Dosing and Solids Extraction: A Retraction

Beginner and pro baristas share tips and tricks.
User avatar
another_jim
Team HB

Postby another_jim » Aug 22, 2007, 3:56 pm

In February 2007, I posted and talked up a paper "Some Aspects Of Espresso Extraction" where I published a set of experimental data which apparently showed that higher doses in a given basket led to lower solids extraction. I wish to retract almost all of that claim. The results were not due to the physics of percolation, but to the inadequate preparation of the higher dosed samples. I will explain this below.

I also claimed that the taste of espresso systematically changed with the degree of solids extraction. This claim stands.

Last year, using the Mazzer Mini and the Versalab M3 grinder, I found that higher doses in the same basket led to a very precise decline in solids yield. Andy replicated the result with the Mini. When I started the Titan Grinder Project tests, I again replicated the result with the Mini and the M3. However, this relationship was not apparent with the conical grinders. The conicals had roughly the same solids extraction at all doses.

When I did the "Beat the Robur" part of the test, I used 13.5 and 16.5 gram doses. In these tests there was no systematic difference in solids extraction with any of the twelve grinders; in fact, the solids extraction was always so close to the 22.5% stated in Illy, that the differences could have been due to measurement error. So the relation between dose to solids extraction I had posited in the paper was clearly wrong.

I busted my head trying to explain why I had got that relation in the earlier tests, and not in the "Beat the Robur" trials, but didn't see a good reason. Fortunately Andy Schecter did, and the explanation is obvious:
  • Everybody noticed how well the conical grinders distributed the grinds and how evenly they poured. I never got a change in solids extraction with these grinders
  • I got the reduced solids extraction at higher doses in the early tests of the flat burr grinders, when I used simple distribution and packing. For the "Beat the Robur" test, I used WDT along with shaking and several other leveling techniques, and I did not get changes in solids extraction.
  • Therefore, the lowered solids extraction is due to sub-optimal packing when using higher doses on flat burr grinders.


This result may be even more significant to barista technique than the earlier one. I am not that great a barista, but I am competent. The naked pours I get with higher doses from a flat burr grinder are not picture perfect, but they do not look like they are channeling. However, they are not extracting properly, and are probably "pseudo-channeling" internally I'm coining this name for what happens in a puck with no cracks, but with denser and less dense zones of grinds. The denser pockets get less water flow because the water preferentially moves through the less dense patches. In a low dose puck, where one gets the soupy puck at the end, the grinds absorb so much water that these density differences even out. In a higher dose puck, where one gets a firm puck at the end, density differences probably remain till the end of the pour.

I have tried some of the lighter roasted coffees used in high doses at barista competitions. With flat burr grinders I couldn't get a decent shot at anything but a lowered dose. Presumably this means that competing baristas can properly distribute at higher doses than I can, without using time consuming grind sifting and fluffing techniques. This is hardly a surprise. But it probably does indicate that almost all home baristas, even those getting very pretty bottomless pours, should either be using something like WDT or a commercial conical when using high doses of coffee which produce a dry, firm puck after the shot.

(crossposted to HB and Coffeed, the sites where I made the original claims)
Jim Schulman

Wescott

Postby Wescott » Aug 22, 2007, 5:11 pm

Thanks for keeping us current on this, Jim.

Let me make sure that I understand properly. If I do, then all other things being equal (which, of course, they never are), one should either...

1) ...Use a conical grinder (easiest but costliest)

2) ...Aggressively de-clump the output of your flat blade grinder (a small nuisance, but doable in a number of different ways)

3) ...Use smaller doses to saturate and swell the unevenly distributed coffee so that the distribution matters less before the extraction proper begins (either "natural" or positive preinfusion will serve this purpose if you leave enough head room)

You could choose either the first or some combination of the second and third to achieve the best soluble extraction.

Please correct me if I'm wrong.

User avatar
another_jim
Team HB

Postby another_jim » Aug 22, 2007, 5:48 pm

Wescott wrote:Let me make sure that I understand properly. ...



That's the gist of it; thanks for the excellent summary.
Jim Schulman

User avatar
Teme

Postby Teme » Aug 24, 2007, 1:20 pm

So do I understand correctly that you find that the solubles yield is not dependant on the grinder or the dose size used? I.e. assuming that the extraction time (A), brew pressure (B) and brew temp (C) are the same, the solubles yield also remains a constant? If this is true, and assuming perfect packing of the coffee, is under and over-extraction then solely defined by the balance of the fast and slowly dissolving flavours that are extracted and end up in the cup (which sounds logical to me)? The dose size would certainly play a role here (with A, B and C remaining constant), but can this balance be measured by means other than tasting?

As for the grinder's effect on the extraction quality when updosing, I think one cannot generalise that one doesn't get sub-optimal extractions with conical burr grinders. I agree that conical burr grinders clearly do appear to be more "forgiving", but in my experience there are also differences between conical burr grinders in terms of the fluffiness of the grind and clumping - enough so to make differences to the extraction characteristics. For example:

* Casadio Instantaneo - Fluffiest grind of all these grinders and no clumping. Easiest grinder for achieving even distribution and extraction, even with a higher dose and withstands aging of the roast surprisingly well.
* Compak K10 - Grind slightly less fluffy than the Casadio. No clumping. Good extraction very easy to achieve with fresh coffee but becomes more picky when the roast ages.
* Mahlkonig K30 - Surprisingly fluffy grind for a flat burr, but clumping can be pretty bad in some conditions. However, the clumps are "soft" and it is fairly easy to achieve good distribution - especially for the flat burr grinder.
* Mazzer MiniE - Grind is the least fluffy of all these grinders and the clumps are harder. Difficult to reach an even distribution / extraction with a higher dose.

The fact that I think that the Casadio produces practically no clumping and the best initial distribution of all of the above grinders (quite a feat for a doserless grinder), I actually like the taste from it the least (with the possible exception of the MiniE if one gets the extraction right with it) even though it appears to be the easiest to achieve an even extraction with. This leads me to believe that there is something to the grind particle size distribution profiles and the shape of the ground particles generated by various grinders that does play a significant role in the optimal extraction as well. Perhaps a measure of the total surface area available in a particular size of a dose (this would vary between grinders)?

One thing that might also be interesting to explore when testing grinders is to measure the amount of static electricity in the resulting grind. I assume that this has a direct correlation to the amount of clumping and the "forgiveness factor" of the grinder. It also may play a role in whether or not some of the fines produced by the grinder end up in the puck (maybe some are retained in the grinder chute/doser therefore forcing the barista to grind finer or dose more than would otherwise be necessary to achieve a good extraction).

Sorry for all the questions and I may be completely off here, but I though I'd throw it in the air anyway...

Br,
Teme

User avatar
peacecup

Postby peacecup » Aug 24, 2007, 3:07 pm

another_jim wrote:When I did the "Beat the Robur" part of the test, I used 13.5 and 16.5 gram doses. In these tests there was no systematic difference in solids extraction with any of the twelve grinders; in fact, the solids extraction was always so close to the 22.5% stated in Illy, that the differences could have been due to measurement error. So the relation between dose to solids extraction I had posited in the paper was clearly wrong.



So, a 13.5 g dose = 3.0375 g solids, and a 16.5g dose = 3.7125 g of solids correct? If the later contains ~18.18% more water, that the tastes ought to be approximately equal (given constant brew temp and pressure)

Doesn't this mean that the issue of up- and down-dosing is a non-issue, if one has a good grind and proper distribution?


PC
LMWDP #049
Hand-ground, hand-pulled: "hands down.."

User avatar
another_jim
Team HB

Postby another_jim » Aug 24, 2007, 10:25 pm

I don't think extraction is the only issue when it comes to taste. For instance, the little Rocky extracted equally at 16.5 and 13.5 grams, tasted great at 13.5 and miserable at 16.5. On the other hand, take a washed yrg, do an 18 gram shot on a non-great grinder without fancy distribution, and it's lemon, lemon peel and tannin, at 14 grams it'll be flower honey. This I still think is underextraction, plain and simple.

The concept of pseudo-channeling, i.e. a perfect looking pour with a puck that has under extracted pockets, is a hypothesis at this point. It explains the data so far, but needs to be proved. If anyone has any ideas how to do that, I'm all ears.

I was away 4 days, when I came back, I had more trouble with my now ten day old test blend using the Compak, so Teme, I think you're right about that. This is the first time I had to work getting the grind setting exactly right.

Does the Instanteneo use the Trespade burr (like the Innova, Lux etc)? The size is the same.
Jim Schulman

User avatar
peacecup

Postby peacecup » Aug 24, 2007, 11:01 pm

Holmes sat for some time in silent thought. .... "That is another line of thought. There are two, and I beg you will not tangle them. ...'

I think this is true of taste and extraction.

It seems that the physical properties of extraction could be addressed using fluid dynamics modeling. There must be some theory relating to how liquids flow through interstital spaces.

As to taste, well that's another matter. Consider all the variables that must influence the human palatte - recent food or drink, environmental odors, mood, etc. I saw a scientific presentation on the role of the sense of smell a taste a while back. Its very complex. Lots of science out there for the inquring mind...

PC
LMWDP #049

Hand-ground, hand-pulled: "hands down.."

User avatar
Teme

Postby Teme » Aug 25, 2007, 4:28 am

another_jim wrote:Does the Instanteneo use the Trespade burr (like the Innova, Lux etc)? The size is the same.


Nope, it is their proprietary burrset as far as I know and a lot sturdier looking than the Trespade set. It looks very much like a smaller version of the Compak's burrs. The burr carriers are beefy and the grinder has a 450W motor, which is a fair amount considering the burr size.

Br,
Teme