PSD changes when settings change

Changing grinder settings isn't simply shifting one average number. The relative position of the grinding mechanism can change the entire PSD, in which the fine fraction, the coarse fraction, and distribution width can all shift together. So if only a representative value is checked, an operator may miss changes in the tail. For extraction control, this is a significant gap since a small or large group of particles can behave differently from the rest during brewing.

Why the same setting can still differ

The same setting also doesn't guarantee the same PSD at every point in time. Temperature, residual grounds in the machine, grinding-mechanism condition, and operating conditions can all change the output. Research on static electricity in grinding shows charge state also depends on coffee characteristics and can affect how grounds distribute after leaving the grinding mechanism. So when comparing before and after a settings change, you need to control variables that aren't on the adjustment dial too.

Testing before and after

A before-after test should include a stabilization stage. Right after changing settings, you shouldn't take just one sample and conclude immediately. You need to grind several consecutive samples of a defined mass, discard or record the transition portion according to a consistent procedure, then measure PSD. At the same time, cleaning and machine condition need to stay stable. Otherwise, the observed change could be due to old grounds being released or the grinding chamber still transitioning between states.

Determining a stable state

A setting can only be called stable once consecutive measurements converge within an acceptable range of variation. “Stable” doesn't mean every sample is identical; there's always measurement variation and material variation. What you're looking for is fluctuation around a relatively stable level rather than a continuously increasing or decreasing trend. Series-format data makes this clearer than a single average number.

Reading the entire distribution

When reading PSD, you should look at the median along with the width and the tail. A 2024 study on fines in espresso shows the very small particle fraction plays a significant role in extraction kinetics. So two samples with the same median but different fines ratios don't necessarily produce the same behavior in the cup. Recent PSD models also emphasize using the distribution rather than a single indicator to describe grind consistency.

A good procedure after each settings change therefore includes three layers of data: machine condition, PSD data, and, if the goal is brewing, an output indicator of the extraction process. Not every machine adjustment needs to become a major experiment, but enough data is needed to know which change is real. At that point, “uniformity” stops being a feeling from looking at the grounds and becomes an attribute that can be tracked, compared, and traced.

After changing settings, the number of transition batches needs to be treated as part of the procedure. Old grounds in the grinding chamber can make the first sample different from later samples. If this phenomenon is ignored, an operator might think the setting hasn't stabilized when in fact the machine is simply transitioning from the old state to the new one. So the number of transition batches should be defined or measured separately when establishing the method.

A before-after PSD chart should also show the variation between repeats. If two settings have similar medians but one has a wider distribution, calling them “the same” is missing information. In quality control, distribution stability can sometimes matter more than hitting an exact average value. This is why raw data, or at least indicators sufficient to describe the whole distribution, should be stored.

A point often overlooked is the time between adjustment and measurement. The machine can reach a different thermal state after a few batches, and the amount of residual grounds also changes. So “10 seconds after adjusting” isn't equivalent to “1 kg after adjusting.” If the operating procedure is defined by transition mass, reproducibility will be higher when production speed changes.

When the goal is uniformity, the sampling method should stay the same from start to finish. If one sample is taken from the entire stream and another from the middle of the stream, PSD can differ even without the machine changing. Using the same sampling rule helps ensure observed variation reflects the equipment and raw material more than the sampling action itself.

After each settings change, recording the number of transition batches helps distinguish a temporary state from a stable one. If the first sample differs but later samples converge, that could be the effect of leftover grounds in the machine. If several consecutive samples keep shifting, you need to further check the grinding mechanism and raw-material conditions. This procedure helps base adjustments on a trend rather than a single unusual sample.

When determining stability, you should also predefine what condition requires repeating the measurement. If one sample has an unusual tail, it should be rechecked rather than immediately changing the setting. This rule helps quality control avoid overreacting to a single data point. Only once a trend is confirmed should the machine be adjusted or the raw material investigated.

References

• Khongphinitbunjong et al. (2025), Journal of Food Process Engineering 48, e70129, DOI 10.1111/jfpe.70129.
• Smrke, Eiermann & Yeretzian (2024), The role of fines in espresso extraction dynamics, Scientific Reports 14, 5612, DOI 10.1038/s41598-024-55831-x.
• Méndez Harper et al. (2024), iScience 27, 110639, DOI 10.1016/j.isci.2024.110639.