Every HEPA leak scan passes or fails against 0.01% of a number that many test reports never print: the upstream concentration.
With a photometer, the aerosol measured upstream of the filter is taken as 100%, and a leak is any steady downstream reading above 0.01% of it. ISO 14644-3 writes the acceptance as that ratio, so a pass is worth only as much as the upstream measurement behind it.
The photometer method works with an upstream challenge of roughly 10 to 100 µg per litre of air. At 20 µg/L the leak threshold is 0.002 µg/L downstream. If the generator runs low, the threshold falls with it: every tenfold drop in the challenge puts the threshold ten times closer to the instrument's zero. A pinhole in the media or a lifted gasket still leaks, but its plume gets hard to tell from background, and the filter is signed off.
We see thin challenges most on large filter banks, where one generator is asked to feed too many filters, and on duct-mounted housings where the upstream sample point sits in air that has not mixed yet.
On a report we want the measured upstream figure for each filter or bank rather than the generator setting, with the point where it was sampled and the photometer's resolution written next to it.



