How to Taste Coffee



That can happen fairly easily. And uneven/not smooth heat/fan profile can cause this, and when making your own profiles it is nice to cut a couple beans with a good knife to get a clean cut and you can see the variation of the internal vs external bean color without having the color measuring tool.


An interesting article on grinding - very interesting about what they say about fines…

And how cooling coffee makes it undergo a “glass transition” (becomes brittle) at lower temperatures - which increases fines and decreases boulders when grinding - because the coffee shatters. And makes for more even grinds and better extraction.

It appears that the best results were achieved grinding coffee at -196C (beans stored in liquid nitrogen) but the majority of benefit was achieved by -19C (beans stored in normal freezer)

The conclusions about fines are that sifting does not remove (all) of them (as many just adhere to the larger particle). And also that fines are actually good (which is an old fashioned idea).

I always used to think that more fines = more surface area = more extraction (but too many = choking). So there may be an ideal balance which can be manipulated by grinder - and by sifting…

Conclusion about grinding - store single doses in the freezer and grind when cold - to produce the best and most consistent extraction.

So perhaps freezing my beans might help me to get a finer grind and better extraction :slight_smile:


yea it’s definitely interesting article and I feel more aligns with the research for the industrial application of coffee, we talked about it here a little bit, It was also nice to note that just regular freezer temps couldn’t get the distribution and required to go to the extreme so room temp was ok. Environmental temp (the one outside the roaster)


Yep … interesting article for sure. Though I dont think that counting particles is better than volumetric measures (though even volume is not the best … maybe surface area? but I dont think surface area is the best too … because what we actually are interested in I think …is availability for extraction? :slight_smile: )
And its because the exact same reason why those numbers seem so impressive … sure … if you have the same amount/weight of 1mm sized particles and 100 micrometer sized particles, the number of particles in each group will be drasticaly different. So WOW … we have sooo many more fines … But … what is the absolute maximum of extractable material there? I would say its same, its just more or less available for extraction. If you really had the time and persistance, you may I think extract almost everything from both groups … but it would be a bigger effort with the larger particles.

And - if it was really true that fines are good, then hario grinder would be the best, and not EK43 and simillar ones …
I think its really important to understand why would you need them, what do they bring into the cup … and thats why I started doing the narrow band espresso tests - because a) I have been told its not possible without the fines slowing down the shot b) I have been told fines are necessary for some taste atributes in the shot …and so on … So I tried, and I can say I can do the shot with really big amount of fines removed /if not all :smiley: / … and I have some idea about what they add. For me, they are not necessary there … I would allmost allways trade of clarity and definition over say body … when using really great coffee. But for those who need a lot of bass, lots of body and richness … for them maybe the fines are possitive (though I am still not absolutely sure I can not make extraction of the narrow band to high enough TDS that it would start being more rich and full then the one with fines)


The distribution graphs should be done as a set: Volume, Weight, Surface area, and particle count — and perhaps other things???

Another interesating graph would be “100 micron penetration volume” or “extractable coffee volume” across the range. The 100 micron being the water penetration level measured by Matt Perger in the earlier video above.

Talking about measures used on the graphs: the y-axis should always be a relative measure like % - so % of total weight, and % of total volume. This allows you to compare the graphs. If you use simple weight and volume measures on the y-axis then you cannot make a meaningful comparison — because can simply stretch the scale to make the graphs match or look totally diferent (sort of like fake news :slight_smile: )

I am guessing that the Volume & Weight graphs will be very “similar” as they basically measuring the “physical amount” of coffee across the range. This is basically what I was trying to achieve by sifting grounds. These graphs should be identical (depending on scale) if the coffee density is the same across all particle sizes (which seems likely).

The only reason that I could see for a difference is if the density of particles varies across the range - perhaps the fines could be lower density (softer = easier to crush) and larger particle more dense (harder = harder to crush). And with both type of graphs then you could look for differences like this which otherwise would be missed.

The other two graphs (particle count and surface area ) are very interesting as they are (in my mind anyway) seem to (get closer to) measuring “where the extraction actually happens”. And they should also share similar characteristics. The surface area will be proportional to the particle count but larger particles will have (a power of 2) more surface area (based on the fact that the surface of a sphere is proportional to the square of the radius Basically the shape should be the “same shape” but with higher peaks (power of 2 surface area ratios) at the different particle sizes (remember to use % on y-axis).

The extractable coffee volume graph would also be very interesting as it should be closest to actually “measuring where extraction happens” :slight_smile: and it would be very interesting to see if optimizing “extractable volume” actually produced the best coffee extraction (highest TDS if you want to measure). My feeling is that it probably would. For example it might show which grinder or which sifting regimen/profile would (potentially) produce the best extraction.


If you look at the graphs in the article it appears that normal freezer temps (-19C) do actually give “most” of the benefit (by moving coffee into the “glassed” state). But lower temperature just further increase the affect.

And of course it is easy enough to use things like dry ice or liquid nitrogen at home :wink:


The interesting thing is that if you think about the surface area of a particle (proportional to the square of the radius). Then larger particles actually will have more potential for extraction = equals fines are BAD. Or rather you should get more extraction (higher TDS) from larger particles with greater surface area.

If you look at the 100 micron penetration (for expresso) then using all 200 micron particles would in theory (if it didn’t choke) give the best coffee extraction. Also you get a very similar extraction from each particle so the flavour should be very consistent (this fits in with you getting a “cleaner” taste across a 200 micron range).

But if you only had then you have lots of empty spaces = maybe need some fines to fill. BUT then you wil get “dirtier” taste (or less consistent extraction from each particle size).

The next thing would be to try a 300 micron particle extraction vs 200, vs 100. Do they all produce similar clean profiles? Or so they start to bring out different tastes from the (same) coffee? Obviously a real world test would be more like using different 200 micron ranges - and comparing taste…

If you do find that different ranges get different flavours from the (same) coffee - then you can tailor the taste more. Maybe used a mix of sizes for more body. Or a small range for clarity (@pavel as your experiments indicate). Then you might vary that range to emphasize different flavours (this is purely speculative - and needs more experimentation).

Trying to draw a conclusion from this probably means removing boulders and fines (say below 200ish) should increase TDS and give a better extraction. However you will have more interstitial space between the coffee particles (previously occupied by fines) - so water will flow through the coffee faster = you will need to increase the amount of coffee to compensate (reduce flow rate).


Yes … well … I think weight is better than volume, because weight is really measuring the matter of coffee, where volume also ads air, and “spaciness” (not sure what the english word for it is, but I mean the amount of air between particles) may be different for different fractions, and also you may have if fluffy or compact it in a way that tamping does, and have different volumes … so … I like measuring weight.


Yes I came to the same conclusion … :slight_smile: though its not real to not choke it … but … :smiley:


Apologies - I think I was not clear earlier :blush:

CLARIFICATION of what (I assume) volume means in this case. As the particle sizes are being measured using an LSA (laser diffraction) device I ASSUME that the volume in this case does not include the space between grinds. If that is true then it should be directly proportional to weight (assuming constant average density of the coffee).

This is how it is measured in all the continuous graphs like the Mahlköenig graph (using LSA)

If however we are doing our (comparatively) crude sifting - then weight is the only way to measure :slight_smile:


Unless we invent a new method of espresso extraction — pressure followed by vacuum???

Also we should add sonification (ultrasonic homogenization) to break down cell walls to get better (more thorough) and faster extraction.

And then we get a much higher TDS - but then we could find that it tastes worse - in the end the taste rules…


Nope … actually for the same weight you will get more surface area with smaller particles … thats why they overextract I guess? :smiley:

My reasoning for the narrow band espresso is exactly whan you mentioned that similar size of particles would give a chance for similar extraction and thus more straight and defined (less blurred by variations in extraction) shot I get. And from my experience it works this way, though not sure if its worth it for any beans …

I would say … since you have the sieves just go and try both 200micron bandwith and 100 micron one (you can then do multiple extrctions for each 100micron band that could still be extracted in reasonable way/with reasonable amount of coffee. What I think (not tested so just how I envision what happens) will happen is that if you can tweak the other variables to get the same ratio at the same time (or even better the same TDS? :slight_smile: ) you will get very simillar taste profile …

The idea of controlling the range of particles did come to my mind too … though for espresso I always want it as straight and clear as possible … so I would do this for pour over rather than espresso :slight_smile:

And … yes when you cut out fines you get faster shots but there is more to it … what you actually can do (by changing where you have positioned the band moving it to finer when you want slower shot, and then when finetuning by manipulating the dose) is pushing the extraction way further without problems with overextraction … thats why EK is so popular too I believe … because you can extract more. So the most extreme what could be done with Kruve is 100 micron or even 50 micron band extracted to very high TDS. I am waiting to play more with this untill I have refractometer though, because its really lot of work and hard to understand sometimes without some insight that TDS could give …


Also elsewhere (I think in the video above) Matt Perger mentioned getting a higher extraction espresso and better tasting espresso from a roller grinder.

Roller grinders tend to give more consistent grind size - so this further tends to confirm that reducing fines and boulder actually improves flavour.

It is from REAL observations (though this is simply one anecdote unfortunately) like this that we can work backwards - and figure out WHY it works better :smile:


You are correct!

My reasoning was flawed - I forgot that the weight of the particles increases with the cube (volume) of the particle. This counteracts the square of the surface area.

Which means that the area is directly proportional to the grind size = fines are better :slight_smile: or perhaps not totally the enemy…

But that then takes us back to 200 micron being (theoretically) optimal because of 100 micron penetration…

And there are a bunch of other reasons that different particle sizes will extract differently - so keeping a uniform grind size still seems good for consistent extraction.

I have a more musing (later) on overextraction of fines, migration of fines etc… I always like ristretto-ish shots (very unfashionable today) as I found they did not bring in nasty (what I called) overextracted flavours (from fines?). However it may well have been from underdeveloped beans (as in center is not roasted enough = underdeveloped).


I think this is allready done for immersion extractions, but … I believe it would disrupt the puck … so, not sure if it would be usable for espresso. There is also even more with ultrasonics added to help extraction, if I understand it well you can extract more at lower temperatures, which could reduce the impact hight temperature has on the outcome and still keep high extraction yield, maybe opening completely different taste profiles in espresso.


You will see after you try it … :slight_smile:
But though I understand how one comes to a conclusion that 200 micron size should be best (you should use all the matter and extract the most of it …) its actually not so obvious if different size is worse (well it is for a reason that you would not extract some parts at all, wasting is bad …) … I mean … even if you can not get the water all the way through if you have to use say 400 micron sized particles, you will still get even extraction so it should taste similar … and if you count the volumes (simplifying the particles to spheres - baaad - but simple :slight_smile: ) you will see the volume of the 100 micron deep hollow sphere is way way bigger than its supposedly unreachable core. So I think - most even sizes achievable should be good - and then tune how big they are to get the flow needed …you will not use a part of weight of coffee maybe … but still you should have very even extraction … right?


Agreed :wink:

Actually my theoretical conclusion would be anything above 200 should extract consistently (if the 100 micron penetration is true).

I was oversimplifying…


Roller grinder sounds interesting will have to look at it. For quite a long time I am playing with this idea that what we use to grind, though fine tuned and refined, is in a way something like a harddrive vs SSD - harddrive being a very old idea of plate with grooves (gramophone …) used for really long time just improving on it. SSD is completely different idea that makes a very big step forward in read write speeds, so I think a lot about if a different kind of grinding could not be helpfull. But so far nothing simple enought to be able to try it came out of this thinking - BUT

BUT - you gave me one idea that could be possibly tested now - though it may seem somewhat strange. What we could do actually … is grind and sift to make all particles very small … say under 200 micron. Than we may possibly try to add some neutral particles that would not extract at all, but just help to extract those that we have there. So the resistance might be tuned but the size of those particles and their amount. Its still not practical at all … lots of steps and some added material that would have to be totally inert and not adding any taste to the result … but it should be able to test if extraction of the fines alone would be the best :))))

And one older idea was to use some automatic system that would grind and sieve at the same time allowing particles smaller than the sieve size to be removed as soon as they are produced … some sort of drum and balls stuff which may be similar to roll grinding? But its just totally not practical … so maybe we have to make some nanobots that would slice the beans to perfect cubes of precise dimensions … easy peasy :smiley:


Rather large commercial spice grinder.

You will probably need to build a new room to put it in… And it probably costs as much as the house…

Cost (and size) is basically the reason we are not using roller mills for coffee shops :smile:


On the contrary = fines + inert matrix - makes perfect sense.

Though actually finding/making a suitable matrix to hold the “fines” may not be quite so easy… The issue being that as the fines erode they will tend to start migrating through the matrix. ‘So the inert matrix probably needs to expand gradually to counteract this (which I think happens in a coffee matrix with the larger coffee particles swelling more than small ones… just a theory).

This is similar to the “old” theory that multiple size grounds makes a matrix to hold the smaller sizes. Of course the coffee matrix is not inert.