Author: Leon Zhang — Power Monitoring and Testing Equipment Team, Baoding Houfeng Trading Co., Ltd., specifying permanent and portable power quality monitoring for utilities, industrial plants and energy managers
Technical review: reviewed by the in-house power monitoring team against IEC 61000-4-30:2025, IEC 62586-2:2017 and EN 50160:2022 practice
Last updated: 19 October 2026
An AC power quality monitor records what the voltage and current at a point in the network actually do over time. It does not answer “is there power here” — a multimeter answers that. It answers a harder question: how far the supplied power deviates from the ideal, how often, and for how long. That is the record a maintenance or energy team needs when equipment overheats, a drive trips without an obvious fault, or a customer disputes the quality of the supply.
The words matter here, because they are standardised. GB/T 32507-2024 defines power quality as the degree to which the electrical characteristics at a specified point deviate from benchmark technical parameters. Which parameters, and how they are measured, is exactly what separates a monitor from a data logger with a voltage input. A logger stores samples. An AC power quality monitor reduces those samples into the quantities a contract, a standard or a failure investigation is written against.
Key takeaways
- A monitor is bought for two families of data, not one: aggregated steady-state values (10-minute frequency, RMS voltage, unbalance, harmonics, flicker) and discrete events (dips, swells, interruptions, transients). A unit that does only one leaves half the record empty.
- IEC 61000-4-30:2025 replaced the 2015 third edition and removed Class B entirely. Any specification still asking for Class B is describing a withdrawn edition.
- Channel count decides where the instrument can be installed and whether a disturbance can be traced back to its source. It cannot be upgraded afterwards.
- Rising and falling edges matter as much as levels: a 2×2 mm² voltage lead and a 200 A CT are the accuracy limits on most installations, not the instrument’s own error budget.
- A permanent monitor and a portable analyser answer different questions. Buying one to do both is the most common reason monitoring projects stall in week two.
- A record with no time reference is a graph. A record synced to relay targets and the plant log is evidence.
What an AC power quality monitor measures
An AC power quality monitor records two families of quantity, and almost every specification argument comes from confusing them: aggregated steady-state values reduced to 10-minute intervals, and discrete events stored individually with their depth and duration. IEC 61000-4-30:2025 sets out both families and the methods for getting repeatable results, and everything below is drawn from that list rather than from a datasheet. Frequency, supply voltage magnitude, flicker, unbalance, harmonics and interharmonics, rapid voltage changes, mains signalling voltage, and the current quantities — magnitude, harmonic and interharmonic currents, current unbalance — make up the steady-state side. Dips, swells, interruptions and transient voltages make up the event side.
| Parameter | What an AC power quality monitor records | Why teams buy it |
|---|---|---|
| Power frequency | Actual frequency and its variation, 10 s and 10 min aggregation | Contractual compliance, generator and inverter behaviour |
| Supply voltage | RMS against nominal, slow drift, sustained over- and under-voltage | Transformer tap problems, long feeders, PV back-feed |
| Voltage dips and swells | Depth, duration, affected phases, one record per event | The single largest cause of unexplained equipment trips |
| Interruptions | Short and long loss of supply, split from dips by duration | Transfer scheme and UPS performance |
| Harmonics and interharmonics | Per-order magnitude and phase, not one summed number | Locating the source of distortion, not just its presence |
| Flicker | Pst and Plt over standard aggregation periods | Arc furnaces, welders, and complaints about lights |
| Voltage unbalance | Negative- and zero-sequence components | Single-phase loads on a three-phase supply, motor heating |
| Rapid voltage changes | Step changes distinct from dips, per the 2025 method | Tap changer and capacitor switching |
| Transient voltages | Very short high-magnitude events, sampled not aggregated | Switching transients, capacitor restrikes, lightning |
The split between the two families is the first thing to understand about a specification, and it is the one most often blurred. Steady-state quantities are reduced to aggregated values — 10-minute intervals are the usual unit, and compliance is expressed as a percentage of intervals within limit over a week. Events have no aggregation that makes sense; each one is stored with its depth and its duration, and the count and distribution are what get analysed.
That difference drives the memory and communications requirement. A week of 10-minute values is a small file. A year of events on a distorted supply is not, and it is the events that fill the memory up first.
What the 2025 edition changed
The fourth edition of IEC 61000-4-30:2025 was published on 24 October 2025, with a corrigendum issued in July 2026, and it is not a cosmetic revision. It folds in the 2021 amendment and the 2016 corrigendum, rewrites the measurement method for rapid voltage changes, extends the treatment of voltage events, splits the old Annex C so that conducted emissions in the 2 kHz to 9 kHz band sit in one annex and a new method covers 9 kHz to 150 kHz, removes the underdeviation and overdeviation parameters, and removes Class B.
The Class B removal is the item with commercial consequences. Class B was a lighter class used for surveys and for instruments that could not meet Class A. A tender document written from a 2015 template can still contain “Class B or better”, which now points at a class that no longer exists in the current edition. The supplier has to guess whether the buyer means Class S or an S-class instrument from a superseded edition. Ask for the edition year in the specification, not just the class letter.
Online monitor or portable analyzer?
This is the first decision to settle, and it is decided by the question being asked rather than by budget, so the useful move is to state the question before comparing models. A permanent installation catches the event that happens at 3 a.m. and never reproduces while an engineer is on site; a portable survey establishes which feeder is misbehaving before any budget is committed. Our power quality monitors are the fixed side of that split, covering single-channel and dual-channel feeders up to 8-channel switchboard monitoring, plus harmonic-only monitoring from the 2nd to the 30th order. Our portable power quality analyzers are the carried side: a portable meter, a handheld analyzer, a three-phase current unbalance analyzer and a waveform recorder.
| Question being asked | Instrument that answers it | What it cannot do |
|---|---|---|
| Which feeder is causing the problem? | Portable analyzer, moved feeder to feeder | Prove how often it happens after the survey week |
| How often does it happen? | Fixed monitor on the identified feeder | Tell you which feeder to choose |
| Did we comply with the contract this month? | Class A fixed monitor with a defensible record | Explain a transient nobody specified |
| What happened during that trip? | Monitor with event log plus relay targets | Replace a waveform recorder at MHz sampling |
| Is the distortion my load or the utility’s? | Dual-channel monitor, supply versus load | Separate two loads on the same feeder |
A portable survey is the cheaper first step and it is not a substitute for the second one. The survey answers “what kind of problem is this”; the monitor answers “how often, and can I prove it”. Projects that install a monitor without a survey usually end up with the instrument on the wrong feeder, and projects that only survey end up with a week of data and a dispute they cannot win.
Monitoring the transformer behind that supply is a different measurement problem again. Dissolved gas analysis watches the insulation, not the voltage, and we cover it separately in transformer online DGA monitoring.
Channel count: the number that decides the installation
Channel count is the one specification that cannot be upgraded later, because it fixes where the instrument can physically be installed and what a single record is capable of saying on its own. Four channels is the usual minimum for one three-phase circuit with neutral; a three-phase circuit without neutral, or a single-phase circuit, needs fewer. Buying more channels than the panel has space for is as much a mistake as buying too few.
- Single channel — one circuit. Suits a single critical feeder, a generator output, or a dedicated load such as a large drive where the question is local. See the single channel power quality monitor.
- Dual channel — two circuits in one instrument, the practical minimum for comparing an incoming supply against the load it feeds, or for separating the utility’s contribution from your own. See the dual channel power quality monitor.
- Eight channels — a whole switchboard or multi-feeder board on one time base, so a disturbance on one feeder can be placed against what the others did in the same millisecond. See the 8-channel power quality monitor.
The reason multi-channel matters is correlation, not coverage. A single-channel record shows that something happened; a synchronised multi-channel record shows which feeder moved first, and that ordering is usually the whole answer. On a board with six outgoing feeders and a common incomer, eight channels is the difference between “the plant had a dip” and “feeder 4 started the dip and the incomer followed”.The channel you leave out is the one that explains it. A four-channel installation on a three-phase-plus-neutral feeder has no spare channel for the incomer, so it can record the disturbance but never the direction it came from. If the site has a recurring event with no known source, one spare channel on the upstream side is worth more than a higher sampling rate.
Harmonics: the parameter that most often changes the answer
Harmonics are the parameter that most often changes the answer, for the simple reason that modern networks carry increasing nonlinear load — converters and rectifiers, variable-speed drives, electrified railways, arc furnaces, PV inverters and EV chargers — and distortion produced by the load cannot be read from a nameplate. It has to be measured, per order, on the site it occurs on.
An aggregate total harmonic distortion figure is enough to raise a flag and not enough to act on, which is the distinction that decides whether a harmonic-capable instrument was worth buying. The order tells you the source. Low-order odd harmonics point at rectifier and switching topologies. Triplen orders — 3rd, 9th, 15th — accumulate in the neutral of a four-wire system rather than cancelling, which is why a neutral can carry more current than its phases on a supply whose THD looks only mildly high. Higher orders point at resonance with power factor correction capacitors, and that is where a modest distortion figure turns into a failed capacitor bank.
That is why a dedicated harmonic monitor such as the HF-PQ-206 harmonic monitoring device reports the fundamental plus the 2nd to 30th voltage and current harmonics individually, on busbars and lines, rather than summarising them. Two numbers from the same record do most of the diagnostic work: the current spectrum at the load, and the voltage spectrum at the busbar. Distortion that is high in current and low in voltage is a local load problem with a stiff supply. Distortion in both is a source feeding an impedance that is not stiff enough, and the fix is upstream of the load.
Interharmonics and the 2 kHz to 150 kHz band are the part of the spectrum where the standards are still being written. EN 50160:2022 names the phenomena and gives limits for mains signalling voltage but leaves interharmonic limits under consideration; IEC 61000-4-30:2025 adds a measurement method for conducted emissions in the 9 kHz to 150 kHz range. An instrument that stops at the 30th harmonic will not see a PV inverter’s switching sidebands at all. If the site has inverters, chargers or a PLC system, ask what the instrument does above 2 kHz before comparing accuracy figures in the range everyone quotes.
What the standards require from the instrument
The standards that matter here serve four different purposes, and mixing them up is a reliable way to buy the wrong instrument, because each one answers a question the others do not address. Four of them do the work, and only one of them defines an accuracy class.
- IEC 61000-4-30:2025 defines how the measurement is made — method, accuracy, time aggregation, flagging — and defines two classes. Class A is the stricter one, and measurements made with it are intended to be repeatable and comparable between instruments, which is what makes a record usable in a contract or a dispute. Class S is the lighter class: comparable results, with less onerous requirements on aggregation and flagging.
- IEC 62586-1:2017 sets the product requirements for power quality instruments and defines the two performance classes, A and S. IEC 62586-2:2017 is the one that matters at procurement, because it specifies the functional tests and uncertainty requirements an instrument must pass to claim a class. A datasheet that says “Class A” without reference to 62586-2 compliance is a claim about the algorithm, not about the instrument.
- IEEE 1159:2019 describes the electromagnetic phenomena themselves, the monitoring devices, and the interpretation of results. It is the document to read when deciding what to monitor and what a record means; it does not set accuracy classes.
- EN 50160:2022 sets the characteristics the supply voltage in European public networks is expected to stay within under normal conditions — for example 95% of 10-minute RMS values within ±10% of nominal over a week, and voltage THD of 8% or less for 95% of the time in low-voltage networks. It is a limit on the supply, and a monitor is how compliance with it is demonstrated.
There is a fifth document that is not a standard but behaves like one in a specification: GB/T 32507:2024, the Chinese terminology standard, which replaced the 2016 edition on 1 April 2025. It fixes the vocabulary of power quality, harmonics, unbalance, dips and swells, and it is the reference to quote when a report has to be unambiguous across a Chinese and an international audience.
| Document | Version in force | What it governs | Replaced |
|---|---|---|---|
| IEC 61000-4-30:2025 | 4th edition, + COR1:2026 | Measurement method, Class A / Class S | 2015 3rd edition |
| IEC 62586-1:2017 | 2nd edition | Product requirements, performance classes | 2013 1st edition |
| IEC 62586-2:2017 | 2nd edition, + AMD1:2021 | Functional tests, uncertainty | 2013 1st edition |
| IEEE 1159:2019 | Revision, active | Phenomena, monitoring practice, interpretation | 2009 |
| EN 50160:2022 | + A1:2025 | Supply voltage characteristics, public networks | 2010 |
| GB/T 32507:2024 | Effective 2025-04-01 | Terminology | 2016 |
The practical rule that follows: if the measurement could end up in front of a customer, a regulator or a contract, the instrument should be Class A to IEC 61000-4-30:2025 and demonstrably tested to IEC 62586-2:2017. If it is for internal trending, a Class S instrument that aggregates consistently is usually the better value — and it is still a Class S instrument to IEC 61000-4-30:2025, not one certified to a retired edition.
Sensors, wiring and time: the accuracy chain nobody specifies
Almost every accuracy argument about power quality monitors is decided before the signal reaches the instrument. Three items make up what is worth calling the accuracy chain, and none of them appears on the monitor’s own specification sheet.
Two of the three are passive and are usually fitted without a second thought, which is why they cause most of the real-world error. The third decides whether the event log can be joined to anything else on site. Current transformers set the current accuracy, and a split-core CT is not a metering CT. A 0.5% voltage channel paired with a 3% clamp-on CT produces a power measurement whose error is dominated by the transducer. For compliance work the CT should be the accuracy class the measurement needs; for a load survey, a clamp is fine as long as the report says so. The same transducer question applies on the voltage side: a 2×2 mm² lead run to a distant panel drops millivolts, and at 1% of 230 V that is the entire budget.
Lead and fuse resistance shift the voltage reading, and the shift is systematic.Voltage inputs draw a small current, typically well under a milliamp, but a long thin lead plus a fuse holder adds a series resistance that appears as a constant voltage offset. It does not average out over a week; it biases every 10-minute value in the same direction. On a compliance record that bias is indistinguishable from a genuine supply drift.
The voltage-lead error is the one that survives every review, because the record looks clean: the monitor is Class A, the CTs are metering class, and the reading is still 1.5 V low all week because the lead and the fuse holder add series resistance nobody measured.
Time reference decides whether the record is evidence.A dip logged at 14:32:07.412 is a graph entry; the same dip logged to UTC with NTP discipline and correlated against a relay target at the same substation is proof of sequence. GPS or NTP synchronisation within a millisecond is what makes a multi-device record usable, and it is the first thing to check when a monitor’s event log disagrees with a relay’s.What a paired record looks like in practice. The useful unit is not a monitor record on its own, but the monitor record beside the relay target list and the plant log for the same minute. When the three agree, the event is understood. When they disagree, the disagreement is the finding — and it is usually the CT ratio, the monitor’s channel assignment, or a clock that has drifted.
Three questions to settle before you buy
The purchase usually goes wrong before any model is compared, so the three questions below are worth settling in writing first. What has to be proved, and to whom? An internal trend and a contractual dispute demand different accuracy classes, different aggregation and different reporting. Decide this first, because it constrains every other choice including channel count and CT class.
Where will the instrument live?A permanent installation means an enclosure, a supply, CT and VT interfaces and a data path. A portable survey means battery, memory, safety rating and weight. A unit that is awkward to install gets left in a cupboard, and a survey set that needs a mains outlet is not portable.
What happens to the data?Stored locally, exported as a record, or pushed into a system someone actually reads? Monitoring that nobody reviews produces a file, not a result. The most common failure of a permanent installation is not a hardware fault; it is a monitor that has been logging correctly for a year into a directory no one has opened.
If the answers point to a permanent installation, the next step is to name the feeders and count the channels before comparing models — the power quality monitor range is organised that way for a reason. If the answers point to a survey, the deliverable to specify is the report format, not the instrument.
Limitations and boundaries
A non-Class A monitor cannot produce a contractual record, however good the trend.Class A is a statement about repeatability between instruments, and it is the comparability that makes a record defensible. An S-class instrument that trends beautifully for a year is still not the instrument to put in front of a regulator.
A monitor measures at one point.Everything upstream and downstream is inferred. A dip at the incomer says nothing about which feeder caused it unless a second channel was recording there at the same time and on the same clock.
Event thresholds are a choice, not a fact.Dip detection depends on the start threshold, the hysteresis and the reference voltage, and different settings produce different dip counts from the same supply. Any event count quoted without the threshold set is not comparable to another count.
Harmonic measurement above the 30th order is not universal.Many instruments stop at the 50th, some at the 30th, and the 2 kHz to 150 kHz band is only covered where the method is implemented. A distortion problem above the quoted range will look like nothing at all.
Aggregation hides short excursions by design.A 10-minute RMS value can sit comfortably within limits while a 200 ms swell destroys a drive. That is not a fault of the instrument; it is the reason events are recorded separately, and the reason a compliance report and a failure investigation use different views of the same data.
A monitor does not fix anything.The single cheapest improvement to an existing installation is usually not a better monitor but a better clock: without time synchronisation the event log cannot be joined to relay targets, and an unjoined event log is only ever read once.
A monitor does not fix anything.It converts an intermittent complaint into a record that can be argued from in a meeting, which is all monitoring was ever asked to do. The remedy — a filter, a UPS, a tap change, a different CT — is a separate decision, made by someone else, using the record.
Need a quotation for a power quality monitor?
Tell us the equipment you need to test, the voltage class and the quantity. Our engineers reply with a technical proposal and a quotation within 12 hours — English manual, remote video guidance and a one-year warranty on exported units.
Payment terms agreed per order · Lead time confirmed per order
Frequently asked questions
What is the difference between an AC power quality monitor and an analyzer?
An AC power quality monitor is installed permanently and records continuously; an analyzer is portable and used for surveys and commissioning checks. That is how intermittent events get caught. Most projects start with an analyzer survey and then install a monitor on the feeder the survey identified.
How many channels do I need?
One channel covers a single circuit. Two channels are the practical minimum for comparing a supply with its load. Eight channels record a whole switchboard on one time base, which is what allows a disturbance to be traced back to the feeder that caused it. Leave one channel spare on the upstream side if the source is unknown.
Do I need harmonic measurement, or is voltage and current enough?
If the site contains converters, drives, rectifiers, electrified railways, arc furnaces, PV inverters or EV chargers, harmonic measurement is the part of the record that explains the symptoms. Aggregate distortion indicates a problem; per-order harmonics indicate its source.
What does IEC 61000-4-30:2025 Class A mean in practice?
It defines the measurement method, accuracy and aggregation so that results from different Class A instruments are repeatable and comparable, and that comparability is what makes a record defensible in a contract or a dispute. Note that the 2025 fourth edition removed Class B; an instrument claiming Class B is described against the withdrawn 2015 edition.
Does a Class A monitor need IEC 62586-2:2017 testing?
Yes, if the claim is to mean anything at procurement. IEC 62586-2:2017 specifies the functional tests and uncertainty requirements an instrument must pass to claim a class, so a specification should ask for both the class and the test evidence behind it.
Can one instrument cover both a permanent installation and site surveys?
Not usually, and not well. Permanent monitoring favours fixed mounting, continuous recording and remote data, while surveys favour portability and battery operation. Treating them as one purchase is a common reason monitoring projects stall after the first week.
If you are not sure which of the two you need, contact our power monitoring team and describe the symptom rather than the specification — what trips, when it trips, and what changed on site beforehand. That is enough to choose between a survey instrument and a permanent power quality monitor, and to decide how many channels the problem actually has.
Conclusion
An AC power quality monitor is bought for a record, not a reading, and the record has three parts that have to line up: the parameters the standard defines, the channels the installation can accommodate, and the clock that makes the events comparable to everything else on site. Get the IEC 61000-4-30:2025 edition right, decide Class A or Class S from who will read the report, count channels before comparing models, and check the CT class before believing any accuracy figure. Those four decisions cost nothing at specification stage and cannot be recovered afterwards.
Technical references
- IEC 61000-4-30:2025, Electromagnetic compatibility (EMC) — Part 4-30: Testing and measurement techniques — Power quality measurement methods, 4th edition, published 2025-10-24, with Corrigendum 1 (2026-07).
- IEC 62586-1:2017, Power quality measurement in power supply systems — Part 1: Power quality instruments (PQI).
- IEC 62586-2:2017, Part 2: Functional tests and uncertainty requirements, with AMD1:2021.
- IEEE 1159-2019, IEEE Recommended Practice for Monitoring Electric Power Quality.
- EN 50160:2022, Voltage characteristics of electricity supplied by public electricity networks, with A1:2025.
- GB/T 32507-2024, Power quality — Terminology, effective 2025-04-01.