98mm Burr Extraction Yields

Grinders are one of the keys to exceptional espresso. Discuss them here.
Posts: 393
Joined: 9 years ago

#1: Post by coffeemmichael »

With so much discussion surrounding burr geometries, materials, coatings, and alignment, notably missing is data to correlate with flavor.

If you have extraction yield data, please share it here.

The discussions have gone on too long based on speculation and anecdotal reports.

Posts: 567
Joined: 9 years ago

#2: Post by cebseb »

I'm not really sure what you're looking for, but this is from my thread from a while back. Hope this helps.
cebseb wrote:My apologies for the delay. I've had the raw numbers for a bit now, but just haven't had the time to tabulate it in a more digestible format.

All the data shown is from one bean (Sweet Bloom: Kolla Bolcha) and all roasted on the same day. TDS readings were taken with a VST refractrometer that was calibrated hourly. I used my typical workflow with each grinder (palm puffs, shaker cup, Lyn weber blind tumbler, Force Tamper, etc).


I only took three samples with the Ek43s TITUS since it was quite apparent that my EKK43 was near identical in output. Not surprising since they were both aligned and also use the same type of SSP burrs.

The highlighted input numbers are to signify the samples not used in calculating the average retention of each grinder since the output ended up being a greater number. I assume this happens on occasion with most single dose grinders due to some compacted grinds deep inside the chamber being knocked loose.

From the data gathered above and the time I have spent with the grinders, I have come my own conclusions. However, before I get to those, I would like to invite anybody else to comment on the findings or bring up any questions that need to be addressed.

User avatar
Team HB
Posts: 13947
Joined: 19 years ago

#3: Post by another_jim »

Looking at the yields, the EKK ones have a lower standard deviation than the Flat and Max, but the sample is not large enough to declare it significant. The lower Flat average is not significant, since the flat's variance is high enough to wash it out unless you take a a sample of about 50 or so. The differences in TDS are also very small and not significant.

My sense is that if you had noticeable differences in taste from these shots, that could not possibly be due to the extraction and TDS. The data set would be very important if this is the case, since it would strongly contradict the common two factor thesis that changes in extraction yield and TDS are the only causal factors when espresso shots from the same coffee taste different.
Jim Schulman

Posts: 567
Joined: 9 years ago

#4: Post by cebseb replying to another_jim »

From my notes, the taste of the shots did not vary enough in taste to be distinguished from amongst each other. I let the shots cool somewhat then do the whole swish and spit thing. I had to have my wife take over tasting duty for a bit since I started feeling the effects of over-caffeination at about 12 shots in. My wife (the superior palate in the household) was also unable to notice any significant differences in taste in general.

Note that this test was done on the "pre-production" version of the MAX and did not have the most recent SSP burrs.

Supporter ♡
Posts: 2483
Joined: 13 years ago

#5: Post by samuellaw178 »

cebseb wrote:I'm not really sure what you're looking for, but this is from my thread from a while back. Hope this helps.
Thanks Seb for sharing. Very interesting data set. Just out of curiousity, are those measurement syringe filtered?

Here's a quick random data set from me. They're non-filtered readings so take them in that context.

Note that the trendline used above isn't the right one to describe the relationship between EY% and brew ratio. MS Excel just doesn't have the right fitting model for use.

To OP: From my own long dabbling with EY, I don't think extraction yield is that useful of an indicator for grinder's performance and definitely no substitute for taste test. One could almost say there's very limited correlation between taste and EY% (and will only correlate under certain/limited context).

Based on the numbers I see, I am inclined to think that the Robur/conical's claim-to-infame or the infamous 19% extraction ceiling for conical is probably due to misunderstanding. It's most likely due to barista's tendency to squeeze massive dose into triple baskets and pulling ristretto, and much less so due to alignment/grinder/burr size... Similarly, EK43 tends to get higher extraction yield because users tend to have trouble getting very tight/slow flowing shots. When pulled to short brew ratio (without any other crazy manipulation), the EY% on EK isn't that much different (about 1 % difference mostly) from a 'normal' grinder.

Extraction yield is affected by a myriad of factors (below). Change any of it and it can affect your extracton yield greatly. So most extraction yields number are mostly meaningless without being accompanied with its full context (a long list of variables in making the shot and the measurement protocol). When everything is controlled to be the same, the extraction yield doesn't really differ by more than 1% EY (+/- variance) as mentioned.
  • Contact time (all else equal, longer contact time can allow more soluble to dissolve)
  • Brew ratio (more water, aka solvent, will extract more solubles)
  • Evenness of extraction. ie. no channeling/bypassing effect (even extraction/no channeling means you have more water-coffee contact surface area available for extraction). When you have too much fines or uneven puck 'density', it can create mini dry spots/low flow region within the puck and hence lower extraction yield.
  • Temperature (dissolvable solids become more easily soluble at higher temperature)
  • Coffee (the input material, aka coffee, dictates what & how much is available for extraction. Roast level usually determines how easily soluble the solids are. eg. green coffee isn't very soluble
  • and of course the particle size/grind size (finer grind = more surface area, but it is also easily trumped by other factors above)
★★ Quite Helpful

Posts: 567
Joined: 9 years ago

#6: Post by cebseb »

samuellaw178 wrote:Thanks Seb for sharing. Very interesting data set. Just out of curiousity, are those measurement syringe filtered?
Yep! It was an expensive day. And my pleasure.