Start with equipment requirements
Before measurement, know the permitted environment of the particular machine: nominal voltage, phases, frequency, permitted deviations and imbalance, requirements for protective conductor, earthing system, short-circuit capability, harmonics and other parameters stated by the manufacturer. Conditions for the chiller, compressor, extraction and automation matter too.
Use the current installation/electrical manual for the delivered configuration and written supplier replies. A nameplate on a cabinet does not always contain all grid-quality requirements. Do not transfer tolerances from another model or assume that internal power supplies resolve every external deviation.
“Bad voltage” is not a diagnosis
One phrase can hide different phenomena:
- persistent undervoltage or overvoltage;
- short voltage sag or swell;
- complete or momentary interruption;
- phase imbalance;
- harmonic distortion;
- rapid transients;
- impulse overvoltages;
- incorrect earthing or conductor connections;
- voltage drop in an internal line;
- unwanted interaction with a large neighboring load.
Each phenomenon has a different time scale, mechanism and solution class. A device that regulates voltage slowly may not react to a short event. An SPD does not correct persistent undervoltage. An ordinary transformer does not create energy during an interruption.
Record the symptom separately from the electrical event
A machine message, chiller stop or controller reboot is a symptom. To link it to the grid, use one time scale:
- exact alarm timestamp;
- machine state and job;
- auxiliary systems that were starting;
- events of other workshop consumers;
- power-quality analyzer record at the connection point;
- status of protective devices;
- external utility events, where available.
Without time correlation, it is unsafe to conclude “the light flickered, so a stabilizer is needed.” The cause can be an internal contact, overloaded line, compressor start, protection setting or a non-electrical machine state.
Measure an event profile, not once with a multimeter
A manual instrument shows a value at the observation moment but can miss a short sag, transient or periodic imbalance. IEC 61000-4-30 describes methods for measuring power-quality parameters. A practical survey needs a suitable instrument, correct connection, aggregation intervals, event capture and time synchronization.
An electrical specialist defines the measurement plan. It usually covers:
- shift startup;
- laser operation on different jobs;
- switching on chiller, compressor and extraction;
- maximum workshop loading;
- periods when the symptom occurs most often;
- enough days to capture rare events.
Measure where the data answer the question: at the machine incoming supply, the cell distribution board and, if needed, farther upstream. One instrument at the wrong point cannot distinguish an external event from a drop in an internal line.
### Minimum contents of a measurement report
The report must let another specialist reproduce the conclusion without verbal explanation. Record the analyzer model and current configuration, points and connection diagram, recording period, time zone and clock-synchronization method. Include aggregation and threshold settings, the list of monitored parameters, equipment start/stop log, and export of raw events rather than only several screenshots.
Link every conclusion to a specific data interval and corresponding event in the machine log. List gaps separately: loss of analyzer power, changed point, invalid channel or period without synchronization. Compare “before/after” with the same method and retain file versions. Such traceability does not determine cause automatically, but prevents a single chart from becoming an unsupported choice of stabilizer, transformer or other protection.
Compatibility environment and individual machine limits
IEC 61000-2-4:2024 specifies compatibility levels for conducted disturbances in industrial locations up to 35 kV in a defined frequency range. These levels help describe the electromagnetic environment, but do not replace the particular equipment manufacturer’s requirements or the local electrical design.
For a decision, use two columns: the actually measured environment and permitted limits of all sensitive consumers. Request a missing manufacturer characteristic in writing. Do not declare compliance or noncompliance by comparing random readings with an excerpt from a standard.
Event–evidence–solution-class table
| Suspected phenomenon | Evidence required | Possible action class, not a finished selection | |---|---|---| | Unsuitable nominal voltage | Nameplate, diagram, stable measurement | Transformation or change of supply point | | Persistent voltage deviation | RMS trend and load | Regulation, network reconstruction, coordination with operator | | Short sags/swells | Event capture with duration and depth | Eliminate source, strengthen network, special conditioner | | Interruption | Record of zero/low voltage and protection | ART-154 scenario, not an ordinary stabilizer | | Phase imbalance | Simultaneous phase measurements | Redistribution, repair or design solution | | Harmonics | Spectrum, THD/TDD and source modes | Source analysis, reactor/filter/other agreed solution | | Impulse overvoltages | Risk assessment and protection coordination | SPD system and earthing by design | | Local voltage drop | Measurement at line start and end | Check cable, contacts and scheme |
The right column intentionally has no specific model or rating. One problem class can have several sources, and a device interacts with transformer, protection, load and warranty requirements.
When a transformer is needed
A transformer can be needed to match nominal voltage, create a defined supply system or provide galvanic isolation, where the design provides for it and the manufacturer permits it. But it has its own losses, inrush current, impedance, heating, protection, ventilation and earthing requirements.
Before selection, clarify:
- primary and secondary voltage and connection scheme;
- required apparent power and load profile;
- consumer nonlinearity and harmonics;
- permitted drop and short-circuit capability;
- inrush and switching sequence;
- location, temperature, noise and fire requirements;
- machine manufacturer recommendations.
“A transformer with reserve” without a system calculation can worsen selectivity or create a new starting sag.
When a stabilizer or regulator is considered
Regulation is appropriate when it is confirmed that voltage persistently or repeatedly leaves the permitted band and eliminating the network-side cause is impossible or economically unreasonable. Check input range, accuracy, reaction speed, overload capability, bypass, harmonic compatibility and behavior with the actual load profile.
Eaton explicitly separates power-conditioner functions: voltage regulation, sag protection, harmonic isolation and attenuation. This is an important boundary: the word conditioner does not mean one device covers every phenomenon simultaneously.
Do not install a stabilizer merely because it was part of another enterprise’s package. First demonstrate that its function matches the measured event.
Protection from impulse overvoltages
An SPD is part of a coordinated protection system connected with the incoming supply, distribution boards, line lengths, earthing, lightning protection and network characteristics. One module near the machine does not compensate for errors in the whole system.
The designer determines types, levels, coordination and protection of SPDs themselves under applicable rules. Installation without checking conductors and current paths can create a sense of protection without demonstrated effectiveness.
Harmonics and nonlinear loads
Variable-frequency drives, rectifiers and switched-mode power supplies are nonlinear consumers. Harmonic currents can affect voltage, heating of transformers and cables, reactive-power compensation and other loads. But the presence of a VFD does not automatically mean an unacceptable harmonic level.
Measure the spectrum and link it to modes. Then model the point of common coupling, sources and possible resonance. Select a passive or active filter, reactor or other means after that analysis. Arbitrarily adding capacitors is particularly unsafe without checking the harmonic environment.
Voltage sags and large starts
Schneider Electric notes that sags can result from transformer and motor inrush or large step loads. If a laser alarm coincides with compressor startup, check source, lines, startup sequence and load control.
Options can include changing start timing, another start method, a separate line, network reinforcement or a special ride-through solution. Selection depends on sag depth and duration and machine tolerance. An ordinary AVR can be too slow or lack necessary energy capacity.
Earthing and protective equipotential bonding
Power quality is not limited to voltage between phases. Compare the site earthing system, protective conductors, bonding between modules, neutral, shielding and manufacturer requirements. Only qualified specialists perform this work.
Do not create a “separate earth for the machine” as an arbitrary electrode without a system design. Incorrect separation or joining of conductors can worsen safety and electromagnetic compatibility.
Coordination with protection and warranty
An additional transformer, conditioner or filter changes impedance, starting modes and short-circuit currents. Protection, selectivity, disconnection, bypass and maintenance procedures must be checked again.
The machine supplier should confirm in writing that the solution is permissible at the connection point. This is particularly important if a device changes the voltage waveform, neutral or earthing system, or can switch to bypass.
Acceptance test
After implementation, do not stop at “the machine switched on.” Verify using the same modes and measures that revealed the problem:
1. Check installation, labels, protection and documentation. 2. Record parameters without load and in characteristic modes. 3. Reproduce permitted starts of neighboring consumers. 4. Compare the event log before and after. 5. Check bypass and safe failures according to instructions. 6. Set a period for further monitoring.
If a rare event does not occur during a short test, that does not prove it has been eliminated. An agreed observation period is needed.
Eliminate the cause first, then compensate for the effect
If a sag comes from a poor contact, an undersized line or simultaneous starting of two large motors, installing a conditioner near the laser can cost more and be less reliable than correcting the network. The same applies to harmonics: a local filter is not always optimal if the main source is elsewhere in the enterprise.
A useful decision tree is:
1. Confirm the event and its impact. 2. Localize the source from the machine connection point to the incoming supply. 3. Check defects and nonconformities of the existing installation. 4. Assess organizational actions: starting sequence, phase distribution and load mode. 5. Consider network reconstruction. 6. Only then compare specialized corrective devices.
A compensating device is justified when the source cannot be eliminated and its function and economics are confirmed for the measured profile.
Compare the options over their lifecycle
Purchase price does not show the full cost of a solution. For a transformer, regulator, filter or conditioner, include energy losses, ventilation, space, noise, service, consumables, periodic inspections, bypass, repair availability and consequences of failure.
Assess the cost of the problem separately: event frequency, lost time, rejects, recovery, damage risk and effects on auxiliary systems. If an event happens once a year but stops a critical process, an average energy assessment is insufficient. If the deviation is continuous, the solution’s operating losses can be decisive.
Document the basis of design
The final record should explain which problem the system solves, which measurements support that decision, which machine limits were used, and which alternatives were rejected and why. Include the single-line diagram, settings, equipment documentation, responsible people, maintenance plan and acceptance measures.
This prevents a future error where, several years later, the device is seen as a “mandatory stabilizer for all lasers” or removed without understanding its function. Review the basis of design when machine or network configuration changes.
Monitoring after the problem is corrected
An analyzer installed for diagnosis is often removed immediately after commissioning the solution. Yet some events are seasonal or depend on a rare production combination. The minimum observation period should cover conditions in which the problem previously occurred.
Track not only voltage but the same machine outcomes: alarms, restarts, lost cycles, and chiller and compressor state. If electrical indicators improve while the symptom remains, revisit the causal link rather than automatically adding another device.
The log retains instrument settings and time synchronization; otherwise new records cannot fairly be compared with the original baseline.
Typical mistakes
Buying a stabilizer before measurement. A device name is not a diagnosis.
Treating a transformer as a universal filter. Its function depends on its construction and system.
Confusing a sag with complete interruption. Loss of the source needs the ART-154 scenario.
Looking only at average voltage. Short events and harmonics can disappear in an average.
Not synchronizing time. Without timestamps, the link between machine alarm and grid event is invisible.
Ignoring auxiliary equipment. A chiller or compressor can be the source or victim of an event.
Not checking protection after changes. A new device changes the electrical system.
Control checklist
- Are current manufacturer requirements available for the whole configuration?
- Do symptoms have exact times and a machine-state log?
- Were measurements made with a suitable analyzer?
- Were the correct points and sufficient recording duration selected?
- Are persistent deviations, sags, interruptions, harmonics and transients distinguished?
- Has it been established whether the source is inside or outside the site?
- Does the solution class match the confirmed event?
- Were the new device’s apparent power, inrush, harmonics and bypass checked?
- Are earthing and protection coordination agreed?
- Is there written approval from the machine supplier?
- Does acceptance repeat the original measures?
- Has a monitoring period after the changes been set?
Conclusion
Selecting a stabilizer, transformer, SPD, filter or other measure begins by classifying the event. Combine the particular equipment requirements, measurements at the right point, machine log and network model.
The best solution often consists of eliminating the source, a correct scheme, coordinated protection and controlled monitoring rather than one expensive unit. If the problem has not been measured, buying “protection against everything” only transfers uncertainty into a new device.
Need service advice?
Provide the equipment model, symptoms and the conditions in which the issue appears so the service request can be prepared accurately.
Discuss a service task