Birth of Turbobloom and Extractamundo!

EDIT: Yang Luo kindly translated this post to Chinese and can be accessed here.

When Cameron et. al. released their peer-reviewed paper on turbo shots, it kinda took the espresso world by storm. It was controversial in a very interesting sense. The science behind it (by the standards of a lot of coffee literature) was pretty solid, the authors included their failure to line up with a mathematical model of puck behavior, which meant that their findings were based on real life shots, and it may have for the first time revealed to a broader (non-espresso afficianado) audience that espresso shots can go beyond the darling 1:2 ratio or that they can end within 15 seconds. But this isn’t a talk on turbos. If that’s what you came here for, look no further than Cameron’s own blog post on the topic (or alternatively Lance Hedrick’s wonderful detailed video that he’s done in consultation with Cameron, if you prefer the movie format).

Instead we’re here to look at the history of profile development behind what is currently arguably one of the most popular family of profiles in the DE1 community, and how they came about as a result of a series of thoughts outside of the box by a bunch of different folks across the world. The end result, as you’ll shortly see, is what is probably the most forgiving and easy-to-use profile designed in the DE1 community. This post also serves as insight into how thoughts and discussions lead to incremental changes to approaches in espresso, with the TurboBloom and Extractamundo! profiles being key snapshots or poles in the ground that somehow “just work”.

So bloom isn’t just for pourovers? (and flowers)

The most popular example of when the concept of “blooming” or “soaking” an espresso puck without letting in extra water can be traced back to Scott Rao. Note that we’re not focusing on the filter sandwich part of the discussion but rather the concept that letting the puck soak in the water that was let in to wet the puck, hypothetically allowed the top and bottom of the puck have lesser difference in temperature, besides also encouraging a more even extraction due to potentially lessened pathways of differing tds (which will act as precursors to channels), both of which result in increased extraction evenness.

Quite a lot has happened since Rao first proposed the idea in 2018. Even though single-dose “titan” grinders had slowly started seeping into the home enthusiast space back in 2016 (maybe a bit earlier if you choose to include the EK), over the last couple of years, the market for purpose-built single-dose grinders and a variety of burr geometries to accompany them, seems to have exploded with options. In particular, I want to bring attention to 64mm SSP Multipurpose (henceforth called 64MP) and 98mm SSP High Uniformity (henceforth called 98HU) burrsets, which I consider the driving force behind the prominently used versions of turbo family of profiles in the DE1 community.

To give some context, Rao’s blooming profile works as a constant-flow profile on the DE1. However, constant-flow profiles are notoriously hard to dial-in due to their sensitivity to grind, and if you care about consistent peak pressures, they get even harder to nail down. Being the tinkerers that they are, DE1 users Stéphane Ribes and Luca Costanzo figured out that using a dynamic bloom programmed to exit at 2.2 bar (instead of the fixed 30 second bloom in the original profile) gave them more consistent results. A few months later, the now famous Turbo paper was published. Another few months down the line, DE1 users Jan and Collin Arneson had both come up with their takes on turbo profiles that could be implemented on the DE1. Add another few months to the mix, and user Jan noticed in mid-2021 that they’re getting tastier results with short blooms (about 5 seconds) added to turbo shots.

Enter JoeD

The blooming turbo shots above though aren’t the one we’ll be looking at today. Down here in Southern California, arguably one of the smartest minds I currently know of in coffee, who goes by the name of JoeD in the Decent community, was developing his own approach to turbo shots. For starters he developed his own take on a turbo shot that looked something like this:

Figure 1: Graph shows a turbo shot that starts water fill at 8-9 mlps, with the input flow rate decreasing as pressure starts rising at the 4 second mark. The profile then holds the pressure at 7 bar for a few seconds, before allowing it to decline to 6 bar starting around the 10 second, till the shot ends at the 14 second mark. The input flow meanwhile dips to 3 mlps around the 10 second mark before increasing again as the puck loses solubles. The output flow starts around the 5 second mark, beginning to track a little higher than the input flow line around the 8 second mark. (click here for color-blind-friendly version)

Around the same time as Jan, Joe discovered that adding short zero-flow blooms of two to five seconds to turbo shots not only improved taste, but also increased extraction yield. This discovery was made using Stéphane/Luca’s easy-blooming albeit at a much faster flow rate than a regular espresso shot, effectively creating the first turbobloom.

Figure 2: Example of an early turbobloom. Water fills in at 8 mlps and stops when the pressure hits 4 bar at around the 6 second mark. It then “blooms” till the pressure drops to 2.2 bar around the 8 second mark, then water starts flowing again to make the pressure reach its 6 bar target at the 14 second mark, after which it gets a slight decline to 5.5 bar till shot end at 19 second mark. The input flow peaks at 4 mlps when pressure peaks and declines slightly after that. The output flow line tracks starts ramping at the 6 second mark and broadly tracks input flow a little bit higher. (click here for color-blind-friendly version)

To really put things in perspective about how short that bloom really is, here’s the shot above compared with its non-bloom counterpart in faded colors:

Figure 3: Previous turbobloom shot overlaid against non-bloom version. Almost everything is the same except the pressure doesn’t dip once it starts ramping and ramps to 7 bar at around the 13 second mark, only declining slightly to 6.5 bar till the shot ends at 19 seconds. (click here for color-blind-friendly version)

If you click on the link above and compare the shots’ details, you’ll notice that the shot with bloom extracted more than a percent higher (based on unfiltered tds). Not only that, when Joe pulled back-to-backs with and without blooms, in both blind and non-blind taste tests, he significantly preferred the shots that had the two seconds of additional bloom.

The day we all met

In summer 2021, when I was still brand new to espresso making, DE1 users in southern California agreed to have a meetup, and I got a chance to host a whole bunch of coffee enthusiasts at my place (fully vaccinated, in an outdoor ventilated space). This is where I first got to taste Joe’s TurboBloom (the experience and aftermath of the grinder part of this is documented here). Using his 98HU on the P100, he was pulling 1:2 ratio shots that were delicious syrupy nectar-like renditions with lovely flavor separation at high extraction yields.

So as a bunch of folks were incessantly pulling shots, calibrating taste, discussing grinders etc., DE1 user Chenchen unveiled his Jian-Yi lever machine. Although amidst hosting I didn’t get a chance to be actively involved in the discussions that day involving the Jian-Yi, one of the things that came up for discussion was how the shots from the lever machine were tasting less harsh.

Figure 4: Chenchen’s Jian-Yi lever machine which may have inadvertently led to the birth of low-temperature turbobloom shots. Image taken from actual day of meetup so this is kinda a moment in history

This train of thought persisted in Joe’s mind, and that ultimately led him to doing a couple of things with the profile – drop the temperature down by several degrees (I have previously talked about why the DE1 may benefit from lower temperatures), and induce a temperature drop as the shot progresses so as to reduce extraction of harsh compounds as the shot progresses. Something magical happened when that change was implemented. All of a sudden, any persistent harshness due to how fast some burrs extract, or whether your machine’s headspace necessitates a finer grind, was magically reduced. Here’s an example of a typical temperature decline (link):

Figure 5: Example of a typical temperature decline during TurboBloom. The sensor that measures temperature right above the puck starts out at around 85C, staying there till blooming ends at around the seven second mark, after which it declines to about 79C by the time the shot ends at 15 seconds (click here for color-blind-friendly version)

Note that even though we’re setting a very low temperature as a target, within the duration of the shot the temperature at best declines between 4-6 C because however quick the DE1 can heat or cool water, there is a finite amount of time it needs to do so depending on the shot’s flow rate (I bring up flow rate because a faster flow rate means more volume of water needs to be heated or cooled per second, which for the way the machine is designed, is harder to instantaneously heat at higher flow rates, where higher is 3 mlps and above, and easier to cool). Also the cool water (green mix temperature line) isn’t just cooling the coffee but also the group parts in its way, which depending on the material will also determine the extent of cooling.

Another caveat here is that the actual target decline will also be determined by your burrs. Grinders with wide boulder distributions like Niche Zero may benefit from both a relatively hotter start and less steep decline as compared to burrs with narrow distributions like 64MP.

Is your water salty?

Now remember that when Joe was developing this profile, he was using the Frankenjolly equipped with 64MP burrs. The more conventional total hardness (GH) and alkalinity (KH) levels of 80GH/40KH was somewhat acceptable to his palette. But by the time he started pulling these shots on 98HU, there was an immense amount of harshness that 80GH was bringing about. Joe mitigated some of this by switching to using his reverse-osmosis water that had a GH of 20. However, there was still some remnant harshness because these burrs not only extract fast, but possibly have such a narrow boulder distribution that both good and bad parts of a roast get highlighted, which was making the astringency a little hard to tame.

Around this time, EAF (Espresso Aficionado Forum) user Fam came up with a clever thought – if alkalinity works in a volumetric range (ppm or mg/L), then it might make sense that espresso needs more alkalinity per unit volume of water than filter, and discovered that they started getting better results in espresso with alkalinity upped to 90KH. Joe also decided to take this approach and that seemed to be the last nail in the coffin to taming the harshness coming from 98HU burrs. Surprisingly, despite increasing the alkalinity to such an unconventionally high amount, there didn’t seem to be a loss in flavor separation or in acidic intensity in the resulting espresso.

It is worth noting here that a big chunk of the Decent user base who use fast-extraction burrs now use high-alkalinity water based on Fam’s original idea. There’s quite a bit of cross-pollination of ideas between EAF and Decent’s user base.

Extract a what?

You would have thought that Joe would take a breather after having popularized TurboBloom as a solid forgiving profile you can use on any grinder. But instead, it bugged him that TurboBloom was plagued with an aspect of puck resistance that is possibly more conducive to channeling – as the puck loses more solubles, to maintain the same amount of pressure, the machine has to pump more water per second (increase flow) through the puck. If we go by the theory that localized regions that have lost more solubles have lesser tds as a result, and therefore lesser viscosity, and thus as a result water may tend to flow with more ease through these regions (long-form way of saying channeling), then an attempt to increase flow to sustain pressure might just be more conducive to the process.

What I’m about to describe is probably one of the most clever implementations of one of DE1’s most complex (to me) capabilities. The DE1 is equipped with limiters for both pressure and flow. However, there is also an option to decide how aggressively you want that limiter implemented called “range of action”. If your “range” is small, the machine will be extremely aggressive about trying to stay at that limit. However, if you give it a very large range, it will try and stay at the limit, but only very weakly. So for example, if you use a flow rate limit of 1 mlps, and apply a range-of-action of 3 mlps, and grind coarse enough, it will not actually restrict the flow to 1 mlps, but rather allow to go it all the way to even 3.5 mlps, before beginning to work hard to restrict it at around the 4 (1+3) mlps mark. However, with output flow rates as high as 3 mlps, you are already in turbo regime. An amazing thing happens though when you do this using range of action – because the flow limiter is weak but still present, this function prevents the flow from increasing as the shot progresses (link to shot example).

Figure 6: Example of an Extractamundo! shot with the water filling in at around 8 mlps till pressure ramps to 4 bar around the 6 second mark, after which it blooms till pressure declines to 2.2 bar around the 8 second mark. Pressure then ramps to 5 bar at the 11 second mark despite the target being at 6 bar, and declines to to 4.5 bar as the limiter constrains the input flow to around 3.8 mlps once pressure has peaked. The output flow starts at around the seven second mark, broadly following the input flow line a little bit below till the shot ends at 17 seconds. (click here for color-blind-friendly version)

In other words, this mimics a gicleur (with the max flow rate determined by your grind size/puck resistance). More often than not, constraining flow ends up resulting in a pressure decline, despite having a 6 bar target. While this concept was originally called the “resistor” (as it resists flow increase), DE1 user Jorgen came up with the more attractive name “Extractamundo!” If you see the name “Extractamundo Dos!” floating around it’s only because Joe figured that the 6 bar ramp wasn’t necessary, with the puck’s own resistance being a better determinant of the maximum pressure it can reach, but it still uses the limiter + range of action.

For those interested in seeing a detailed example of how the range-of-action works when it’s at its limit, please refer to the Instagram highlight I made of it a while ago.

Some folks may also notice that the behavior of Extractamundo! resembles the Yeet. EAF user Shotwell confirmed a while ago that the ancestry of Yeet can be traced back to Rao’s original IG post on his Allongé. There’s a subtle difference though – Yeet usually depends on the pump’s max output flow to reach a pressure level below target, whereas Extractamundo! allows one to set the range of action to achieve the flow rate they desire.

Why should I care?

Ever since I’ve used Extractamundo!, I have rarely had to sink a shot. With a combination of coarse grind (turbo regime), bloom (extraction evenness) and flow restriction (reduced channeling probability), this profile is so forgiving it has become kind of a North Star for a lot of us in the DE1 community. And it’s kinda fascinating to realize that even though Joe put in a lot of the hard work, there were a bunch of significant events in the recent history of espresso-making that has led to this profile.

Acknowledgements

JoeD – For proofreading this post
Matthias – For pointing me to viridis colormap
Yang Luo – For translating to and publishing the post in Chinese

If you’ve made it till here, consider supporting NOT me, but the hourly coffee worker by donating to GoFundBean

Subscribe to keep your pocket updated with coffee science

4 responses to “Birth of Turbobloom and Extractamundo!”

  1. Hi! I’ve translated this article into Chinese language so that more Decent users can learn about this awesome history behind. Would it be okay if I publish it as a personal blog post?

    Liked by 1 person

    1. Hello Yang, thank you very much for translating it. I’m good with you publishing the translation on your blog as long as folks like Joe and others who I’ve mentioned in the post get credit in the translation as well. If you’re okay with it, please share the link once it’s published so I can link it here as well.

      Like

      1. Okay here’s the Chinese translation – I linked to the original post (this one) in the beginning and kept the Acknowledgement in the end in verbatim. Lemme know if I miss anything!

        View at Medium.com

        Liked by 1 person

      2. Thanks a lot Yang, very kind of you! I have cross-linked to your post at the top.

        Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: