Coffee Data Science
Hacking the Espresso Puck
Finding other ways to understand extraction
I recently got a CoffeeJack, and it doesn’t have a bottomless portafilter, so I didn’t have a way to understand how well water was flowing through the puck. Additionally, the puck was difficult to knock out, but I had the bright idea to use a straw. This could allow me to take a few samples in the center or the side, and then for each sample, I could make a few slices, add some water, and take a grounds TDS (gTDS) measurement. I then parlayed this into understanding steam pre-infusion.
Initial Experiments
I don’t have a transparent portafilter for the Decent Espresso machine, and I wanted to see how steam pre-infusion was affecting the puck. So in these tests, I stopped the shot at different times, pulled the puck, and took some images and measurements.
10 Seconds Steam PI
The puck comes out upside-down, so the top (with the grounds that aren’t wet) is from the bottom of the basket.
20 Seconds Steam PI
The water penetrated deeper, but an unevenness was formed.
30 Seconds Steam PI
The water made it to the bottom, but again, it was uneven. Further experiments showed this caused the shot to channel. It’s interesting to see the dark line indicating that the dissolved solids are being pushed to the bottom of the puck.
50 Seconds Steam PI
Water finally made it to the bottom, but there was a pattern of channeling.
With respect to channeling, it seems to start after 10 seconds of PI. This happens to coincide on the puck with there the first tamp is in a staccato tamped shot, so about half way down the puck.
Flat vs Long Pulses
We can compare 50s flat to smooth transition pulsing (10s intervals), using a fast fast transition and 0.1 ml/s flow instead of 0.2 ml/s. Both had similar patterns, and I didn’t see a particular benefit.
gTDS Metrics
For these tests, I was also cutting some core samples from the center and the side of the puck. Then I added some water and measured the TDS. The aim was to look for a trend and better understand how the center differed from the side. For each center and side core, I took a sample from the Top, Middle, and Bottom vertically in the puck.
The side measurements are usually lower. This is as expected since there is side channeling and the shower screen pushes water into the puck more from the side than the center.
We can look at center vs side, and at the top, it is certainly the case that the side had a much lower value than the center throughout. However, things get much closer at the bottom.
Overall, this was an intermediate experiment to give a look at how I try to improve my shot profile through a lot of trial and error.
If you like, follow me on Twitter, YouTube, and Instagram where I post videos of espresso shots on different machines and espresso related stuff. You can also find me on LinkedIn. You can also follow me on Medium and Subscribe.
