White Papers Forum

Abstract As electric utilities, industrial facilities, and renewable generation sites expand into remote off-grid locations, reliable cellular connectivity is not always available. Low Earth Orbit (LEO) satellite communications, specifically Starlink, offer a high-bandwidth, low-latency alternative for field telemetry. Power Monitors, Inc. cloud-enabled power quality recorders, including the Tensor, Bolt, Seeker, Seeker+ (Wi-Fi and Ethernet configurations), and Revolution, can be paired with Starlink systems to stream power quality data into the PQ Canvass platform. This whitepaper details the network architecture, Carrier-Grade NAT (CGNAT) mechanics, local DHCP requirements, and device-specific communication profiles required to successfully deploy PMI instruments with Starlink. Technical Overview & Network Architecture Connecting industrial field instruments to satellite internet requires navigating two critical realities of Starlink’s network design: public WAN address routing and local LAN address enforcement. Standard Starlink residential, mobility, and enterprise hardware plans assign the satellite terminal a private, non-routable WAN IP address (within the 100.64.0.0/10 block) using Carrier-Grade NAT (CGNAT). Under CGNAT: The Starlink terminal cannot receive uninitiated inbound IP traffic from the public internet. The satellite terminal is not assigned a static public IP address. Standard inbound port forwarding is ineffective, as incoming traffic is blocked upstream a

ABSTRACT Every PMI recorder before the Seeker+ starts a recording with an initialization step: the measures, averaging interval, and capture thresholds are chosen up front, bundled into a settings file, and uploaded to the meter. Initialization is how a Tensor, Seeker, Guardian, or Bolt tailors a recording to the job at hand, and with a saved preset it’s quick work — but it happens before you know what the circuit is going to do. The Seeker+ drops the step entirely. The no-initialization workflow is presented here, along with what a Seeker+ records by default and how to work with a recording while it’s still running. THE INITIALIZATION STEP On a familiar circuit, initialization is routine. The measures, interval, and thresholds that worked last month will work this month, and a monthly compliance recording can run from the same preset for years. Troubleshooting is a different job. There, the right settings depend on what the circuit is going to do, and that’s the very thing the recording is supposed to find out. Choose an averaging interval too coarse and the event you were sent to find is averaged into the background; too fine and storage fills before the phenomenon recurs. Set a capture threshold too tight and you come back with thousands of uninteresting waveforms; too loose and you come back with none. Experience and a rule of thumb usually land the settings close enough. When they don’t, the fix is a second site visit with new settings. The Seeker+ takes no recording set

Abstract The Seeker+ runs a power quality survey right out of the box. Mount it, energize it, and it detects the circuit hookup and logs every measure it supports, each aggregated on whatever interval the governing standard calls for. Nothing gets selected in advance, so nothing gets left out. It samples voltage at 1 MHz and current at 200 kHz, fast enough to hold a microsecond transient intact and to cover the supraharmonic band from 2 kHz to 150 kHz. Inverters, EV chargers, and variable frequency drives all emit in that band, well above where a conventional survey instrument stops looking. Capabilities Voltage is sampled at 1 MHz, or 16,666 samples per cycle at 60 Hz. An impulse a few microseconds wide, of the kind a lightning strike produces, registers as a shape and not a single anomalous sample. An analyzer working in the tens of kilohertz sees the same event as a bump of unknown origin, if it sees it at all. Rise time, peak, and the ringing frequency that follows are usually enough to name the source. The current channels sample at 200 kHz, 3,333 per cycle. That covers current harmonics through the 70th, and it resolves the current excursion that goes with a voltage event. A sag with a matching current rise started at the load. A sag without one arrived from upstream. There are four voltage channels and four current channels, rated 0 to 600 V RMS per phase and 0 to 5,000 A RMS. Each input covers its full range, so there’s no clamp setting to get wrong on a crowded panel

Abstract If you were to ask a handful of customers what they consider a ‘power surge’ to be, you would get a multitude of answers — anything from a lightning strike to the lights flickering off briefly, to the lights getting brighter for a moment before returning to normal. This last symptom is caused by the topic of this paper: Voltage Swells. Utilities tend to log far more sag reports than swells, though swells can stress equipment in ways that a sag never could. What is a voltage swell? IEEE 1159 defines a voltage swell as an increase in RMS voltage between 1.1pu (per unit) and 1.8pu (Nominal voltage is equivalent to 1.0pu, or 100%). For example, on a 120V nominal system, 1.0pu is 120V, 1.1pu is 132V, and 1.8pu is 216V. These swells have a duration from 0.5 cycles to one minute, with typical magnitudes seen settling in between 1.1-1.2pu. Swells are characterized into three different categories based on their magnitude and duration: Instantaneous swell: Duration of 0.5 to 30 cycles and being 1.1-1.8pu Momentary swell: 30 cycles to 3 seconds, 1.1-1.4pu Temporary swell: 3 seconds to 1 minute, 1.1-1.2pu. Looking at Figure 1, we can see two classifications adjacent to voltage swells – transients and the over-voltage area. If the duration of the event is longer than one minute, the condition is no longer classified as a voltage swell but as an over-voltage condition. Over-voltage conditions are steady-state regulation issues measured against ANSI C84.1 limits. Events that are sh

Abstract As electronic power converters are increasing in prevalence on the grid, especially EV chargers, photovoltaic inverters, variable frequency drives, and LED lighting systems, a new power quality issue is arising and gaining in popularity within the research community. This relatively new problem is leading to symptoms like conductor power losses, aging and degrading of insulating materials, damage to MV cable terminations, interference with appliances and energy meters, nuisance tripping, flickering lights, and severe issues with power line communications [1]. This is known as supraharmonic distortion, which is any quasi-stationary distortion on the voltage or current waveform in the frequency range of 2 kHz – 150 kHz [2]. The quasi-stationary clause in the definition is meant to exclude transient behavior, which should be treated as a different issue with different causes. The lower end of the supraharmonic frequency range is chosen to correspond with the point where traditional harmonic standards stop measurement, while the upper end is chosen to distinguish from the CISPR 16-1-1 frequency band B, which covers 150 kHz to 30 MHz and is primarily concerned with conducted emissions [3]. Properties of Supraharmonics Here are some properties of supraharmonics. Supraharmonics are generally not phase locked to the 60 Hz line, because they usually come from switching power converters instead of from the 60 Hz line interacting with some nonlinear load. Supraharmonic current

Abstract PMI View’s live data pages show waveforms, phasors, and tabular values from a connected device, updating as you watch, without interrupting an active recording. During initialization, the live data pages can help the user catch reversed CTs, swapped phases, and bad ratios before anyone leaves the panel. Viewing Live Data for a Device The device page shows a Live Data card once you are connected, with the available pages depending on the device model and its hardware and firmware capabilities. Each page pulls readings directly from the instrument, independent of any recording configuration. Figure 1. Live Data Card on the Device Page The pages are complementary. Live Waveform, Vector Diagram, Live Meter and Live Harmonics each present the same electrical inputs from different points of view — shape, phase relationships, numeric values, and frequency content. On devices with a cellular modem, the Cellular Status page provides information about signal quality and network registration. Live Waveform Live Waveform is a continuously updating single-cycle capture of voltage and current — the fastest way to confirm that every input is alive and that the signals look right. Two viewing modes answer different questions. Per Channel isolates each channel with its voltage and current overlaid, so you evaluate one input at a time. Per Measure stacks all voltages together and all currents together, so you compare phases against each other — useful for confirming phasor rotation be

Abstract In the complex landscape of electrical distribution, identifying the origin of power quality disturbances requires more than just high-speed sampling; it demands precise temporal synchronization across the entire monitoring network. As power systems evolve with the integration of distributed energy resources and sophisticated industrial automation, the margin for error in event correlation has narrowed significantly. This white paper examines the critical role of microsecond-level timing accuracy in root-cause analysis, focusing on how Global Positioning System (GPS) integration overcomes the inherent limitations of traditional Network Time Protocol (NTP) methods. By exploring the feature set provided between the PMI Seeker’s hardware-level synchronization and the advanced visualization capabilities of PQ Canvass software, we demonstrate how precise timestamps allow investigators to transform disparate data points into a coherent chronological narrative. Ultimately, the ability to align waveforms from geographically separated monitors enables engineers to distinguish between upstream grid events and downstream facility issues with greater certainty. The Role of Precise Synchronization in Monitoring Hardware The foundation of any forensic power quality investigation into a sub-second duration disturbance is the integrity of the timestamp assigned to a measured disturbance. Historically, many monitoring devices relied on internal clocks or Network Time Protocol (NTP) t

Abstract A loose neutral is one of the most dangerous faults a single-phase service can develop, and one of the least likely to announce itself. The connection does not fail outright. It degrades, intermittently breaking and remaking the return path while the customer reports little more than flickering lights and the occasional dead outlet. PMI has covered the fault itself across a decade of white papers; what has changed is the finding of it. Merlin , the AI analyst built into PQ Canvass , now examines a recording automatically, explains in detail what is wrong and how severe it is, answers questions about the findings through Merlin Chat, and writes the finished report. This paper uses the loose neutral as the test case: a brief review of the mechanism and its signature, then a demonstration of Merlin carrying the investigation from raw recording to finished document. The Anatomy of a Loose Neutral In a 120/240 V split-phase service, the neutral conductor has one job: carry the difference. Each of the two hot legs serves its own 120 V loads, and the neutral returns whatever current the two legs do not share. When the loads are balanced, the neutral carries almost nothing. When the connection degrades (a corroded lug at the weatherhead, say, or a set screw worked loose by thermal cycling), the return path picks up resistance; the service then begins to misbehave in a very particular way. With the return path compromised, the two 120 V loads are no longer referenced to a sol

Abstract Revolution meters are compact, cellular-connected power monitors. Unlike the Tensor , Seeker , Guardian , and Bolt devices, which initiate their own uploads to PQ Canvass , Revolutions do not push their recordings to the server on their own — the recording has to be retrieved on the user’s behalf. Revolution download addresses this by allowing users to initiate a server-side pull of the current recording directly from the PQ Canvass web interface. The server connects to the Revolution through the Revolution’s cell modem, retrieves the recording, validates the file, and stores it in PQ Canvass. Users are notified by email — and optionally by SMS — when the recording is ready to view. This paper describes the download workflow, the optional restart-recording behavior, and how the resulting recording is integrated into the standard PQ Canvass viewing and reporting pipeline. Initiating a Download from PQ Canvass From the device page of any Revolution meter registered in PQ Canvass, users can click the Recordings button to access the recordings page (Figure 1), and then click the Download Recording button to begin a download (Figure 2). The Revolution meter must be online to access the Download Recording button. This button appears only on Revolution device pages; devices in other PMI families upload their own recordings and do not require a server-initiated pull. Figure 1. Revolution Quick Access View page. Figure 2. Revolution device page with the Download Recording but

Abstract As American homes shift from gas furnaces to heat pumps, the load that used to burn gas now arrives at the residential transformer — in winter, at the coldest hours, on every house at once. Shipment data from the Air-Conditioning, Heating, and Refrigeration Institute show heat pumps have outsold gas furnaces every year since 2022, and the gap keeps widening — heat pumps led by about 32% in 2024, the largest margin on record,¹ with annual U.S. shipments now running near 3.5 to 4 million units.² Figure 1. U.S. shipments of heat pumps and gas furnaces, 2022–2024: heat pumps have led every year, with the margin widening to about 32% in 2024.³ The industry’s attention has gone to demand; this paper covers what lands on the power-quality engineer: compressor-start sag, cold load pickup, flicker, and harmonics, appearing in more recordings every season. Using a real field recording from a 120/240 V residential heat-pump service, we show you what each signature looks like, how to attribute every event to the customer or the utility with three simple checks, how to rule out the failure modes that mimic each other — and how to automate the entire investigation, so the hours go into the fix instead of the first pass. How a Heat Pump Shows Up in a Recording A single-speed compressor draws a locked-rotor inrush of five to eight times its running current at every start.⁹ That inrush is strongly inductive, so it depresses service voltage more than a resistive load of the same size

Abstract Most distribution feeders were planned around one-way power flow: from centralized generation, through substations and feeders, to customer loads. The rapid growth of Distributed Energy Resources (DERs) — rooftop solar, customer-sited batteries in export mode, and community solar — is changing that model. When local generation exceeds local demand, the surplus flows upstream toward the substation. This is reverse power flow. Reverse flow is no longer rare on DER-rich feeders; it can occur daily during high solar production and low load. It is not automatically harmful, and modern systems can accommodate some bidirectional flow. The challenge is knowing where it occurs, how often, and whether it is contributing to voltage, protection, or planning concerns. That challenge is largely a visibility problem. Reverse flow often begins at customer services, secondaries, or laterals that aren’t continuously monitored, and traditional data may lack the direction, resolution, or location to catch it. Power quality (PQ) monitoring closes that gap by recording real power, RMS voltage, reactive power, power factor, and interval min/avg/max values. Throughout this paper we use the load convention: positive real power = import from the grid; negative real power = export back to the grid. Once monitor setup is verified, negative kW is the clearest field indication of reverse flow. How Reverse Power Flow Occurs Distribution feeders typically operate as radial systems, with power flowi

Abstract Power Quality recordings are the primary source of evidence in any PQ investigation. The problem is that these recordings can be weeks long and contain hundreds of captures, which makes sorting through them all a time-consuming process. PMI’s newest addition to Merlin allows it to reference and embed clickable snapshots directly into its analyses, taking the reader straight to the relevant evidence, whether that is a zoomed graph, a waveform, or an event table. This paper covers what a region of interest is, how Merlin identifies one, how it surfaces that region as a snapshot, and how to find these snapshots in PQ Canvass . Regions of Interest in a Recording PQ recordings can contain an enormous amount of data, with the vast majority of it looking nominal. This makes scrolling through a recording by hand both tedious and prone to overlooking the moments that actually matter. A region of interest (ROI) is a specific time interval that Merlin ’s analysis has flagged as worth a closer look, such as the window surrounding a voltage sag or a burst of harmonic distortion. In the past, these regions were referenced in the written analysis, but the burden was on the investigator to locate the corresponding portion of the recording manually. This often meant cross-referencing a timestamp against a graph and scrolling until the right window came into view. Now, Merlin provides a link that opens a snapshot of the referenced ROI in a pop-up window, so the user can view the evide