DE1 Teflon Grouphead Shower Upgrade

A while ago, Ricco (officially the nicest person in the Decent user community), reached out to me saying he heard from John Buckman (Decent’s owner) about prototype teflon grouphead shower parts being available for purchase. This is significant in a number of ways. On the surface, as written by Decent’s official channel, the pieces theoretically act as insulators, and should therefore allow more accurate temperature measurement. However, this has a lot of implications when retrofitting into the current machine and firmware. But first, let’s talk about why these parts matter to me in the first place and why John agreed to let me purchase these prototype parts.

Do you even temp bro?

While a lot gets talked about the DE1’s pressure and flow profiling, there’s not a lot of discussion to be found about temperature profiling. This is a bit disappointing because the amount of thought that’s gone into allowing lots of temperature control and modulation is mind-boggling. Over the course of this post, the term “temperature control on DE1” is synonymous with Ray Heasman, who’s the inventor/lead internal designer of this machine. So think of temperature control on this machine being somewhat like this – there’s temperature sensors at various points along the water path that act in sync. So if you’ve set your temperature target to 93°C, there is a sensor right above the puck in the dispersion block of the grouphead that’s measuring it. The machine pumps water from the tank, through the pumps, tubes and heating elements, and sometimes discarding some back into the tank based on what the sensors along the pathway decide in conjunction with how much heat it will take to keep the group measurement at 93°C a few seconds down the line. It’s fascinating! Not only does the machine have to take into account how to sustain the water temperature as the shot progresses, but also account for heat losses it incurs along the way (remember this as we’ll have to revisit this later on).

The earliest example of temperature profiling on DE1 I came across was actually by Scott Rao on his blog, where he describes the original blooming espresso shot in detail. If you notice in the blog post, even though he intends to pull the shot at 92°C, he starts preinfusion at 98°C to account for the heat the puck loses during its long 30 second soak. Other than that though, when I got on diaspora, the only mention I found of temperature was the predominantly asked question by new users – “Why is the default profile’s temperature set so low to 89°C as opposed to 93°C on <insert previous machine’s name>?” There’s probably a few reasons for that as far as I can surmise. The stock machine has more grouphead space between the puck and shower screen as compared to other machines. Folks have speculated that this necessitates a finer grind. Decent’s stock 18g basket is also a low resistance precision basket, which could be a factor behind the finer grind. If you are grinding finer though, provided your puck isn’t getting clogged , you are extracting more than you would have on your previous machine. And if you were already at the highest possible tasty extraction at coarser grind and 93°C on your previous machine, you’ll be pulling nasty shots at finer grind and 93°C on the DE1 for a similar profile. In addition, the DE1 tries to avoid as much thermal loss as possible in its water path. So if your previous machine needed the heat of a 93°C temperature setting, remember that a big chunk of that heat was being leeched by components in the water’s path, leaving the remaining for extraction. So the lower default temperature setting of 89°C could be indicative of lower thermal loss in the DE1’s water path.

Then why are you not happy?

Although this story deserves a post of its own (and will get one), sometime last year, a bunch of changes were culminating towards what would become known as the turbobloom profile. The market was awash with fast-extraction burrs, there were now options beyond EK43 for 98mm burr carriers (Option-O, Kafatek and the now obsolete Levercraft Ultra). There’s a few things about SSP’s 98mm burrs that can make espresso-making annoying. The way particles shatter with those burrs creates pucks that can get extracted upto ~50% of max tasty point extraction within ten seconds of the shot getting started, if not sooner. I’m not kidding. So everything gets thrown at the puck in hopes of getting less astringency out of the shot – coarser grind, short ratios, lowered water GH, WDT, ICM (or alternatively filter on bottom), puck screen on top, and yet, even though you had 1:2 syrups with high flavor separation, more often than not there was still astringency. Now, it’s not abnormal for an espresso to be astringent, but controlling it within a short ratio when targeting high EY can be difficult.

The guy who came up with the version of turbobloom that I use (there’s multiple versions that came about at around the same time) is JoeD. During summer of last year, Joe helped coordinate a meetup between a bunch of espresso enthusiasts locally (outdoors, ventilated and vaccinated), and a byproduct of that was that folks who owned a machine in addition to the DE1, ended up bringing it along. One such machine was the JianYi lever machine owned by DE1 user Chenchen. When pulling shots using the JianYi, there were discussions surrounding its temperature control that led Joe to wonder what would happen if he did a couple of things – reduce overall temperature further (85°C and below) and then once the puck starts losing solubles, decrease temperature further. Something like this:

Figure 1: A typical declining temperature profile in turbobloom

You can view more details of the shot at this link. There’s quite a few things going on if you look carefully. The shot has begun preinfusion (1) at around 84°C (solid red line) but is being made to target 72°C (dotted red line). However if you look closely at the temperature mix (green line) which is the water being streamed to the grouphead to achieve target temperature, it reaches about 75°C, which in turn results in the actual grouphead temperature coming down to about 79°C. You may think 5°C reduction is a small amount, but look at the duration in which it does that reduction – 12 seconds!

It’s all in the mix

Let’s first address why we’re going for this sharp a decline at that particular stage in the shot. The point in the shot where you see mix temperature begin to decline (2) is where the “bloom” ends (think of bloom as a phase in the shot where no water is being pumped into the puck, which in practice adds a lot to evenness of extraction). As a general rule of thumb, for fast-extraction burrs, it’s very normal for a big chunk of extraction to be done by this point. This in turn implies that the puck has lost a lot of solubles, and the pathways that have lost more solubles will have lowered viscosity and will provide lesser resistance for water to flow through, i.e. channeling. Remember again that this is inevitable. We have thrown everything we can at the puck to improve evenness but it’s still “falling apart”. So banking on the fact that a big chunk of extraction is done, we lower the temperature to reduce probability of leeching astringent compounds. Astringent compounds are usually harder to extract but in a pressurized environment it just implies delaying the inevitable.

With this, some kind of a dam broke and shots were not only extracting high at these lower temperatures, but astringency and harshness were significantly reduced. And that kinda was like the last nail in the coffin.

“Not sure if you’re interested, but…”

Or so we thought. Ricco’s ping one fateful morning resulted in me getting a couple of ridiculously lightweight circular thingies via UPS which would end up giving me fascinating insight into how temperature control works in the DE1.

Figure 2: Teflon grouphead shower block parts

That I was overjoyed is an understatement. This was the first time in life I was getting to test a prototype part. I didn’t know what to expect from these to be honest. But I wanted to make sure that I set a baseline for what to expect in terms of accuracy from the grouphead temperature sensor. So I stuck a k-type thermometer probe on top of a tamped coffee puck with a mesh screen on top (to have the exact conditions like when I’m pulling a shot). To my surprise and relief, the grouphead sensor was within 0.5°C of the k-type probe! But that’s also where I hit a snag.

The shots were tasting alright, no doubt, but, upon looking closely at the temperature curve, shot after shot, I saw persistent behavior like this:

Figure 3: Start temperature plumetting drastically

If you notice carefully, the start temperature plummets a lot as compared to the set target, and so does the mix temperature. It later compensates and pulls it back up but it got me thinking – why’s it doing it in the first place? And then the beauty of the firmware behavior struck me. Keep in mind that this is still a prototype part, so almost all users currently use the original brass parts and the firmware is designed to deal with the thermal mass of those. What I’m about to say is pure speculation and counts as pocket science. So think of it this way – the firmware “knows” that in a stock machine the brass parts act as a heat sink. So not only will it leech heat from the incoming water, it will also act as a source of heat. Before the shot begins, the firmware doesn’t know what’s to be expected of it a few seconds into the shot. But it assumes that the group has been sitting “hot” for a while and based on that assumption it sends out cold water to help it normalize to the target temperature once the shot begins. You can kinda see it happen in figure 1, where despite the mix trying real hard to bring the puck temperature down, it has to also work to bring the temperature of brass down at the same time. Not to mention that water itself has an extremely high specific heat capacity, which is another way of saying that you have to provide a lot more energy to heat every ml of water by 1C as compared to most other materials (we’ll come back to this in a bit).

The faster you go the slower you are

Coming back to the problem at hand, now that I had a fair idea of how the firmware will usually work, I had to figure out a way to compensate for it. Luckily Ricco had a trick available up his sleeve (borrowed from Decent user Dan who came up with the clever idea in his NuSkool espresso profile). Essentially, gauge how much the temperature drops (delta) at the start on an average and add a 2 second “temperature compensation” step with the same parameters (flow rate etc.) as the preinfusion step and make it sit that delta amount higher than the target.

Figure 4: Temperature compensation step and increased actual decline

And it worked! Not only is the temperature now starting and stabilizing as expected but the decline is also noticeably higher – upto 9.5°C with the teflon parts vs. 4.5°C with brass! This is where it gets further complicated. All the heat that the brass was soaking up now stays with the water (remember its high specific heat) and in turn that now gets transferred to the puck. Those with keen eyes may have noticed that my target temperature in the shot in figure 4 is even lower than before (figures 1 & 3) and this is because I want the same amount of heat transferred to the puck as before during preinfusion and the significantly improved temp decline capabilities of the teflon parts takes care of the post-bloom part of it (win-win).

There’s some more interesting things happening under the hood. Sometimes when dialing-in, if the grind is too fine, the effect of making it coarser gives some amazing insight into flow and how it affects temperature. Figure 5 for example shows two shots of the same coffee but at different grinds with their end of blooms aligned for the sake of this discussion.

Figure 5: Temperature behavior of shots with different flow rates

If you look carefully, both shots end at the same temperature. Which, if you think about it seems a bit counter-intuitive given they’re flowing at different rates and the shot that ends later has had more time to cool. However if you think of water as a “coolant” during this phase, the more cold water in volume that you pump in, the faster it will cool.

Grouphead above water

These teflon parts apparently had another surprise wrapped (or rather drilled) inside of them. To be honest, I wasn’t expecting much more than improved temperature accuracy and behavior as a result of these parts. But when I looked at the bottomless extractions I was left a little confused. To gain more insight into what I’m about to describe, I highly recommend taking a pause and reading Jonathan Gagné’s wonderful post on how a paper filter at bottom affects a puck’s hydraulic resistance. Now assuming you somewhat understand what a donut extraction is, please also know that I’m in camp donut (i.e., I think it’s a good indication if espresso starts appearing at the edge of the basket than all at once since it indicates that probably similar amounts of water flowed through the edges of the basket as through the rest of the puck).

I almost exclusively use a filter at bottom to aid evenness of extraction. With the new parts though, donut extractions disappeared. And this concerned me. Shot after shot, spent puck bottoms were more dark in the center than at the edge, indicating that in an ironic turn of events, the edges were getting extracted more than the center.

Figure 6: Darker center of puck as compared to edges

This made me wonder if removing the filter at bottom would suit me better. Ever since Rao first made usage of a bottom filter mainstream (although it’s roots can probably be independently traced to discussions on coffee forums way back in the day), they are a certified gold method of achieving higher extraction at the cost or grinding finer. I say “cost” because a finer grind may not always be a desirable thing. However in this case because a filter at bottom evens out flow, I argue there’s benefit to that slightly finer grind. But now, with these new parts, is this even true anymore? And trusting my gut, after almost a year of using one daily, I chose to forego the bottom filter.

Figure 7: The edges now get extracted much easier

The results were quite the contrast – the puck bottom got significantly more even (there’s still a bit of increased edge extraction that I need to figure out how to mitigate), I didn’t have to grind coarser, the shots got more flavor separation with lessened astringency and there was no discernible change in EY. I was at a loss to understand why changing the grouphead parts would improve water distribution when the claimed benefit was only on temperature. And then as Joe and I were discussing this phenomenon, we noticed that the new shower holder seemed to be dispersing more water towards the edges by design.

Figure 8: Teflon vs. brass shower holders

Not only were there more holes in the holder, but they seemed to be more evenly distributed with a lot more of them going to the edge. This design in combination with the mesh screen on top of puck, has for me eliminated the need for a paper filter at bottom. It may seem like all this happened in a day, but remember that this machine’s journey began more than 6 years ago and what may seem like a culmination is just another step in the process of continued innovation.

EDIT: Seems like John Buckman had announced back in 2021 December that they indeed worked on intentionally improving water dispersion over the puck. Also turns out most of the design and testing heavylifting was done by Decent R&D engineer Ben Champion, and long-time Decent users Stéphane Ribes and Luca Costanzo.

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

One response to “DE1 Teflon Grouphead Shower Upgrade”

  1. […] 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 […]


Leave a Reply

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

You are commenting using your 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: