Coffee Data Science
Better know a refractometer
Using a refractometer to measure total dissolved solids (TDS) in coffee is based off of some data correlating TDS and a refractive index. I’m not sure what this data was based off of, but I saw some weirdness in the TDS measurement compared to groundtruth for espresso.
This discrepancy got me thinking, what about the inside and the outside of the bean? Could the dissolved solids from each part of the bean cause a different measurement?
I first split the inside fines out by grinding coarse and sifting out everything less than 300um. Then I took the coarser grounds, put that back through the grinder at a fine setting, and sifted out less than 300um to make up the outside fines.
Then I mixed 4g of each sample with 11g of spent coffee. I pulled each shot with the same profile, and I split each shot into three cups. I focused on the first two cups (First, Second) because the third cup had such a low TDS.
For each sample, I diluted the sample a few times to measure how the TDS responds assuming the first and highest TDS reading for each sample was groundtruth. If you add water, it should dilute the total dissolved solids linearly, all else being equal.
The data didn’t quite show this linear trend with dilution. We can look at the first cup, which was the 1:1 shot. For the outside fines, the readings end up being higher than groundtruth.
We can zoom and see some differences in readings.
We can also look at the inside and outside separate but comparing the first and second half of a shot. The first half (First) over-estimates groundtruth while the second half (Second) slightly under-estimates groundtruth. The difference for outside fines is more pronounced than inside fines.
If we focus just on the seconds, the inside and outside fines behave similarly, which is somewhat expected considering that the easy to extract solubles are already gone.
The aim of this study was to suggest we don’t have a good understanding of how refractometers characterize coffee extraction. I suspect such characterization was originally done on longer brews and didn’t consider that the bean is not homogenous. When calculating the conversion from refractive index to TDS, the input of output ratio is not included, but even then, there is an assumption of homogenous extraction. Usually water temperature is considered though.
Refractometry is the current best and ubiquitous tool to quantify coffee extraction and strength, but we need better tools to be able to get a better insight into coffee.