Scott Rao

View Original

My two-week espresso education on the road

Serving coffee at Sweet Spot Kaffee in Munich

I recently spent two weeks with John and Bugs from Decent Espresso, doing demonstrations of the DE1 as well as my forthcoming Filter 3.0 basket, throughout the Netherlands, Germany, Switzerland, and Austria. We did sixteen demonstrations in twelve cities, and it was a blast. Our hosts were incredibly gracious and generous, and everyone who came to the events seemed to enjoy the vibe and the coffee as well as the insights about extraction we have gained from using the DE1. 

As always happens when I spend time with John, we honed many of our ideas about coffee extraction, and came up with new ideas to work on in the future. Below is a blog-post version of discussions we had at our demonstrations. No matter how experienced you are at making espresso, I trust you will find some interesting food for thought in this post. 

The history of espresso machines

In the early days of the Decent DE1 machine, John created various profiles to mimic every significant espresso profile from history. The profiles included versions of lever profiles, the E61 profile, and the flat nine-bar profile common today. John did this to demonstrate the versatility of the DE1 and to show that anything any other machine can do, the DE1 can do it as well. That is still the case, and the DE1 does many things no other machine can do (a few manufacturers make claims about flow and temperature profiling, but don’t provide the data needed to see the accuracy of the claims.) During our demos, John talked about the evolution of espresso-machine design, where it where it went wrong, and how today’s most popular design is not well matched with lightly roasted coffee. 

Early espresso machines had manual levers (the barista applied pressure using a lever) while other machines had spring-loaded levers to apply pressure more consistently, and with an arc (preinfusion followed by a pressure rise, peak, and gentle decrease) similar to the pressure curve of the allongé graph in the next section of this post . The spring-loaded lever machines were arguably the peak of Italian espresso-machine design in terms of optimizing coffee quality. Over the years, most manufacturers abandoned the spring-lever design in favor of pump-driven machines that apply a constant nine bars of pressure. Such machines require less barista time and effort, but yield coffee flavor and extraction levels inferior to that of spring-loaded levers. 

Allongé

At the beginning of every demo, I served two allongés, a style of very long espresso. I pulled allongé in part because it’s a delicious way to extract lightly roasted coffee, and in part because it yields a lot of coffee, making it easier to serve a large crowd. 

Back when I founded Café Myriade in Montreal in 2008, every cafe in the city was expected to serve allongé, and it was the second-most popular drink after filter coffee. Back then, baristi simply pressed the shot button multiple times, or hit the continuous button, to make a very long espresso, using their usual espresso blend, grinder, and grind setting. Of course, such shots were very bitter and astringent. We did things differently; I dedicated a grinder to allongé, used a coarser grind than for espresso, created a diligent puck-prep routine to decrease channeling, and served only fruity, lightly roasted, washed coffees. 

I served allongé because it was expected; what was unexpected was how much I liked it. With our light coffees, allongé enhanced the fruit acids and made some memorable shots. Visiting coffee pros were often shocked by how much they enjoyed a 110-ml espresso! Subsequent trends such as “coffee shots,” “turbo shots,” and “extractamundo dos” were all descendants of my allongé, and I’ll discuss later how and why they differ from allongé. 

When I told John about allongé eight years ago, he programmed it as as 4.5 ml/s flow profile on the DE1.  That meant the machine would apply whatever pressure was necessary to yield 4.5 ml/s. Although the allongé we had served at Myriade were quite good, they didn’t compare to the allongé we served this month on the Decent, since at Myriade we had a Kees Van Der Westen Mirage, an excellent machine, but it could not control flow. 

For the record, the allongé we made on tour used 15g of grounds to yield 73g of coffee in 25 seconds.  

Note that the constant flow caused the pressure to naturally rise, peak, and decline steadily.  Pressure rises as the puck compresses and drops as the viscosity of the slurry decreases (Thanks to Jonathan Gagné for the insight that it is decreasing viscosity, more so than puck erosion, that causes pressure to drop.)

Graph of textbook, flow-profile allongé;  green=pressure, blue=flow,  brown=shot weight


Fines and channels

The pressure curves of most allongé look like the one above. However, when I pulled a few allongé using an Ethiopian coffee roasted by DAK, something interesting happened: the pressure and the flow levels both stayed constant. Many Decent owners are aware that the curves of Ethiopians and decaffeinated coffees behave differently from other coffees; that is because Ethiopians usually generate more fines than any other origin, and decaffeinated coffee yields more fines than even Ethiopians do. Larger amounts of fines help maintain puck integrity for longer.

Allongé of Ethiopian coffee with flat pressure and flow curves


Turbo Shots, Extractamundo Dos

A few years ago, Chris Hendon, et al. published a paper proposing to optimize the consistency and flavor of espresso by pulling fast and short “turbo shots.” They didn’t explicitly say their findings were only for lightly roasted coffee, but whether they knew it or not, their system wouldn’t work as well with dark roasts. The study’s authors used 15g in a 20g basket to yield 40g, using flat-six-bar pressure, and an EK43 grinder set coarser than the standard espresso grind size. They claimed more consistency and better flavor by pulling such shots vs conventional slower, higher-pressure espresso shots. 

When I read the paper, I appreciated their efforts, but lamented that they didn’t use a Decent, because had they used one for a while, they would surely have framed the paper differently. They were discovering things John, some DE1 owners, and I had known for years: lighter roasts taste better at faster flow rates, the optimal grind size is not the one that yields the highest extraction, but is slightly coarser than that (in order to limit excessive puck compression/flow suppression, which promote channels), and shots from most large, flat-burr grinders such as the EK, should be pulled shorter, presumably because of lower fines production than most traditional espresso grinders, as a lack of fines causes the puck to lose integrity earlier in a shot.

I believe the choice of six-bar pressure was merely a hack to decrease the severity of channeling, whether the authors were aware of that or not. Had they had better puck prep and a proper, complete preinfusion, they would likely have had different results, and preferred slightly higher water:grounds ratios. More fines would have also nudged them toward higher ratios. 

When I made allongé in Montreal, I used 22g in a 20g basket to create a deeper bed to decrease the risk of astringency. Deeper espresso beds generally increase body, decrease clarity, and lower the risk of astringency. I also used a Robur, which has conical burrs and produces more fines than an EK. The extra fines allow pressure to persist longer, and decrease channeling by blocking larger flow paths. These days, one can see all of those dynamics in the Decent’s pressure and flow curves. 

Although I was the first person in specialty coffee to use an EK grinder, I have never enjoyed the quality of espresso extraction from an EK, at any flow rate or ratio. It’s simply not well-suited to espresso, which I believe skewed the results of the paper. That points to a weakness of scientific method when studying a complex system like coffee: it’s difficult for a scientist to know what is “true for coffee in general” and what is “true only for our particular equipment and coffee and methods.” The DE1’s graphs plus the 5000+ members of the Decent Diaspora forum together make up a more effective system for studying espresso extraction, as the DE1 owners’ collaboration is akin to a “big data” approach. DE1 owners have uploaded data from over one million shots to https://visualizer.coffee, a very useful resource for coffee researchers. I believe such an approach is more effective and appropriate than traditional science for learning about a complex system. 
In summary, the “turbo shot” is shorter than allongé because one needs to use a lower water:grounds ratio to compensate for the channeling and uneven extraction produced by lightly roasted coffee with modest amounts of fines, pulled on flat-pressure machines.

EKs and constant-pressure machines are suboptimal for getting the most from lightly roasted coffee. 

A recent variation on allongé/turbo shots is the “extractamundo dos” profile. Extractamundo is a shorter version of its predecessors. John and I believe extractamundo is popular with DE1 owners mostly because pulling shorter, fast-flowing shots decreases the flavor damage from roast defects, imperfect puck prep, and channels related to low fines production. 


My prediction for the future is that as more people use the DE1, and the forthcoming Barista Hustle AutoComb improves puck prep, baristi will pull light roasts at higher flow rates and longer ratios, on average, than they do currently. 

Extraction of Light Roasts vs Dark Roasts

Over the years, roasters have attempted to optimize coffee flavor from nine-bar machines. However, while such machines do a passable job with darker roasts, they do a poor job of optimizing light roasts. I know I’ll take flack for that from people who use nine-bar machines and love their espresso, but any barista who has spent years using a variety of profiles, instead of only nine-bar machines, is likely to agree that a flat-nine-bar machine does not optimize light roasts. I’m not saying what one should like, I’m saying that certain systems tend to produce better overall results from lightly roasted coffee. For reference, precious few Decent owners choose the Italian nine-bar profile on the DE1 to pull straight shots with light roasts. 

To optimize dark roasts and minimize bitter, burnt, and acrid flavors, one may want to pull shorter shots, target lower extraction levels, and extract at lower temperatures. Paul Chan of Decent recently created the “80s  espresso profile” to decrease the harshness of darker roasts. (You can also add salt to espresso to decrease bitterness.)

The allongé and my “blooming espresso” profile on the Decent were breakthrough profiles for getting the best flavor from lighter roasts. Blooming was inspired by drip brewing methods: the profile consists of a preinfusion phase, a 30-second bloom with a small amount of applied pressure to prevent the puck from expanding (and possibly losing integrity), a pressure ramp, and a slow pressure decline. When I craeted the blooming profile, I was shocked to experience extraction levels as high as 29% with light roasts and 2.5:1 ratios. Unlike dark roasts, quality light roasts tend to taste best at high extraction levels, and some of the most memorable espresso shots of my life were blooming shots of Kenyan coffees extracted to 29%. 

During our demos, John perceptively pointed out that both allongé and blooming use non-standard methods to boost extraction from lighter roasts. Allongé runs a lot more solvent (water) through grounds than any other espresso profile, and blooming uses a 30-second bloom to increase contact time between the grounds and water. Notably, neither technique makes great espresso from darker roasts, since they yield such high extractions, which make harsher dark-roast flavors more prominent. 

Blooming espresso: preinfusion is done at 4 ml/s, pressure gradually and naturally declines during a 30-second bloom, and pressure rises and falls during the secondary flow stage of 2 ml/s

Filter Coffee

During the second half of our presentations I made filter coffees for the audience using my forthcoming Filter 3.0 basket. It’s called 3.0 because over the years, in my attempt to turn the DE1 into a filter-coffee brewer, there were two earlier iterations. 

Filter 1.0 made use of a blank portafilter basket with numerous holes in it to turn the basket into a spray head. The barista set a manual brewer under the basket, pressed START, and the DE1 ran a pulsing flow profile to make filter coffee. Unlike, say, a cheap home filter machine, since the DE1 can apply a small amount of pressure, spray from the holes was even, and the result was pretty good filter coffee with no barista effort or skill required. The results were good, but I wanted better. 

In Filter 2.0, I used 18g of finely ground coffee in a 25g basket. I used a flow profile that occasionally applied up to 0.2 of pressure. The goal was to make a coffee concentrate with as little pressure as possible, and to dilute the concentrate before serving. Results were sometimes amazing, but inconsistent. The problem with 2.0 was it demanded flawless puck prep. Great puck prep led to excellent results, but when puck prep wasn’t perfect, astringency was noticeable.

Filter 3.0 in action

Filter 3.0 is almost ready for production, and I’m happy to report, makes excellent filter extractions every time, with no barista skill or time required. The barista drops a 57mm paper filter in the basket, pours in 22g of grounds, shakes the basket to level the grounds, and presses START. The DE1 runs a flow profile consisting of a prewet, bloom, and successively slower flow rates, with a declining temperature profile. Extractions hover in the low 20s and brew times are 4:00-4:15. The basket is a version of a “no-bypass brewer,” a term coined by Jonathan Gagné to describe brewers such as the Tricolate, NextLevel LVL10, and the forthcoming NextLevel Pulsar. Since there is no possibility of bypass, the grind for 3.0 is extremely coarse, similar to that one would use for a commercial batch brew. I’m thrilled with the results, and the audiences at our demos seemed universally happy as well. Several of the host cafe owners on our tour expressed their interest in buying a DE1 in order to make Filter 3.0 in their cafes. 

Q&A

At the end of each demo, we talked about topics such as water chemistry (see my recent posts here and here), grind quality, how fines affect espresso extraction, Decent’s new “forever warranty” program, and our plans for a new commercial Decent machine, the Bengle. Stay tuned for more, and thanks to everyone who hosted us and joined our demonstrations!

Questions and comments are welcome.

Join the Prodigal Mailing List to get first crack at new releases