As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.
Flow rate is the divisor. An optical particle counter detects individual particles and reports a concentration, and the only bridge between those two things is how much air went past the laser while it was counting. Get the flow wrong by ten percent and every concentration the instrument reports is wrong by ten percent, in a direction you cannot see from the display. Get the flow unstable and two runs in identical air will not match, which destroys the one thing a particle count is actually good for: comparing a place against itself.
A count is meaningless until you say how much air it came from
Suppose an instrument registers four thousand pulses in the half micron channel. That number on its own carries no information. Four thousand pulses from half a litre of air describes a dirty room. The same four thousand pulses drawn from thirty litres describes a fairly clean one. The pulses are the observation, the litres are the context, and a concentration is simply the first divided by the second.
Instruments handle the divisor in one of two ways. The honest approach is to meter the flow, either with a dedicated sensor or by running a positive displacement pump whose output per revolution is fixed by geometry. The common approach is to assume the rated flow, run a timer, and multiply. Assumption works perfectly while the hardware matches the datasheet and fails silently the moment it does not, because nothing on the display changes when the assumption stops holding.
This is why the flow figure deserves to be read as a specification rather than skimmed as trivia. Two of the instruments here publish the same one. The six channel air particle meter states a built in pump drawing air at 2.83 litres per minute, described as a tenth of a cubic foot per minute, alongside laser scattering detection. The cleanroom and HVAC particle counter states the same 2.83 litres per minute and calls the flow stable. Neither states a tolerance on that figure, so how far the real flow may sit from the nominal one is not stated in the listing bullets.
Pumped sampling against fan draw and passive diffusion
There are three ways to get air in front of a laser, and they differ in how well they define the divisor.
- A positive displacement pump sweeps a fixed volume on every cycle, so the flow is set by mechanical geometry and motor speed rather than by what is in the way. Back pressure from a dusty filter slows it, but the relationship is predictable and the flow can be verified externally. This is the arrangement behind a stated figure such as 2.83 litres per minute.
- A fan moves whatever the pressure difference allows. Flow depends on how clean the path is, how the case is oriented, whether a hand or a pocket is covering an intake, and how the surrounding air is already moving. Fan drawn sensors are cheap, quiet and long lived, and they rarely publish a volumetric flow because the honest answer would be a range.
- Diffusion relies on particles wandering into the chamber on their own. There is no defined sampled volume at all, response to a change in the room takes minutes rather than seconds, and the coarse channels are badly under represented because large particles diffuse slowly and settle quickly.
The Eujgoov dust particle counter sits in the first group, stating a built in micro vacuum pump and describing the response as fast, although the volumetric figure itself is not stated in the listing bullets. The presence of a pump is the architectural fact that matters: it tells you the instrument is trying to define its own divisor rather than hoping the room will supply one.
What a drifting 2.83 litres per minute does to the number on screen
Work the arithmetic in the direction the hardware actually fails. The instrument counts particles for sixty seconds and divides by a nominal 2.83 litres. Air genuinely holding one thousand particles per litre should deliver 2,830 counts. Now let the real flow fall to 2.55 litres per minute, a reduction of a tenth, because the inlet screen has loaded with the plaster dust you were sampling last week. Only 2,550 particles enter the chamber, but the firmware still divides by 2.83 litres and reports 901 particles per litre.
The room did not change. The instrument now understates it by a tenth, and it will keep understating it by a growing margin as the screen loads further. Nothing in the display hints at this, because a counter has no way to distinguish a particle that was never there from one that never arrived. The same mechanism works in reverse on a freshly serviced pump that runs slightly fast, inflating every reading by a few percent.
A second consequence is subtler. Sampling efficiency for the coarse channels depends on inlet velocity. Large particles have enough inertia to miss a bend in the flow path and deposit on the wall rather than crossing the beam, and that loss grows as velocity falls. So a slow pump does not scale every channel down by the same fraction: it hits the ten micron channel hardest and the sub micron channels least, quietly changing the shape of the size distribution as well as its height.
Sample volume, counting statistics and the shortest honest run
Even with a perfect pump there is a floor on repeatability, and it comes from the particles rather than the instrument. Arrivals at the laser are random events, so if the true concentration would deliver a certain number of counts, an individual run will land near that number rather than on it. The expected spread is the square root of the count, so the relative uncertainty is one divided by the square root of the number counted. Count one hundred particles and successive runs scatter by about a tenth. Count ten thousand and they scatter by about one percent.
| Run length | Air sampled at 2.83 litres per minute | Particles counted if the air holds 1,000 per litre | Spread between repeat runs from counting statistics alone |
|---|---|---|---|
| 10 seconds | 0.47 litres | About 470 | About 4.6 percent |
| 30 seconds | 1.42 litres | About 1,420 | About 2.7 percent |
| 1 minute | 2.83 litres | About 2,830 | About 1.9 percent |
| 5 minutes | 14.2 litres | About 14,200 | About 0.8 percent |
| 10 minutes | 28.3 litres | About 28,300 | About 0.6 percent |
Read the table twice. The first reading says that short runs are noisy, which is unsurprising. The second says that the noise depends on the number counted and not on the duration, so the same ten second run in air a hundred times cleaner counts about five particles and scatters by nearly half. In clean conditions, in the coarse channels, a short sample is close to worthless, and the instrument will still display a confident looking figure. That is the mechanism behind a great deal of apparent instrument disagreement, as set out in why particle sensors disagree.
Recording the sampling conditions alongside the count
A particle reading travels well only when four things travel with it. The flow rate the instrument used, because without it the figure cannot be converted into any other unit. The run length, because it sets the statistical floor above. The channel, stated explicitly as cumulative or differential, because a count of particles at or above half a micron and a count of particles between half and one micron are different quantities. And the location and conditions, because inlet height, proximity to a wall and the state of the ventilation all move the result more than most instrument differences do.
Build the habit into the routine rather than the notes. Run every comparison at the same duration and the same inlet height. Repeat each measurement at least once and keep both figures, since a pair that disagrees badly is telling you the run was too short. Check the inlet screen on a fixed schedule and verify flow against an external meter when the instrument matters to a decision. The practical trade offs between instrument types are laid out in how to compare laser particle sensors and the particle counter overview, while the air particle monitor overview covers the fixed and fan drawn end of the range where no flow figure exists to record.
Frequently Asked Questions
How long should a single sample run?
Long enough that the instrument has counted a few thousand particles in the channel you care about, which is a different duration in clean air than in dirty air. At a flow of 2.83 litres per minute a one minute run draws under three litres, and in genuinely clean conditions that can mean a handful of counts in the coarse channels. The practical rule is to extend the run until a repeat of the same run lands close to the first, rather than to pick a fixed duration and trust it everywhere.
Does a restricted inlet make the reading high or low?
Low, which is the dangerous direction. The firmware divides the counted particles by a nominal volume calculated from the rated flow and the elapsed time. If the real flow has fallen because of a blocked filter, a kinked tube or a tiring pump, fewer particles enter the chamber but the divisor stays the same, so the displayed concentration falls. A clogged instrument reports an improvement that never happened.
Can I compare a pumped counter against a fan drawn monitor?
Only loosely, and only for trends. A pump holds a defined volumetric flow and can state it, so its counts convert into a concentration with a known divisor. A fan assisted or diffusion fed sensor moves air at a rate that depends on orientation, obstruction and internal fouling, and usually publishes no flow figure at all. Both can show you that a room got worse, but only one can tell you by how much per litre.
What is a tenth of a cubic foot per minute in litres?
A cubic foot is 28.317 litres, so a tenth of a cubic foot per minute is 2.8317 litres per minute, which is where the familiar 2.83 figure comes from. The unit survives because contamination control grew up around counts per cubic foot, and a tenth of a cubic foot per minute makes a ten minute run equal to exactly one cubic foot of sampled air. If your instrument reports per litre and a reference document asks for per cubic foot, multiply the per litre figure by 28.317.
Does the flow rate need rechecking over time?
Yes, and it is the maintenance item most people skip. Pumps lose displacement as seals age, inlet screens load with the material you have been sampling, and a tube that has been coiled tight in a case develops a permanent restriction. Flow verification against an external meter is the only way to know, and it is the step that keeps counts taken far apart in time comparable with each other.