Telemetry that drops out intermittently is uniquely frustrating, because the link isn't broken — it's marginal. It works most of the time, then loses packets just often enough to corrupt data, trip alarms or break an SLA. The cause is almost never the one obvious thing people swap first; it's usually one of a handful of issues that only bite when conditions line up.
First, read the pattern
Before chasing causes, notice when the dropouts happen. Do they cluster at a certain time of day, follow a machine's cycle, get worse in rain or heat, or only affect nodes at the far end of the network? A marginal link betrays itself through its pattern, and a few notes on timing will rule out half the suspects before anyone opens a toolbox.
The usual causes
Most telemetry dropouts trace back to one of these:
- Congestion and airtime limits — too many nodes, or duty-cycle rules on licence-exempt bands, so transmissions collide and retry until they fail.
- Interference and a raised noise floor — nearby drives, power supplies or other radios lifting the noise so the receiver can no longer hear weak nodes cleanly.
- A marginal link budget — the link was always running on the edge, so a little rain, foliage growth or a temperature swing tips it over.
- Antenna and cabling faults — a loose connector, water ingress, a damaged or wrong-impedance feeder, or an antenna knocked out of alignment.
- Receiver desense — a strong transmitter co-located with the receiver overloading its front end so it briefly goes deaf.
- Power problems at the node — brownouts or a failing supply that resets the radio under load.
Why it gets misdiagnosed
The instinct is to swap the radio, then the antenna, then the node — and sometimes the symptom moves around enough to feel like progress. But if the real issue is marginality or interference, replacing healthy parts changes nothing and buries the evidence. The link doesn't need new hardware; it needs measuring.
How an RF engineer pins it down
The work is about turning "it drops sometimes" into numbers. That means measuring the received signal level and the noise floor over time at the problem nodes, logging exactly when drops occur, and running a spectrum survey to see whether an interferer is present and how it behaves. If there is one, direction finding traces it to source. If the air is clean, attention turns to the link budget, the antenna system and the node's power. Either way you finish with evidence of the cause, not another box swapped on a hunch.
Two things worth doing first
Log the timestamps of the dropouts and note what else was happening — weather, which machinery was running, time of day. Patterns are gold. Then physically check the antenna and feeder at the worst node: connectors tight, cable undamaged, no water in the connector. Beyond that, resist swapping radios blind.
See wireless diagnostics →When to get someone in
If the dropouts are corrupting data you rely on, breaking an SLA, affecting safety or monitoring, or you've already swapped kit and they persist, that's the point for RF measurement. Telemetry that's only mostly reliable is often a quick fix once it's measured — and an expensive guessing game until it is.