the cloud analytics platform usually comes with a monthly subscription. FitCore used a vendor that charges around $200 per month per location. For a single-location gym with 20 machines, that's $2,400 a year. If they're only saving one or two emergency repairs annually — maybe $800 each — the math gets tight. Luna: So you'd need cheaper analytics, or a simpler threshold-based alert instead of machine learning. Lucas: Exactly. Some startups now offer a lite version: just a vibration threshold alarm. No harmonic analysis, no trend prediction. You set a limit — if the vibration exceeds, say, 5 millimeters per second — and you get a text. That can run on a $10 microcontroller and cost nothing monthly. It's less sophisticated, but it catches the big failures. Luna: I imagine the privacy angle comes up too. These sensors track usage patterns — how often a machine is used, at what times. Could a gym use that to see which members are slacking off? Lucas: That's a real concern. FitCore was careful to aggregate data at the machine level, not the member level. They only track 'this treadmill was used for 45 minutes today,' not 'Luna ran on it for 45 minutes.' But the capability exists. Some gym management software already integrates with check-in systems, so linking sensor data to individual members is technically possible. Industry guidance from the International Health, Racquet & Sportsclub Association says sensors should never be used for member surveillance — only equipment health. Luna: Good to hear that's being addressed. So beyond treadmills, what other gym equipment benefits from this? Lucas: Ellipticals and stair climbers have similar drive systems — belts, pulleys, motors. Stationary bikes with magnetic resistance are lower risk because they have fewer moving parts. But selectorized weight stacks — the ones where you move a pin to add plates — those have cables and pulleys that fray. A cable snap can be dangerous. FitCore added a simple tension sensor to the cable pulley on those machines. It measures deflection under load. If the cable stretches beyond a threshold, it triggers an inspection. Luna: That's smart. A snapped cable can cause serious injury. So the sensor prevents both downtime and safety incidents. Lucas: Yes. And there's another layer: data that flows back to equipment manufacturers. Life Fitness and Precor have started offering 'connected' machines that report their own telemetry. But the installed base of older machines is huge — some treadmills last 15 years. Retrofitting sensors on legacy equipment is where the real market is. Luna: That retrofitting trend reminds me of the cold chain sensors we covered in episode 38. Same idea: add a sensor to an existing asset, get data, reduce failure. Lucas: Exactly. And the sensor ecosystem is getting simpler. The latest generation uses Bluetooth Low Energy mesh networks, so each machine acts as a relay. No need for a Wi-Fi password or complex IT setup. You screw a sensor onto the motor housing, it joins the mesh, and data flows to a single gateway that might cover the whole gym floor. Luna: How long do those sensor batteries last? Lucas: Depending on the transmission interval, typically two to five years. FitCore set theirs to report every 15 minutes under normal conditions, and every minute if an anomaly is detected. They've been running for 14 months without a battery replacement. Luna: That's practical. So what's the biggest barrier to wider adoption right now? Lucas: Honestly, awareness. Most gym operators still run a reactive maintenance model — wait for a machine to break, then call a technician. The idea of preventing a failure weeks in advance is still new to them. The upfront cost, even if it pays back in eight months, requires a capital budget line that many small operators don't have. And the analytics platforms need to be dead simple — no dashboard that requires a data scientist to interpret. Luna: Right, because the facility manager isn't an engineer. They need a red light / green light system. Lucas: Exactly. And that's where the industry is heading. Some vendors now offer a single 'health score' from zero to 100. Anything below 70 triggers a maintenance alert. No raw vibration data, no spectrum analysis. Just a number. That's the kind of simplicity that will drive adoption beyond the early adopters. Luna: I'm curious — is any of this data used to inform equipment design? Like, if certain motors fail more often, do manufacturers change them? Lucas: They're starting to. FitCore shared anonymized failure data with one of their equipment suppliers. The supplier found that a specific model of treadmill motor had a higher than expected bearing failure rate after 2,000 hours of use. They redesigned the bearing housing in their next revision — added a larger grease reservoir. That change came directly from field sensor data. So there's a feedback loop forming. Luna: That's a great example of IoT not just preventing failures, but improving future products. That's the kind of long-term value that's harder to quantify but maybe more important. Lucas: Absolutely. And it only works if you have enough data across enough machines. One gym's data is noise; a thousand gyms' data is a signal. That's why the larger chains have an advantage — they can aggregate across locations and get statistically meaningful patterns. Luna: So is there a tipping point where retrofitting becomes standard practice? Like, maybe within five years? Lucas: I think so. The hardware cost is already low enough. The analytics platforms are getting simpler. And insurance companies are starting to take notice. If a gym can show it has predictive maintenance on its equipment, some liability insurers are offering premium discounts of 5 to 10 percent. That's a direct financial incentive beyond just repair savings. Luna: That's a compelling argument for any gym owner. You save on repairs, you save on insurance, and your members have a better experience. Lucas: Right. And the member experience piece is real. FitCore's member satisfaction scores in the pilot gyms went up 8 points compared to control gyms. Fewer machine outages means less frustration. People notice when the equipment is always available. Luna: So what's the next frontier for this? Beyond gyms, obviously. Lucas: I'd point to any environment with high-usage electromechanical equipment that's expensive to repair and where downtime frustrates customers. Think hotel laundry facilities, apartment building elevators, even escalators in shopping malls. The same vibration and current sensors apply. The same analytics. The business case is essentially identical. Luna: So the gym is just a really relatable example of a much bigger trend. Lucas: Exactly. And for me, that's the beauty of IoT in this space. The sensors are commodity hardware now. The intelligence is in the pattern recognition. And the value is in a machine that never breaks while you're using it. Luna: Here's hoping my next treadmill run is uninterrupted. Thanks, Lucas. Lucas: Thanks, Luna. And thanks to everyone listening. We'll be back next time with another angle on the connected world.