Coffee Data Science

Coffee Refractometer Accuracy: DiFluid R2 vs VST

Experimenting with single-device performance

By: Robert, Joe, and Jeremy

DiFluid originally came out with a small, affordable refractometer earlier this year. The accuracy by a few studies suggested the device may perform well enough for some coffee, but concerns increased as signal strength decreased in a sample (i.e. filter strength coffee). Even more troubling was the manufacturing variance across devices. DiFluid continued to iterate, and they came out with the R2 this fall.

All images by author

So more experiments were done with the R2 to compare it to the VST and Atago refractometers. The R2 is priced at $200 which is far more affordable than the VST at $700, but does the R2 pay the price in accuracy and precision?

Data was collected at Socratic Coffee, and we iterated through a few experiments to better understand the performance of the R2.

The Data

The data was collected using 5 DiFluid R2 devices, 1 VST, 1 Atago Coffee, and 1 Atago RX-5000i. Several solutions were used, each providing different insight:

1. Sucrose solution (the basis for Brix measurement; well-established normative data; a “clean” assessment of hardware)
2. Instant Coffee at Espresso Strength (high coffee soluble concentration with minimal interference from non-solubles; increased difficulty from sucrose, requiring software conversion of refractive index reading to coffee solubles)
3. Instant Coffee at Filter Strength (low coffee soluble concentration with minimal interference from non-solubles; reduced signal strength compared to instant coffee espresso but relatively low noise compared to real-world solutions as instant coffee is almost entirely coffee solubles — 99.9%
4. Espresso (real-world application at high coffee soluble concentration; a difficult testing solution with increased noise but strong signal)
5. Filter Coffee (real-world application at low coffee soluble concentration; the most difficult testing solution with decreased signal and increased noise, testing robustness of both hardware and software)

All refractometers were zeroed before each experiment. All data was collected at room temperature. A precision scale measuring to 0.001g was also used.

For this first round of experiments, we focused on single device performance. One device was chosen at the beginning of the experiments to be the R2 Official device. This was chosen at random before any data was collected. Further data on multiple device performance will be discussed at a later date to keep this discussion more concise.

Espresso Strength

DiFluid previously performed well enough in the espresso range for the typical barista. For sucrose, all three refractometers have similar performance (Figure 1).

Figure 1: Sucrose Solution

The same is true for espresso, but the groundtruth is not known (Figure 2).

Figure 2: Espresso Strength

Filter Strength Instant Coffee

Recall our previous results with filter strength samples showed the VST and Atago were slightly below groundtruth (Figure 3). We can isolate that graph to the VST and Atago. This was the difference across a few solutions with a known groundtruth.

Figure 3: Left: instant coffee for original DiFluid; Right: Zoom-In for VST and Atago, from this article

We can repeat this experiment with the DiFluid R2 Official. As you see in Figure 5, relative to groundtruth, the R2 unit appears to have less variability compared to the original DiFluid device. Interestingly, it appears to adhere closer to the groundtruth than VST and Atago.

Figure 5: VST, Atago, R2 for Instant Coffee at Filter Strength

Filter Coffee

For filter strength, we can compare the Atago and R2 to the VST assuming the VST is the most accurate, although this assumption is thrown into doubt based on the previous plot.

This test was done over 4 brews with slightly varying strengths, and a few test samples were taken from each brew. All four brews started at the same strength (15g in, 220g out), and the last two brews had 12g and 32g respectively of water added after brewing.

Figure 6: R2/Atago vs VST in Filter Coffee

We can sort these sample triplets to see how they perform, and the R2 tracks pretty well with the VST.

Figure 7: R2/Atago vs VST in Filter Coffee, sorted pairs

This data does not address one of the issues observed from the previous data — manufacturing variability. This topic will be discussed in the near future.

These findings give evidence to a few conclusions:

1. The DiFluid R2 performed well across multiple test conditions.
2. For data with groundtruth, specifically instant coffee made to filter strength, the R2 outperformed the VST with respect to groundtruth.

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.

Further readings of mine:

My Future Book

My Links

Collection of Espresso Articles

A Collection of Work and School Stories