How to Find a Failing Radio Before You Send the Truck
Most radio failures are not sudden; they are gradual degradations that nobody was looking at until the site went dark. Finding them early turns a 2 AM emergency dispatch into a planned visit with the right parts, in daylight, bundled with other work.
What a failing radio tends to look like
Radios rarely die with no warning. More often they decline, and the decline leaves fingerprints if you know where to look:
- Intermittent missed polls. One site that drops a poll now and then, while its neighbors are clean, is a candidate. The pattern often shows up first as longer poll cycle times rather than outright loss.
- Rising retries or response latency. A link working harder to deliver the same data is a link with less margin.
- Time-of-day or weather correlation. Problems that appear in the heat of the afternoon, after rain, or at dawn and dusk point to temperature-sensitive hardware, water ingress, or atmospheric effects.
- Drift in received signal level. A slow downward trend in RSL with no configuration change is one of the earliest honest signs of antenna, cable, connector, or amplifier trouble.
- Rising noise or interference. A climbing noise floor can come from a new neighbor transmitter, a failing power supply nearby, or a degrading front end.
- Repeated rebooting or power events. Low battery voltage, a weak solar charge, or a failing power supply often shows up as comms loss before it shows up as a power alarm.
- Asymmetry. A link that works well in one direction and poorly in the other points toward one end's transmitter, antenna, or local interference.
What monitoring usually misses
Most SCADA environments watch for one thing: did the device respond. That is necessary and it is not enough.
- A poll response is binary. It says the link worked, not how close it came to not working.
- Retries hide degradation. If the second attempt succeeds, the data looks fine and nobody is alerted.
- Alarms fire at the cliff. Comms-loss alarms trigger after the link is already down, by definition. They are a notification, not an early warning.
- Radio data lives in separate tools. Signal levels, noise, and radio diagnostics typically sit in each vendor's own management software, away from the SCADA screen, and are reviewed only when someone is already troubleshooting.
- Mixed fleets are hard to compare. Networks built up over years often combine several radio vendors reporting similar quantities in different units. Without normalization, no one sees the whole picture, and a bad radio can hide in the noise of a mismatched data set.
- Telecom and SCADA teams look at different screens. The comms loss that looks like a field device fault to operations may look like a path problem to the radio team. Without a shared view, it is easy to dispatch the wrong person.
A practical way to look for trouble
You can do much of this with the data you already have, even without new tools:
- Baseline every link at a known-good point. Record RSL, noise (where the radio reports it), and poll success at commissioning or after a clean maintenance visit.
- Trend on a regular cadence. Weekly or monthly review of RSL and poll performance per link, sorted by change from baseline, finds the sites worth a closer look.
- Rank by criticality. A custody-transfer site or an unmanned station with no local backup deserves a higher priority than a redundant link to a minor point.
- Check the cheap causes first. Connectors, weatherproofing, antenna alignment, grounding, and power supply health account for a large share of field radio problems.
- Look at the neighborhood. If several radios on the same repeater or hub show degradation together, the problem is probably upstream of the individual radios.
The economics: planned visit versus emergency dispatch
The case for early detection is not mysterious; it is scheduling.
An emergency dispatch carries overtime or after-hours labor, a technician who has to diagnose on arrival without knowing which part is needed, sometimes a second trip for the part that was not on the truck, and the operational cost of running blind or on manual readings while the site is down. A planned visit lets you choose the day, bring the likely parts, combine the stop with other maintenance on the same route, and avoid the outage entirely.
For illustration only, using a modeled assumption of $500 to $2,000 per truck roll (the real figure varies enormously by terrain, distance, and labor model): if early detection converted just 10 to 50 emergency dispatches per year into planned visits, the avoided cost lands in the tens of thousands of dollars before counting any avoided outage. These are illustrative modeled numbers, not measured results. Replace them with your own fully loaded dispatch cost and your own annual count to see what applies to your network.
Compare it against your own outage history.
Aevus Comms (RF Link Intelligence) is designed to put cross-vendor radio signal data and fade-margin trends in one read-only pane, so degradation like the patterns above can be seen while the site is still polling.
Related Articles
The story: 8 radios degrade for 2 months unnoticed.
A modeled 14-day timeline of a radio drifting toward failure.
Bands, degradation modes, and redundancy patterns for SCADA radio.
Ready to see how this applies to your operation? Start a pilot conversation, no commitment, no field changes.