Off-Grid System Monitoring: Why Voltage Lies About State of Charge and What to Log Instead

Most off-grid systems do not announce their failures. They drift. A bank that charges slightly less each month, an inverter that idles a little warmer than it used to, a load nobody remembers adding — none of these trip a fault code on day one. By the time an alarm sounds, the expensive part has usually been degrading for months, and nobody can say when it started, because nobody was recording.

Monitoring is the difference between a trend you can correct in autumn and a blackout you diagnose in December. This guide covers what to measure on an off-grid inverter system, why some measurements are trustworthy and others are not, and how to set up logging that actually gets reviewed.

Why voltage is the least honest number on the display

The first thing every system shows is battery voltage, and the first mistake almost every owner makes is reading it as a fuel gauge.

Under load, the voltage at the terminals sags — a 12 V battery resting at 12.8 V can sag to 12.1 V with the pump running, with no real change in stored energy. Right after the sun lifts the bank out of bulk charging, the same battery can sit at 13.5 V or more while only half full. On top of that, the voltage curve of a lithium battery is nearly flat between roughly 20% and 80% state of charge: four different charge levels can occupy almost the same voltage.

The numbers below show how far a “12 V” reading can wander while the actual state of charge barely moves.

Condition	Typical 12 V LiFePO4 reading	What it looks like on a gauge
Resting, ~50% charged	13.2 V	“Half full” — plausible
Discharging at high load, ~50%	12.1–12.3 V	“Low battery” — misleading
Just after absorption ends, ~70%	13.4–13.5 V	“Nearly full” — misleading
Cold morning, high load, ~50%	11.9–12.2 V	“Nearly empty” — misleading

Voltage is not useless — it is the right quantity for protection thresholds such as low-voltage disconnect. It is simply the wrong quantity for answering “how much energy is in the bank”, which is the question monitoring exists to answer.

The instrument that answers it: shunt-based coulomb counting

The reliable way to track state of charge is to count the energy in and out. A shunt-based battery monitor measures the voltage drop across a precision resistor in the battery’s main negative cable and integrates that current over time — a technique called coulomb counting. Because the shunt sees every amp that enters or leaves the bank, it accumulates a real state of charge instead of guessing one from voltage.

Shunt installation is where most readings go wrong. Two rules decide accuracy:

• Every conductor crosses the shunt. All loads and all charge sources connect on the system side; the battery terminal itself is the only thing on the battery side. One inverter standby wire or one MPPT sense lead bolted directly to the battery silently skips its current past the shunt, and the state of charge drifts from that day forward.
• Calibrate at a known full. After a genuine full charge — absorption completed, current tapered to nearly nothing — synchronize the monitor to 100%. Repeating this occasionally keeps accumulated error from growing.

A shunt also answers a second question that voltage cannot: how much current is flowing right now, and in which direction. That single number tells you whether the array is actually charging, whether the standby consumption you predicted is real, and whether an unexplained draw exists at 3 a.m.

One data source per quantity

A common commissioning mistake is bolting on several monitoring devices and letting each compute its own state of charge. The inverter estimates from voltage, the MPPT controller estimates from its own model, and a shunt counts coulombs — three numbers that will never agree, and no way to tell which is wrong when they disagree.

The cleaner architecture assigns one authoritative source per quantity:

Quantity	Authoritative source	Why
Battery state of charge	Shunt-based monitor, or BMS over comms	Only a current-integrating device measures it directly
Battery current	The same shunt	One counting point for all in and out flow
PV array output	MPPT controller or inverter MPPT telemetry	It is the only device that sees array current and voltage
AC loads	Inverter output telemetry	It measures what it delivers
Cell-level data (lithium)	Battery BMS	Only the BMS sees individual cells

If the inverter and battery communicate over CAN or RS485, the BMS state of charge is generally the better master — it is computed by the device that sees the cells — but the same principle holds: one number wins, and every display agrees to show it.

What to log, and what the trends mean

Live numbers fill screens; trends prevent failures. The events worth catching are slow ones, and they only become visible in a log:

• Falling full-charge current taper. If the current at the end of absorption is higher this month than last month on the same array, the bank is accepting less and reaching full more slowly — an early sign of capacity loss, chronic undercharging or a weakening charge source.
• Rising idle draw. Standby consumption that creeps up often points to a new parasitic load, a device stuck in a non-eco mode, or corrosion adding resistance somewhere it should not be.
• Shorter time between low-battery events. When the bank hits the low cutoff in progressively fewer days under a similar load, capacity is shrinking — or the load has quietly grown and the sizing no longer matches reality.
• Generator hours per month trending up. On hybrid solar-generator sites, fuel hours are a summary statistic for everything else. Rising hours with unchanged loads mean less PV harvest or a bank that no longer holds what it did.
• Midday float at a lower voltage than before. If the array no longer holds the system at full charge through midday hours it used to dominate, either the array output has degraded or consumption has grown past the original design point.

None of these appear as a fault code. All of them are visible in a weekly glance at logged curves — which is why the log has to exist and somebody has to look at it.

Remote monitoring for unattended and distributed sites

For a weekend cabin, a Bluetooth shunt read on occasional visits is adequate. For any site that matters commercially — telecom or monitoring stations, holiday rentals, distributed fleets of systems — monitoring has to reach the person responsible without a site visit.

Three practical requirements separate useful remote monitoring from a marketing checkbox:

• The system works when nobody is looking. Data must be logged on the device itself and uploaded when connectivity returns, not lost during the outage. An alarm that only appears in the app while the app is open is not an alarm.
• Alerts arrive with context. A useful push message names the quantity, the site and the threshold — “Site 7: battery below 30% at 06:40” — so the recipient can decide whether a truck roll is needed before breakfast.
• The data outlives the visit. Historical curves let an engineer compare this winter against last winter. A system that only shows today’s numbers cannot answer the trend questions in the previous section at all.

For B2B projects this is also a service opportunity: many distributors and installers now resell monitoring as a subscription, with the fleet dashboard generating recurring revenue and, more importantly, early sight of every failure mode described above across the whole customer base.

What the inverter itself should expose

Monitoring architecture is easier when the inverter participates. When evaluating an inverter platform, the checklist to run:

• Metering breadth. The display should show battery voltage, current, AC input and output power, and PV power where an MPPT is built in — the quantities you need for the one-source-per-quantity architecture above.
• Configurable protective thresholds. Low-voltage cutoff, maximum charge current and AC input limits should be settable, and the set values should be visible on the display or app — a limit you cannot read cannot be audited.
• A communication port that speaks to your ecosystem. A serial (RS485 or RS232) or CAN interface for connecting to external monitors, BMS and monitoring boxes. Confirm the protocol is documented before specifying it into a project.
• Logging or at least alarm history. If the unit stores past events, intermittent faults become diagnosable; if not, plan for an external logger.

Our pure sine wave off-grid inverters carry digital displays that present the live operating quantities — battery voltage, load level and status — and the off-grid solar inverter series with built-in MPPT adds array-side visibility on models from 1.5 kVA to 11 kVA. The display gives you the trustworthy instant numbers; pair it with a shunt monitor and a logging device and the system gains the trend layer that displays alone cannot provide.

A commissioning checklist for the monitoring layer

• Shunt on the main negative conductor, everything else on the system side
• Monitor synchronized to 100% after a verified full charge
• One authoritative source per quantity; duplicate estimators removed or ignored
• Protective thresholds (LVD, charge limit) set, then read back and photographed
• Alarm setpoints defined for low SoC, high idle draw and charge failure
• Data destination chosen: local logger, cloud platform or both
• A named person and a fixed interval for the trend review
• Baseline saved: day-one curves of a healthy full charge and a normal evening

The last item is the cheapest insurance in the whole list. Curves from a known-healthy system are the reference that makes every future anomaly obvious.

FAQ

Monitoring FAQ

Can I trust the percentage shown on my charge controller?

Rarely. MPPT controllers estimate state of charge from voltage models, and those estimates drift badly on lithium banks, whose voltage is nearly flat between 20% and 80% charge. Treat the controller percentage as a rough indicator and install a shunt-based monitor — or use the BMS state of charge over a communication link — as the authoritative number.

Where exactly should the shunt be installed?

On the main negative cable, between the battery negative terminal and everything else. Every load and every charge source must connect on the system side of the shunt. A single device wired directly to the battery bypasses the counting and permanently skews the state of charge.

My inverter display and my shunt monitor show different state of charge. Which is right?

The shunt, if it was installed with all conductors crossing it and synchronized at a verified full charge. The inverter usually estimates from voltage. Decide once which source is authoritative, configure the system to show that one, and stop comparing the two — they will never agree, and that is expected.

Do I really need remote monitoring if the system is working fine?

Working fine today is exactly what a drifting system looks like in its early months. Remote logging is what converts a slow failure — declining charge acceptance, rising idle draw, growing generator hours — from a December surprise into an October correction. For any site you cannot visit weekly, it pays for itself the first time it does its job.

Monitoring is not a device purchase; it is an architecture decision made once and reviewed forever. Build it correctly and the system tells you about its problems while they are still cheap to fix.