Cracking into the Extraction: when and where the espresso puck brews. - Page 2

[jmatt]

What if you dried the pucks on their side? Would that possibly eliminate the top to bottom potential movement of particulate in the drying stage? Or maybe you did this and I missed it: What about drying some pucks upside down, then comparing to pucks dried rightside up? The difference may indicate how much of your results are due to migration after the extraction is completed rather than during the extraction?



I'm not being critical, just making suggestions. All of your measurements and instrumentation are way beyond me. Your thoughts are very intrigueing.

[timo888]

Fines is what got me started on this.

There's an old alt.coffee exercise of using three cups for a shot, putting the first few seconds in the first, the next few in the second, and the final part in the third. Unsurprisingly, the third section tastes weak, and slightly bitter. It's the first two sections that are puzzling: the first section tastes very intense, both bitter and sour; while the second section tastes sweet and creamy.

I always thought that the first section's taste was marked by overextracted fines. But the Illy chapter says the fines migrate down toward the bottom of the puck, and are necessary to control the rate of flow. Now, if you have a column of ground coffee, and send water through, the coffee will brew from the top down, since the water will extract all the top coffee, get saturated, and be unable to pick up solubles further down.

Bottom line: the fines brew faster, but are further down the puck; so when do they brew, into the first or second third of the shot? Turns out, it's in the second, sweet third of the shot, and that they may have been getting a bad rap.

The graph below represents flow data rates for a proper extraction (according to the Illys in their Book of Coffee) -- with flow data taken at 5, 10, 15, 20, 25, and 30 seconds into the shot. Seconds 0-4 are not on the graph. They have a bar chart but I have recast the data in the manner shown and have highlighted the middle time-slice of the shot (seconds 11-20). The old alt.coffee exercise, in which the espresso drawn during the second of the three time-slices tastes "sweet and creamy," happens to correspond with the greatest flow rates. One presumes that the increased flow during this stage produces an optimal extraction, which, at least in part, explains the more balanced taste of the middle time-slice portion.



As the coffee exits the filter, it contains solids (in colloidal suspension) that were extracted from all layers of the puck column. The column of water, as it passes through the puck, is extracting solids as it moves downward, the colloidal suspension becoming more and more concentrated along the way.... unless and until the concentration becomes so great that the water's ability to dissolve additional solids is impaired or approaches zero. I assume this is what Jim means when he speaks above of "saturation"... namely, concentration of the suspension so great that no more solids can be dissolved.

Having underspent coffee in the bottom section of the puck could be caused by any number of things that contribute to flow rate, whether isolated or in combination with each other: too high a dose for the amount of water being passed through the coffee (brew ratio); insufficient cake porosity (either from too heavy a tamp and/or too fine a grind); insufficient brew pressure or too much brew pressure, both of which can result in a less than optimal flow; basket geometry and filter-egress characteristics.

Other things could contribute to this behavior too. It's a multivariable calculus. An optimality problem.

It could be that the extraction of the column of coffee always occurs along a gradient, running from most to least from top to bottom, and that the bottom of the puck will always retain more solids than the top of the puck. It stands to reason that the problem, if and when it occurs, would be most dramatic with tight gloppy ristrettos and least dramatic with generous thin lungos.

Regards
Timo

[JR_Germantown]

3 way valves don't suck anything -- the burp discretely as they vent the built up pressure.

Jim, I like to see a scientific approach to understanding what really goes on, rather than just accepting the conventional wisdom. Nice job, and I hope somoene can eventually understand all the data.

However, I must disagree on this part. Whether the 3-way valves initially "suck" or "burp" depends on your perspective. But there is actually some siphoning action going on (in my machine, anyway) when the moving liquid gets below the level of the group head. Otherwise, the puck flow would stop when the group head pressure is equal to the outside pressure.

Jack

[another_jim]

Jim, I like to see a scientific approach to understanding what really goes on, rather than just accepting the conventional wisdom. Nice job, and I hope somoene can eventually understand all the data.

However, I must disagree on this part. Whether the 3-way valves initially "suck" or "burp" depends on your perspective. But there is actually some siphoning action going on (in my machine, anyway) when the moving liquid gets below the level of the group head. Otherwise, the puck flow would stop when the group head pressure is equal to the outside pressure.

Jack

Funny you should say that. I was cleaning my machine, and removed the blind filter while I had the driptray out. Sure enough, I got a puddle from the 3-way as air flow was restored to the group. The puck is relatively high resistance too, so the siphon style flow probably has to wait till it's removed

[popeye]

Two thoughts:
1. Can that syphon be calculated by simply using the difference between the bottom of the filter basket and the bottom of the three way valve? It seems to be extremely small compared to the usual 9 bar. (1atm ~ 35' of water, our syphon is 4"?)

2. If time zero is when the first drop of espresso is observed, hasn't the top of the puck been brewing for longer? Even assuming that preinfusion is complete (the entire puck gets wet) hasn't the top of the puck been pre-infused longer? And how does the pressure ramp up from top to bottom?

I post the above thinking two things:
1. I am unqualified to post the above.
2. I wonder if anyone has done those preinfusion and pressure studies elsewhere (I heard someone reference illy science)

[popeye]

Another thought: The TDS for all three segments at time zero (roughly) averages to the TDS for the coffee. Shouldn't this be expected, because all the coffee - minus one drop - is still in the filter basket? In fact, it seems to be a good reality check for the measurements. So is the bottom absorbing the top, or is the dissolved top just in the process of flowing through the bottom? Sorry if i missed something in the description.

[timo888]

Another thought: The TDS for all three segments at time zero (roughly) averages to the TDS for the coffee. Shouldn't this be expected, because all the coffee - minus one drop - is still in the filter basket? In fact, it seems to be a good reality check for the measurements. So is the bottom absorbing the top, or is the dissolved top just in the process of flowing through the bottom? Sorry if i missed something in the description.

After the puck has been preinfused (i.e. a fraction of the brew water has been pushed through it at perhaps 2 or 3 bars of pressure), there is still plenty of water above the puck with a TDS approaching zero. That water hasn't touched any grounds yet. So it is misleading to think of the colloidal suspension itself as having levels. The spent puck has levels or gradations. The column of water passes through the puck and is dissolving more and more solids as it goes. Every drop of coffee suspension emerging from the filter and dripping into the cup contains dissolved solids that were taken from all levels in the puck.
Regards
Timo

[another_jim]

So is the bottom absorbing the top, or is the dissolved top just in the process of flowing through the bottom? Sorry if i missed something in the description.

Good question, there's no way to tell, barring a way of drying the puck which also removes the dissolved solids.

However, I think the question could be moot:

-- In the "blind" part of the extraction, the coffee particles absorb liquid, and the puck swells. For a double, it's roughly 30 mL, for a single, roughly 15 mL.
-- The top of the puck absorbs water; the bottom of the puck absorbs strongly brewed coffee.
-- Whether this absorbed liquid counts as "ground coffee" or "liquid coffee" is just pedantery; the point is that whatever it is, it stays highly concentrated for the first 20 seconds of the extraction.

One upshot is quite simple. Why is stopping the shot by blonding so much more consistent than going by time? The blonding occurs when the bottom of the puck exhausts. This happens well after the rest of the puck is extracted. It is thus an amplified indicator - any fast blonding shows that the entire puck is exhausted. That is why it works even though we all have different ideas on how light "blonde" is -- we're, quite correctly, responding how fast it's happening, rather than how light it is.

[timo888]

Why is stopping the shot by blonding so much more consistent than going by time? The blonding occurs when the bottom of the puck exhausts. This happens well after the rest of the puck is extracted. It is thus an amplified indicator - any fast blonding shows that the entire puck is exhausted.

The observation above is "on the money" and is a good practical guide for the barista, as what Jim writes always is. But I have to quibble a little with the next formulation because the observations have implications for my own hypotheses about the brewing characteristics of domestic spring levers. I would like to reformulate it slightly.

-- The top of the puck absorbs water; the bottom of the puck absorbs strongly brewed coffee.
-- Whether this absorbed liquid counts as "ground coffee" or "liquid coffee" is just pedantery; the point is that whatever it is, it stays highly concentrated for the first 20 seconds of the extraction.

The bottom of the puck IS the last to go, as Jim says. But if there has been a proper preinfusion, the bottom of the puck will have absorbed water and become swollen, and it shouldn't be absorbing much, if any, brew water as that flows down through the puck. And it wouldn't be absorbing solids, but water only. Because the water's capacity to dissolve solids is greatest at the top of the puck, where it is pure water, and least at the bottom of the puck, where it is a colloidal suspension containing large amounts of solids, the extraction of solids occurs along a gradient, from top to bottom. The water is gradually losing its power to dissolve solids as it moves down through the puck. I believe that is why the bottom of the puck retains (most of) its solids till the end.

This flow-influenced gradient extraction dynamic has implications for my hypothesis (stated elsewhere) that the distinctive quality in the cup (more coffee flavor with less body) of domestic spring levers is the result of extraction under 6 bars of pressure, using a taller, narrower diameter basket whose height-to-width ratio has implications for the flow rate at that reduced pressure. If the conical basket is used, with a smaller effective egress filter, the flow is further restricted, which can intensify the qualities in the cup.

Regards
Timo

P.S. Just noticed GScace's interesting conjecture about the temperature and pressure differences along the axial length of the puck, and their implication in the extraction. Hottest at top, coolest at bottom. Most pressure at top, least pressure at bottom. Hmmmm

[popeye]

So it is misleading to think of the colloidal suspension itself as having levels. The spent puck has levels or gradations.

Ok, thanks for explaining the colloidal suspension model. That helps me think about it in a new way. However, since TDS measurements at time 0 (And all times) were based off spent pucks (that were dried and re-brewed) doesn't my original question still stand?

[timo888]

Ok, thanks for explaining the colloidal suspension model. That helps me think about it in a new way. However, since TDS measurements at time 0 (And all times) were based off spent pucks (that were dried and re-brewed) doesn't my original question still stand?

Perhaps I am misunderstanding the description of Jim's experiment. But if I understand correctly...

Jim prepared six baskets of identical dose, grind, and tamp, and produced six extractions, stopping them dead in their tracks at varying times: after the first drop (time zero) and then at six-second intervals. The six variously spent pucks were collected and sliced into three layers, presumably equal to the eye, not by mass. At this point we have 18 slices.

These 18 slices were then oven-dried, as was enough coffee to make two more fresh espressos. After drying, the two fresh unspent doses were brewed at different dose-to-water ratios. The eighteen slices were also used to make more espresso, at another ratio, but all of them at the same ratio.

NOTE: I do not understand Jim's nomenclature for expressing these ratios. A ristretto is about 1:1 in dose-to-beverage weight and normale 1:2 and lungo 1:3, per Andy's scheme. Jim, are you going by volume?

The TDS of each of these subsequent extractions was noted. These 18 rebrews were then taste-compared by Jim against the two fresh cups made from the oven-dried 100% unspent fresh coffee.

In light of my (mis?) understanding of the protocol, I am not sure what you are referring to by "TDS for all three segments".

popeye asked:

The TDS for all three segments at time zero (roughly) averages to the TDS for the coffee. Shouldn't this be expected, because all the coffee - minus one drop - is still in the filter basket? In fact, it seems to be a good reality check for the measurements. So is the bottom absorbing the top, or is the dissolved top just in the process of flowing through the bottom? Sorry if i missed something in the description.

Regards
Timo

[another_jim]

Jim prepared six baskets of identical dose, grind, and tamp, and produced six extractions, stopping them dead in their tracks at varying times: after the first drop (time zero) and then at six-second intervals. The six variously spent pucks were collected and sliced into three layers, presumably equal to the eye, not by mass. At this point we have 18 slices. ... These 18 slices were then oven-dried,

Yes

as was enough coffee to make two more fresh espressos. After drying, the two fresh unspent doses were brewed at different dose-to-water ratios. The eighteen slices were also used to make more espresso, at another ratio, but all of them at the same ratio.

No.

The oven dried puck-parts were brewed like regular coffee: 4 grams to 80 grams of water; I tasted the brew for strength (against the regular coffee brewed at 4, 2 and 1 gram per 80 grams) and measured the TDS. The more the puck part tastes like coffee, the higher the TDS, the less extracted it is.

[timo888]

The oven dried puck-parts were brewed like regular coffee: 4 grams to 80 grams of water; I tasted the brew for strength (against the regular coffee brewed at 4, 2 and 1 gram per 80 grams) and measured the TDS. The more the puck part tastes like coffee, the higher the TDS, the less extracted it is.

OK. I'm still a little confused about what happened after the oven-drying.

We have 18 oven-dried slices:

..................T0......T1......T2......T3......T4......T5
Top Layer.....1.......4.......7........10.......13......16
Mid Layer.....2.......5........8........11......14......17
Bot Layer.....3.......6........9........12.......15.....18

The puck from the cup where the extraction was stopped at the first drop is the T0 puck. The puck from the cup where the extraction was allowed to run to 30 seconds is the T5 puck. Each cup's puck was divided into three layers.

So far, I am clear.

When you made the regular brew from these slices after they were oven-dried, did you make 18 cups of regular coffee, with 4g being the actual oven-dried dose added to 80g of water, and then take 18 actual TDS readings?

Regards
Timo

[another_jim]

When you made the regular brew from these slices after they were oven-dried, did you make 18 cups of regular coffee, with 4g being the actual oven-dried dose added to 80g of water, and then take 18 actual TDS readings?

Yep, those are the TDS readings I report (except I did two series of shots, so there's a pair of lines for each puck section). The horizontal lines come from 4 readings each of oven dried fresh coffee at 4, 2 and 1 gram per 80 grams. The water used was distilled, and had a TDS of around 6. TDS readings are not linear as you can see from the 3 concentration levels of the same coffee; but the three solutions of the unbrewed coffee give a scale for the puck concentrations.

[timo888]

Yep, those are the TDS readings I report (except I did two series of shots, so there's a pair of lines for each puck section). The horizontal lines come from 4 readings each of oven dried fresh coffee at 4, 2 and 1 gram per 80 grams. The water used was distilled, and had a TDS of around 6. TDS readings are not linear as you can see from the 3 concentration levels of the same coffee; but the three solutions of the unbrewed coffee give a scale for the puck concentrations.

There are two series because the same experiment was run twice?

One thing remains. The colored lines on the graph represent the TDS of the top, middle, and bottom slices of pucks T0, T1, T2, T3, T4, and T5, brewed as regular coffee in a 4g to 80g concentration. The y-axis represents TDS and the x-axis represent the six different pucks in the experiment. But there are seven items represented on the x-axis.

Regards
Timo

[another_jim]

There are two series because the same experiment was run twice?

But there are seven items represented on the x-axis.

I was too lazy to set the X axis ticks at 6 second intervals, so they are at five. But the line segments run 0 to 6 to 12 to 18 etc. The segments connect the lower readings of each pair and the higher readings.

[timo888]

I was too lazy to set the X axis ticks at 6 second intervals, so they are at five. But the line segments run 0 to 6 to 12 to 18 etc.

You will have to change that for the hardcover edition

The segments connect the lower readings of each pair and the higher readings.

So the lines connect readings from different pucks belonging to the same extraction-time cohort? The data series were intermingled within cohorts for the graph?

Could you post the raw TDS readings for all 36 slices?

Did you weigh each oven-dried dose after the oven-drying, before brewing the 4g to 80g ratio samples? The 4g is an actual measurement and not an extrapolation from an initial 12g dose?

Regards
Timo