Why fleet telematics hardware fails — and how to catch it early.

A field guide from the crew that does the wiring · P.G. Auto Installs, Laurel, MD

When a gateway stops reporting or a camera goes dark, the first assumption is usually a bad unit. Sometimes it is. Far more often the hardware is fine and the install is failing — and the difference matters, because replacing a good unit with another one wired the same way buys you a few months and the same phone call. Here is how to tell the two apart.

1. Read the failure pattern first

Before opening a dash, look at how the failures are distributed. The pattern usually names the culprit.

• Intermittent, worse on rough road — mechanical. A connector backing out, a splice working loose, a unit hanging on its cable. Almost never the electronics.
• Clustered by install batch — if the twelve vehicles wired the same week are the twelve failing, that is a workmanship pattern, not a hardware lot.
• Clustered by vehicle type — points at the harness or the connector spec for that type being wrong.
• Dies with the ignition, or drains the battery overnight — the power source was chosen wrong: switched where it needed constant, or constant where it needed switched.
• Permanent, single vehicle, no pattern — now you may genuinely have a bad unit.

2. The six install failures we see most

Power taken from the wrong place

The most consequential decision in a 12-volt install is where the power comes from. A device that needs constant power wired to a switched circuit stops recording the moment the key comes out — which is precisely when a parking incident happens. A device wired constant that should have been switched will sit and drain a battery over a weekend.

Worse is the back-probe guess: piggybacking on a circuit without metering it, on the assumption that it looks right. That is how you end up sharing a circuit with something that spikes, and how you find out months later when both things misbehave.

Splices that were never really splices

Scotch locks, wire nuts, and electrical tape do not survive a work truck. Vibration, heat cycling, and moisture take them apart on a schedule. A taped splice will test fine on install day and be intermittent by spring. Proper crimps, and solder with heat-shrink where it belongs, are not craftsmanship for its own sake — they are the difference between a connection that lasts and one that is a scheduled future failure.

No strain relief

A connector that carries the weight of its own cable will eventually back itself out. Secured routing with strain relief means vibration cannot work the connector loose. This one failure accounts for a large share of the "unit keeps dropping" calls we get.

Routing through places wiring does not belong

Cable run across a pedal box, through an airbag deployment zone, or pinched under a trim panel is a problem waiting for a specific day. Airbag zones are the serious one: wiring routed through a deployment path is a safety issue, not a tidiness issue. Routing belongs clear of pedals, steering, and every airbag.

Wrong harness for the connector

The diagnostic port on a heavy tractor is not the port on a cargo van, and neither is the port on a pickup. Light-duty runs OBD-II; heavy-duty runs multi-pin connectors that vary by make and model year. A universal adapter forced onto the wrong port produces a unit that reads nothing, or reads garbage that looks like data until someone checks it.

Never verified

The quiet failure: the install is complete, the light is on, and nobody confirmed the unit is actually reporting into the platform. The vehicle gets counted as done. The gap surfaces weeks later when someone pulls a report — or during an inspection.

3. Symptom-to-cause table

4. What a bad install actually costs

The install itself is a small line item. What it protects is not.

• Compliance exposure. An ELD that logs gaps produces exactly the record an inspection is looking for. That is an installation outcome.
• Missing footage on the day it matters. A camera that was not recording during the one incident you needed it for provides no defense, and the entire investment in the safety program was made for that day.
• Decisions made on incomplete data. Routing, coaching, and utilization decisions built on a dataset with holes in it are worse than no decisions, because they carry false confidence.
• Rework at fleet scale. One vehicle is a service call. Forty vehicles wired to the same bad spec is a second rollout you are paying for twice.
• Driver trust. Hardware that rattles, hangs, or visibly looks thrown in tells drivers exactly how seriously the program is meant to be taken.

5. A fleet audit you can run yourself

You do not need an installer to do a first pass. Pull a representative sample — a few of each vehicle type, ideally from different install batches — and walk this list. Anything you cannot check off is worth a closer look.

• Every unit in the sample is reporting in the platform right now, not "last week"
• No device is hanging on its own cable or resting loose under the dash
• No visible scotch locks, wire nuts, or taped splices behind the dash
• Cable is secured along its run, not draped or zip-tied to a single point
• Nothing routed across the pedal box, steering column travel, or an airbag zone
• Factory trim is fully seated, with no missing or broken clips
• Camera sits in the wiper sweep zone, high on the glass, out of the driver's sightline
• Both camera lenses produce a clean image — road-facing level, driver-facing framed
• No vehicle in the sample has unexplained data gaps in the last 30 days
• No ELD in the sample is logging recurring data diagnostic events
• Ask two drivers whether anything rattles, hangs, or gets in the way

6. What a correct install looks like

This is the standard we hold on every vehicle, and it is a reasonable standard to hold anyone to:

• Fused power on the correct circuit, metered rather than assumed
• The correct harness for that vehicle — never a universal adapter forced onto the wrong port
• Proper crimps, and solder with heat-shrink where it belongs. No scotch locks, no wire nuts, no tape splices
• Routing secured and strain-relieved, clear of pedals, steering, and every airbag zone
• Factory trim back on like it never came off
• Verified live in your platform before the keys go back — powered, connected, reporting, clean image on every lens

None of that is exotic. It is ordinary competence applied consistently, which is exactly what is hard to buy at fleet scale and exactly what determines whether your telematics investment produces data or produces service calls.