Skip to main content

Power quality for laboratories: stabilisers, UPS, generators and the earth nobody checked

Voltage swings and load-shedding damage laboratory instruments in ways the warranty will not cover. How to decide which loads need a UPS and which need only a generator, size the protection honestly, and get it into the equipment budget.

The instrument ran for eight months and then the power supply board failed twice in six weeks. The supplier asked what the mains had been doing. Nobody had logged it, there was no recorder on the board, and the claim went nowhere. That sequence repeats in laboratories across Bangladesh. The protection that would have prevented it costs a small fraction of the instrument, and the evidence that would have won the argument costs almost nothing.

Four separate things the supply does to your equipment

These are distinct problems with distinct remedies, and a device that fixes one does not fix the others. Most disappointing installations come from buying one box and expecting it to cover all four.

Deciding what needs what

Protecting everything to the same standard is how power budgets get rejected. Sort the load list by how much interruption it can tolerate and buy accordingly.

LoadInterruption toleranceWhat it needs
Servers, data systems, chromatography and imaging workstationsNone. An unclean shutdown risks the data as well as the machineOnline double-conversion UPS, autonomy covering generator start plus an orderly shutdown
Instruments with a sample running: spectrometers, chromatographs, electron microscopesNone. A lost run is a lost sample and sometimes a damaged column or filamentOnline UPS backed by generator, and the manufacturer's documented shutdown sequence
High vacuum pump sets on those instrumentsNone. An uncontrolled vent can contaminate a chamber or wreck a columnUPS sized to complete the controlled vent or hold vacuum until the generator picks up
Ultra-low temperature freezers, cold rooms, incubators, cell cultureMinutes to hours; thermal mass buys timeGenerator supply, plus independent temperature logging and an alarm that reaches a person
Fume hood extract, gas detection panel, scrubber controlsNone, on safety groundsEssential supply from the generator, with the detection panel on its own battery backup
Autoclaves, ovens, furnaces, water baths, washersTolerant. A restart is acceptableGenerator, with a stabiliser if the incoming voltage is poor
Air handling and chilled water serving the laboratoryShort outage acceptable, long outage is notGenerator. Remember that the hoods need this supply air to contain properly
Medical gas or nitrogen plantNone where clinical supply depends on itEssential supply covering compressor, dryer and controls together, not the compressor alone

What gets missed most often is the auxiliary that shares a duty with the main machine. Compressor on the generator, dryer on the mains. Extract fan on the generator, supply air on the mains. Instrument on the UPS, its chiller on raw mains. Split supplies of that kind create failures that look random and are entirely predictable once you see the single line diagram.

Stabiliser, UPS and generator solve different problems

Earthing is the part that gets skipped

Earthing is invisible, it is finished before the equipment arrives, and it is the first thing a service engineer measures when an instrument behaves strangely. Manufacturers publish a maximum neutral to earth voltage and a maximum earth electrode resistance in their site preparation documents. Obtain that document at quotation stage and hand it to the electrical contractor before the civil works, because retrofitting an earthing system through a finished building is expensive and disruptive.

The most common defect we find is a well-intentioned one: a separate rod driven outside the laboratory for a dedicated 'instrument earth', not bonded to the building's main earth. That arrangement creates a potential difference between the instrument chassis and the building steel, cable trays and pipework around it. It is a safety hazard and it injects exactly the noise it was installed to remove. Where a clean earth is genuinely required, run it as a separate conductor back to the main earth bar, so that the whole installation remains one bonded system with one reference.

Measure and record earth electrode resistance at handover and annually, in the dry season as well as the wet, because soil resistivity moves a long way between them. Record neutral to earth voltage at the instrument socket under load, not at the distribution board with the laboratory switched off. Both belong in the handover file, and both are the evidence you will want if a warranty conversation starts.

Sizing without guessing

  1. Log the supply before you buy anything. Leave a power quality recorder on the laboratory board for at least a week, covering the evening peak and a weekend, capturing voltage extremes, interruption count and duration, and waveform distortion. Every decision below depends on this data. Without it you are buying a product rather than solving a problem.
  2. Build a load schedule with nameplate rating, measured running load, starting inrush, power factor, and whether the load is linear or not. Nameplate alone will oversize everything, and oversized equipment runs inefficiently and costs more to own.
  3. Split the schedule into essential and non-essential. Only the essential goes on the generator. This is where generator capital cost is won, and what keeps the machine off a very light load.
  4. Size UPS autonomy on the actual requirement: the time for the generator to start and accept load, plus the time for a controlled shutdown of anything that needs one, plus a margin for the day the generator does not start. Choose the number for a reason, not because it is a round figure.
  5. Ask both vendors, in writing, whether the generator can accept the UPS. The UPS rectifier and the step loads have to be compatible with the alternator, and the two suppliers must agree on it before purchase rather than during commissioning.
  6. Specify and then physically test the changeover scheme. Witness a mains failure test at handover with the laboratory loaded, watch what recovers automatically and what does not, and write down what has to be restarted by hand.
  7. Leave the recorder installed. A permanent power quality monitor on the laboratory board gives you a warranty case when something fails, and tells you when the supply has changed because something new was connected upstream.

Put it in the budget line, not the variation order

The pattern is familiar to anyone who has run an e-GP tender for laboratory equipment. The bill of quantities lists the instrument, installation, training and a warranty, and says nothing about the supply. The instrument arrives, the site is not ready, and the power protection becomes a variation for which no budget code exists. The instrument then runs on raw mains for a year while the paperwork is argued about, and the damage is done quietly before the protection is ever installed.

Back to all Insights