Over the last few months, me and a friend of mine have sifted several sets of grinds using the 400/800/1200/1600 sets of sieves. The procedure is to grind 10g, sieve for 2+ mins and then measure each part using 0.01g calibrated scale, thus finding the % of mass within each sieve set pair. No re-sieving is allowed. The results of sieving have been very repeatable, I've tested the same grind size several times and have found no more than 1-2% difference at most. There's also slight difference between dark and light roasts, but again not overly significant (again a few percentage points), but for consistency sake we stuck to typical speciality filter roast levels.

Most of the grinds are at roughly at filter grind sizes. So far we have samples from a big flat burr, some conical hand grinders and a blade grinder.

The graphs are linked below. You can click and double click on the legend to select/deselect different grinders.

http://cangencer.github.io/noncumulative.html
http://cangencer.github.io/cumulative.html

The raw data is available here:

https://github.com/cangencer/kruve-grin ... inders.csv

You can see that it's easy to show a grinder as less or more uniform based on the grind size chosen, and where the peak corresponds to among the selected sieves. The cumulative graph is more useful for finding "average" grind size (you can draw a horizontal line from 50% and find where it intersects with the line for that grind setting)

Given that hardly any objective data is available about any grinder, and people make expensive purchase decisions based on essentially hearsay, it seems like a good idea to start collecting and publishing some data. How this data correlates to taste / extraction yield is another matter entirely though, but all the grinders tested were able to reach the 18-22% EY range.

I would be interested to collect more data from people who have access to the same set of sieves. Unfortunately the XL sieves for Kruve seem like sold out (perhaps no longer produced?), which is a bit strange given those were the most realistic ones to use in typical filter grind sizes.

P.S: The EK43 S uses SSP burrs

Nice data: It's interesting how little difference the adjustments on the EK43 made.

You've picked a very coarse grind setting for these tests. The cupping protocol specifies 70% to 75% by weight should pass through an 800 micron screen. This is also the grind size most useful for French Press and all steeped methods. Presumably, filter grinds should be finer than that.

From what we've tested, the Kruve numbers aren't 1:1 with mesh sieves - though the Kruve is very repeatable within itself it doesn't match to mesh sieve numbers so you can't use one to translate to the other. I don't have a mesh sieve but the other tester in this case has a few of them. For example something that's 32% > 1200 in Kruve is only 14% >1200 using the same size mesh sieve. So the Kruve measurements seem coarser than mesh ones.

For cupping and immersions I typically use #8 or #9 on the EK43 S which is something like %55 < 800 according to Kruve, so I think it would be something like 70-75% when measured on an 800 mesh sieve.

Thanks for the clarification. I guess the hole pattern makes a difference on how many particles pass through.

It's interesting how the EK43 shifts at different settings, with less peaking and a fine and coarse skew at 8 and 14, and peaking more, with no skew, at 10 and 12. If all the talk about taste versus dispersion is right; you'd think devising a brew method that suits a setting at 10 or 12 would produce the best brew.

another_jim wrote: It's interesting how the EK43 shifts at different settings, with less peaking and a fine and coarse skew at 8 and 14, and peaking more, with no skew, at 10 and 12. If all the talk about taste versus dispersion is right; you'd think devising a brew method that suits a setting at 10 or 12 would produce the best brew.

Some look more uniform since the average happens to be right at the center of two sieve sizes. So it's better to compare grind settings with similar shape for example Lido@22 vs EK43@14 or Feld@2+6 vs EK43@12. This is also the reason a lot of grind distribution comparison graphs (i.e. from socratic) are not so meaningful. If you want like for like comparison you'd dial in each grinder so that you'd have the average at say 1000 (right between two sieves) and then do a comparison, this would likely be quite time consuming though.

What sequence did you use the sieves?
Was it 400 paired with 1600, then the 800 and 1200?

For EK, Kinu and Vario I used 400-1200 and 800-1600, and the rest of the grinders were done with 400-1600, 800-1200. Doing it 400-1600/800-1200 it's easier to calculate the intervals. I never re-sieve, new batch is ground for each individual measurement.

I've got a full set of Kruve sieves so will try the same tests as you on my grinders.Previously I have only used the Kruve to calibrate my grinders at filter and french press grinds and used 300/1600, 400/1400, 500/1200, 600/1100,700/1000,800/900. As you can imagine very time consuming to do!. I used 30g samples for this.
I've got an EK43, Compak E10, Mazzer Royal and a number of hand grinders, Pharos v2, Hausgrind, Feldgrind and Feld 47 so hopefully will add to your data set.

I've got a full set of Kruve sieves so will try the same tests as you on my grinders.Previously I have only used the Kruve to calibrate my grinders at filter and french press grinds and used 300/1600, 400/1400, 500/1200, 600/1100,700/1000,800/900. As you can imagine very time consuming to do!. I used 30g samples for this.
I've got an EK43, Compak E10, Mazzer Royal, Niche Zero and a number of hand grinders, Pharos v2, Hausgrind, Feldgrind and Feld 47 so hopefully will add to your data set.

That'll be great thanks! I'm happy to add them to the list as you gather the numbers. Yes it can be rather tedious and repetitive, esp the 400 tends to cling everywhere.