Each EC fan's EMC filter leaks current to earth in normal running, and a fan wall adds it up. At an assumed 3.5 mA per fan, eight fans put 28 mA on earth, while a 30 mA RCD may trip from 15 mA. Split circuits, choose the right RCD type and size the earth.
At −60 °C dew point air holds about 0.007 g of water per kilogram, roughly 3,000 times less than outdoor air in southern Viet Nam. By a rough sum, 1 m³/h of leakage uses up around 4,000 m³/h of dry supply air, so the envelope, airlocks and occupancy belong in the dryer specification.
AHU casing classes such as L1 leakage or TB2 thermal bridging usually come from a standard EN 1886 model box. The unit on site has more doors, joints and penetrations. At FAT, check them on the real unit, especially filter frame gaskets: bypass around the fine-filter frames loads the terminal HEPA filters faster.
A filter that passed its factory scan can still fail the installed leak test at the frame. Gel-channel grids seal without clamp force and allow quick changes; gasket grids depend on even clamping and take a compression set. Gel usually suits Grade A and B; gaskets can suit Grade D.
Each cleanroom luminaire breaks the ceiling's pressure boundary, and whether it is serviced from the room or the void drives most of the specification. Ask for a flush, gasketed face and chemical resistance data for your disinfectants, leave the ceiling to the filters in unidirectional airflow, and set illuminance from the task with EN 12464-1.
Airlock door interlocks protect a 10–15 Pa step between grades; a fire needs the doors to open. Specify both together: interlock logic, status lamps, fire-alarm release, fail-safe locks, a logged BMS alarm on every override, and a release tested at FAT, at OQ and periodically.
A ±5% airflow tolerance cannot be tested until the specification says where it is measured, with which instrument and at which filter condition. Fan inlet, duct traverse and terminal hood readings can differ by more than the tolerance. Write the method in the same sentence as the number.
A 50 mm PU-core AHU panel can match a 60 mm mineral wool panel on heat loss, but PU is combustible and mineral wool is not. In a humid plant room, condensation starts at the frame's thermal bridges whatever the core. Decide fire class first, then the EN 1886 TB class, then thickness.
Filter transmitters get bought twice, or not at all, when the tender never says who supplies them. An AHU/BMS interface schedule lists every sensor, drive signal, actuator and fire alarm contact with who supplies, mounts, wires and configures it, attached to both tenders before they go out.
The installed HEPA scan covers media, frame and the filter-to-housing seal, but not the flange between housing and ceiling. Where the ceiling void runs at higher pressure than the room, unfiltered air enters there. Seal it with an evenly compressed gasket and sealant bead, then probe the flange and file the result.
The pressure drop at which a supply AHU's final filters are changed should be the dirty-filter allowance the fan was selected for at design stage, checked at FAT and quoted in the maintenance task. Raise it later and room pressure drifts down inside its alarm band, visible only on the trend.
When a terminal label disagrees with the drawing, fault-finding turns into guessing. At FAT, trace ten random wires end to end, read device tags against the drawing and bill of materials, and confirm the door-pocket drawing carries the same revision as the equipment file. After site changes, return marked-up drawings.