A power monitoring system should be sized around normal and peak load, required measurement accuracy, waveform capture, reporting interval, and the available panel space and communications infrastructure.
The decision starts with one monitored point per electrical circuit, then adds the CT ratio and burden, meter accuracy, network topology, and installation work before a platform is chosen.
Define measurement scope and accuracy before selecting hardware
Energy-management metering normally covers consumption in kWh, demand in kW, and power factor, while power-quality monitoring adds sag and swell detection, harmonic measurements, transients, and waveform capture [S1]. A basic energy meter therefore does not satisfy a request for waveform capture, and a power-quality meter may be unnecessary where the operating requirement is only demand and consumption.
The right accuracy depends on the load range, not only the nameplate meter class. Under IEC 62053-22, a Class 0.5S meter at unity power factor is permitted within ±1.0% from 1% to 5% of rated current and within ±0.5% from 5% to maximum current. At reduced power factor, the limits are ±1.0% from 2% to 10% of rated current and ±0.6% above 10% [S3].
For billing-grade or compliance work, the meter also needs an accuracy level appropriate to its intended use. The PowerMonitor 5000 is positioned for Class 0.2 revenue-grade applications, advanced power-quality measurement, and main incoming service monitoring [S1]. A buyer should obtain the combined meter-and-CT error at each circuit's lowest normal load, because the meter class does not include CT error [S3].
Size CTs to operating current and required burden
CT selection should start with the circuit's lowest normal load, expected maximum load, conductor arrangement, and available secondary wiring burden. Sizing a CT to the breaker rating can produce several percent measurement error when the circuit operates at a small fraction of that rating, which is a common failure in lightly loaded production circuits [S3].
IEC 61869-2 Class 1 allows ±3.0% ratio error and 180 minutes of phase displacement at 5% of rated current, tightening to ±1.0% and 60 minutes at rated current. The tighter S classes include Class 0.5S at ±0.75% ratio error and 45 minutes phase displacement at 5% of rated current [S3]. These figures describe standardized limits; actual performance also depends on burden and operating point.
Burden must remain within the CT rating for both the meter and the complete secondary circuit. Longer control cable runs increase conductor resistance and therefore burden, while terminal resistance, wire size, and the number of shared secondary conductors further affect the installation. For a retrofit, document metres of cable, panel shutdowns, and containment changes before approving the architecture [S3].
Compare the main monitoring architectures
Three architectures fit most industrial and commercial projects: wired Ethernet for permanent plant networks, RS-485 with Modbus RTU for panel-bus installations, and wireless for retrofits where cable access is expensive. [S4]
Ethernet offers the straightforward path for fixed meters with an available port, VLAN, and firewall policy. Modbus RTU over RS-485 supports up to 32 unit loads on one segment without a repeater; every meter needs a unique address, the same baud rate, a cable drop, and termination at both ends [S3]. A wireless mesh can reduce installation work, but a radio survey is still required because walls, metal enclosures, and router placement affect coverage.
LoRaWAN is suitable for a small number of widely separated points, not dense one-minute electrical data. In EU868, the three default uplink channels have a 1% duty cycle; at spreading factor 12, a 12-byte payload occupies about 1.5 seconds, allowing roughly one uplink every 2.5 minutes [S3]. A power monitoring system using this architecture should be selected for slower remote readings rather than high-resolution plant telemetry.
Match the platform to reporting and integration duties
A useful platform must preserve detailed historical data, support configurable time ranges, and permit downloadable reports and data export. These functions are necessary to compare energy use by shift, process, or site and to investigate changes that a live dashboard does not explain [S4].
Alerting should be tied to measurable events such as voltage spikes, unexpected loss of monitoring, or consumption above a defined threshold. Delivery by email, SMS, or an application can be useful, but the threshold and the response owner must be specified during selection [S4]. For control-system integration, determine whether the selected meter supports EtherNet/IP or serial communication and whether the plant network can carry the resulting data [S1].
For facilities that also need temperature, vibration, or air-quality measurements, select a platform that can accept those sensor classes without replacing the electrical backbone. Open APIs and native building-management or automation integrations reduce duplicated data paths [S4]. Existing electrical assets, including power transformer incomer monitoring, should be represented in the same data model where cross-equipment analysis is required.
Account for installation, commissioning, and ownership
Retrofit economics are driven by panel shutdowns, cable distance, containment availability, and the number of modified panels. Count those items separately for every point, then compare the combined installation cost with the accuracy gained. A wireless sensor can reduce cable and containment work, while a wired Ethernet meter can provide a more predictable permanent network [S3][S4].
Commissioning should verify the CT ratio, phase sequence, polarity, meter-to-CT pairing, reporting interval, time stamp, network address, and alarm delivery at the lowest normal operating load. If the CT is purchased separately and paired during installation, ask for calibration of the complete meter-and-CT assembly so systematic CT error can be corrected rather than averaged out [S3].
Ownership also determines the communications choice. An IT-managed site network creates a formal change process for ports, VLANs, and firewall rules, while an industrial bus may be easier to isolate and commission locally. Neither option removes the need for data validation, cybersecurity review, and a defined maintenance responsibility [S3].
Use these limits to build the final shortlist
A sound shortlist contains one instrument architecture for each required function: Class 0.5S or better for precision energy metering, Class 0.2 for qualifying revenue-grade duty, and a power-quality platform when sag, swell, harmonics, transients, or waveforms are required [S1][S3]. CTs must then be selected for the measured current range, phase-error class, and secondary burden rather than breaker rating alone.
Do not select a mainstream wireless mesh for dense one-minute electrical data across a large plant, a 32-device RS-485 segment for additional instruments, or a basic energy meter when the specification requires waveform capture. The final acceptance test is the lowest normal load, not a full-load bench demonstration, because lightly loaded circuits are where oversized CTs and generic accuracy claims fail in service [S3].
Before purchase, record for every monitored point: CT ratio, minimum and maximum expected current, meter accuracy class, required power-quality measurements, reporting interval, protocol, cable route, and shutdown count. The next trackable signal is a completed point schedule paired with a combined accuracy statement and a communications load calculation for the proposed meter network.