How to read a power quality report
A logger produces a hundred pages and a recommendation on the last one. Four measurements decide what you actually need, and knowing which they are turns the report from a sales document into a specification.
Any supplier will log your incoming board for a week. The report that comes back is genuinely useful — and it is also the document that justifies whatever they are about to quote. Reading it yourself takes twenty minutes and changes the conversation.
One: voltage, minimum and duration
Do not look at the average. Look at the minimum, and then at how long the supply spent below your equipment’s tolerance. A supply averaging 405 V and dipping to 340 V for two seconds forty times a day is a stabiliser problem. The average conceals it entirely.
Then check whether those dips cluster. Shift-start clustering points at motor loads on the estate. Midday clustering, on clear days, points at solar export. Random distribution points at the feeder itself.
Two: interruptions, and how they were counted
This is the number that decides UPS against stabiliser, and it is the easiest to present selectively. Ask what threshold counted as an interruption and what the logger’s sampling interval was. A device sampling once per second cannot see a 200 ms break — and a 200 ms break is long enough to reset a CNC controller.
A report showing zero interruptions from an instrument that samples every second is not evidence that there were none.
Three: phase imbalance
Compare the three phase voltages against each other across the week, not against nominal. If one sits consistently three or four per cent below the others, a balanced stabiliser will correct on the average and leave that phase low.
This single line decides balanced against independent-phase, it is the most common specification error we are called out to correct, and it is frequently not called out in the summary at all.
| Measurement | What to look at | What it decides |
|---|---|---|
| Voltage | Minimum and time below tolerance | Stabiliser, and its range |
| Interruptions | Count, duration, sampling interval | UPS, and which topology |
| Phase imbalance | Phases against each other | Balanced or independent-phase |
| THD | Voltage and current, separately | Isolation, filtering, or nothing |
Four: distortion, and which kind
Reports quote THD without always saying whether it is voltage or current distortion, and they are different problems with different owners. High current THD is generally yours — non-linear loads inside the plant. High voltage THD with low current THD generally arrives from outside.
It matters because the remedy differs. Your own harmonics are addressed with filtering or by separating the offending load; distortion arriving on the supply is addressed with isolation. Buying the wrong one solves nothing and is not cheap.
Three questions for whoever hands you the report
What was the sampling interval? What threshold counted as an interruption? Are the phase voltages compared against each other anywhere in this document? A report that cannot answer all three has not measured what it claims to have measured.
The recommendation on the last page
Read it after the data, not before, and check that each line traces back to a measurement. A recommendation for a UPS on a report showing voltage sag and no interruptions is not supported by its own evidence.
We log supplies and we quote against what we find, which is a conflict worth naming rather than pretending does not exist. The defence against it is that the raw data goes with the report — you can check the reasoning, and you should.
Questions we get asked
Anything not covered here, call us — most of it is faster to answer on the phone.
How long should a power quality survey run?
Seven consecutive days, minimum, covering every shift and at least one weekend. Anything shorter misses the weekly pattern, and the weekly pattern is usually where the answer is — estate load on a Monday morning looks nothing like a Sunday night.
What sampling interval should the logger use?
Fast enough to catch what damages you. Sub-cycle for transients, and no slower than 200 ms if interruptions matter. A logger sampling once per second cannot see a 200 ms break, which is comfortably long enough to reset a CNC controller — so a clean report from a slow instrument proves very little.
Is high THD always a problem?
No, and the type decides. Current distortion is generated by your own non-linear loads and matters mainly for transformer and neutral heating. Voltage distortion arriving from outside matters for sensitive equipment. Around 5% voltage THD is a common threshold for concern, but a figure means little without knowing which kind is being quoted.
Should I trust a survey done by the company quoting for the equipment?
Trust it, and check it — the conflict is real and the data is usually sound. Ask for the raw logs rather than the summary, confirm the sampling interval, and verify that each recommendation traces back to something measured. A supplier who will not release the raw data is asking you to take the conclusion on faith.