In this guide 6 sections
A live dashboard pays off where a missed fault costs more than the sensors and the upkeep of their data feed. Think of large inverters, remote plants and battery banks under a service contract. Small sites with easy access can stay on monthly visits. Every alarm needs a ticket and an owner.
Which sites deserve live telemetry
Start with the cost of silence. A string inverter on a rooftop that trips at noon loses a few hours of generation. A central inverter on a remote ground plant can sit dead for a week before anyone drives out to reset the breaker.
Rank sites by three things: distance from your nearest technician, contract penalty for downtime, and how often the site has failed before. Sites that score high on all three go on the dashboard first. The rest can wait.
Ask what the last three faults cost. Count the lost generation, the penalty clause and the second trip for a part nobody had on the van. If the total is small, skip the sensors.
- Remote ground plants with no caretaker
- Battery banks under an uptime clause
- Central inverters and transformers
- Sites with repeated nuisance trips
What the sensors must actually report
Do not stream everything. Pick the readings that change a decision. For a solar plant that means inverter status code, string current, DC input voltage, grid frequency and a plane of array irradiance reading to compare against output.
For a battery bank, track state of charge, cell voltage spread, pack temperature and BMS fault flags. Poll them over Modbus or the vendor’s cloud API, whichever the device supports. Ten well chosen tags beat two hundred noisy ones.
Name each tag in plain words, such as inverter 3 DC input voltage. A technician reading the alarm at night then knows which cabinet to open.
Turn an alarm into a service ticket
An alarm that only flashes red is a notification, not a process. Each alarm type should open a ticket carrying the site, asset serial, fault code, time first seen and a named owner. Nobody should copy details from a screen into a spreadsheet.
Set rules for repeats. If the same inverter throws the same code three times in a day, merge it into one ticket instead of opening three. Add a due time based on the contract. Escalate when it passes.
Close the loop at the end. When the technician closes the ticket, the dashboard should show the alarm as cleared, with the part replaced and the reading that proved the fix.
- Site, asset serial and fault code
- Time first seen and time acknowledged
- Assigned technician and due time
- Parts used and closing note
Costs that people forget to count
Sensors are the visible part. The quiet cost is connectivity: SIM plans, gateway replacements, and a dead data logger that looks exactly like a dead plant. Someone must own the telemetry hardware itself, down to the antenna and the power supply.
Add alarm tuning to the list. A new dashboard produces false alerts for weeks. If technicians learn to ignore it, you have paid for a screen nobody trusts.
Budget a yearly review of firmware on gateways and loggers too. An unpatched edge device is a security hole, and a forgotten calibration on the irradiance sensor makes every performance ratio on the screen quietly wrong.
A cheap way to test the idea first
Pick five sites and run the dashboard for one quarter. Compare the faults it caught against the faults customers reported first. Count how many truck rolls it saved and how many it caused by false alarms.
If the dashboard found faults before the customer did, extend it. If not, go back to scheduled visits and keep the money. Write the decision on the sheet, with the date.
Share the quarterly result with the people who answer the alarms. They know which sensors lie, and their notes shape the next round of sites.
Telemetry site scoring sheet
A sheet with one line for each site. Columns cover distance to the nearest technician, contract downtime clause, past fault count, access difficulty and a total score. Below that, every alarm type is listed with its ticket owner, due time and escalation contact. Print it, fill it for every site on your contract list, and keep the scores beside the date you last visited.
Open a blank worksheet to printQuestions owners ask
Is a dashboard worth it for a handful of rooftop systems?
Usually not on its own, because rooftop systems sit close to your base and are easy to visit. A vendor portal from the inverter maker may already show status and yield. Start there before building anything custom.
How many readings should we collect?
Only those that change a decision. Inverter status, string current and cabinet temperature cover most plant faults. Add more once a real fault exposes a gap, and retire any tag nobody has read in a quarter.
Who should own the alarms?
One named person per shift, with a backup for leave. Ownership sits in the ticket, not in a shared inbox, so an alarm that nobody accepts within its time limit moves up to the service manager.
What if the data logger goes offline?
Treat a silent logger as its own alarm. Raise a ticket after a set gap, so a SIM or router fault is never mistaken for a healthy plant. Then check the gateway power supply first.