henri wrote:What I don't understand so well is how one goes from this information to median particle size, as reported in the diagram.

I'm guessing a little to get there, at the moment. I've got a pretty good idea of what the burr spacing is, but the "outfalls" on the burrs--the grooves in the outer edge--make a big difference in the size of the particles which escape the burrs. I suppose I could measure the outfall depth with reasonable accuracy using calipers if I removed the burrs...

As you point out, presumably that would give me maximum particle size, not median. However... Particle size distribution is usually bimodal, with fines on the left and the nominal particle size on the right:



The fines peak is usually fit using a log-normal distribution, and the nominal peak is usually fit with a Weibull distribution.

As a first approximation, I think we can say that the burrs won't let a particle through which is larger than the outfall depth plus burr spacing. This isn't quite true, because the particles can be larger in the dimension parallel with the burrs--i.e., a long, thin particle could roll through the gap with its long dimension parallel to the burrs.

So with the Weibull distribution I think what we're seeing is a peak at outfall depth plus burr spacing, and then some allowance for larger particles rolling through sideways.

There's definitely a bit of conjecture on my part here. I'd love to hear any differing points of view!

henri wrote:In practice, I find that the Eureka dial has "reproducibility" up to about a 1/4 of a step (it's very difficult to make smaller adjustments, let alone see what one is doing). That would be 12.5 µm, which I believe is what the clicks on some of the 1Zpresso grinders correspond to, for example.

Have you tried modifying the dial at all? I built this indicator for the Specialita:

https://www.thingiverse.com/thing:4941720

I always approach the setting from the bottom side, to avoid issues with backlash in the adjustment screw. I would say with this setup I get repeatability to 0.1 steps for sure, and maybe even 0.05.