I have prepared 3 different coffees in the past three days, all in a French Press, following exactly the same recommendations of "dose" and "brew water" suggested by the system for a 20% extraction yield and 1.3% TDS and using exactly the same extraction time and grind setting on my Baratza encore. I then analyze my TDS% using an Atago Coffee Pal(TDS) and input it plus the beverage weight into VST tools.

The first two brews had very similar measurements (measured ext= 19.22%, measured TDS=1.25% AND measured ext= 20%, measured TDS=1.30%, respectively) and tasted amazing. The third brew also tasted amazing, but VST tools calculated a measured ext= 29.57%, measured TDS=1.91%). How is that possible? These numbers for the third brew are completely off the chart and would mean a bad tasting beverage in most cases, but it actually tasted really good and similar to the first two.

Thanks!

Remember that French Press, having a metal filter, means that your cup will not only contain dissolved solids, but it will also contain undissolved solids (suspended fines), and also lipids (coffee oils), all of this can severely alter your readings. That's why syringe filters such as the one sold by VST are essential to measure TDS in any method that contains suspended fines and lipids, an alternative could be using a Melitta or V60 paper filter to filter your coffee samples.

That's true and a good point. I find it funny those that the third result is so different, even though I used the French Press with exactly the same parameters for all three coffees. I would love to find a cheap way of filtering just a few drops of FP coffee to use with my refractometer. VST is way too expensive.

You may realize this already, and Okty implied it, but the calculation result you got was due to the higher %TDS reading for the third cup.

I agree with Okty that the most likely explanation is that you didn't filter your French Press. Of all the non-espresso brew methods, that one probably needs filtering the most.

Another common source of error is not allowing the temperature of the sample to equalize with the temperature of the refractometer. This is explained in the instructions that come with the VST refractometer, but may not be covered by Atago's instructions. Although both units use temperature compensation algorithms, these apply to a relatively narrow range of temperatures surrounding the temperature at which you calibrated (zeroed) the unit with distilled water (assuming you calibrated your unit). Despite various claims by Atago and Socratic, it's unlikely that the Coffee Pal can produce accurate readings a temperature that differs significantly from the calibration temperature. And, perhaps even more relevant to your situation, if the temperatures of two samples differ significantly, you probably won't get accurate relative measurements.

Also, you didn't supply the other key measurement, which is beverage weight. If the beverage weight varies from cup to cup, then the calculated EY will vary, too.

Other sources of variation could be the actual weight of the dose and/or brew water not matching the design. You're probably OK with a 1g resolution on the water, but the dose should be measured to the nearest 0.1g.

Another possible explanation is that you're getting variation in the %TDS reading from your Atago Coffee Pal. Although the unit Socratic evaluated had sufficient accuracy, Atago's own specifications suggest that accuracy may vary from unit to unit. Socratic never responded to questions about where they got the unit they tested. If they bought it on the open market, then there's a good chance other units will have the same accuracy. If it was supplied by the manufacturer as a review or test unit, all bets are off. Note that the Atago Coffee Pal is no longer sold in the US.

An easy way to check the accuracy of your refractometer without buying syringe filters is to do a series of careful measurements with filter brewed coffee. You should get the same %TDS measurement for multiple readings from the same cup. I wouldn't try the test with unfiltered French Press because the undissolved solids might settle between measurements. Bear in mind that this test checks repeatability, which is a key component of accuracy, but it doesn't check absolute accuracy. For that, you need lab-quality calibration solutions.

1.91 seems like a bad sample, e.g., drawn from the bottom of the cup instead of from just under the surface of a settled cup.

I honestly don't see the utility of using a syringe filter for brewed coffee. I've tested with and without various lab and coffee filter media and any differences were buried in noise - given the correct sample drawing technique. (As an aside, there are nice videos on YouTube demonstrating how to sample for the PAL, but to find them you have to expand your search to include non-English videos - I'd link, but I can't remember how to get YouTube to expand the search .)

For the difference between the refractometer's temperature and the sample's temperature, Atago makes a formed weight called the "MAGIC" (yes, in caps) to equalize temps between a sample and sensor for the PAL-series refractometers. It works amazingly well and enables transfer of the coffee directly from the cup to the refractometer without the need for an intermediate cup to cool the coffee. The MAGIC is available from Atago distributors in the US and Canada.

FWIW, each Atago (should) come with a certificate showing that particular refractometer's calibrations against a series of reference solutions.

FWIW, you can prepare reference solution yourself by using pure cane sugar (sucrose). Add 10.0g of crystaline sugar into 391.6g of distilled water (recipe from Coffeegeek forum posted by Glen & confirmed by Vince later). That should get you around 2.0% 'coffee' TDS at room temp (depending on your ingredient's purity).

Coffee stratifies, always - mix uniformly every single time before you're taking a sample. If you let coffee sit for too long in the well/prism, they will stratify as well.

Temperature matters, A LOT - ideally measurement temperature should be done at close to room temp as possible (cool in a separate room temp narrow ceramic cup). Allows sufficient time for the liquid and prism to be at equal temperature. The temp compensation algorithm doesn't know your sample's temperature (from Atago directly) and it assumes your prism temp = sample temp. In my experience, I find that the temp compensation introduces more inaccuracy than correcting it so try to measure at close to room temp as possible.

Evaporation is real, monitor/eliminate it - set up a protocol for the first time and monitor the evaporation rate of your sampling protocol (during cooling especially). Compensate if needed.

Espresso specific - If you measure espresso immediately post brewing, you get a lower reading even if the temperature is already equilibrated - this is because dissolved CO2 will mess with the apparent refraction index (this is a real effect, not due to meter inaccuracy).

Another point that I am not too sure - filtration. I think it doesn't matter that much as long as you're consistent and only compare internally. Dissolved solid will affect reading, but not super significant unless it's to compare reading with another barista. For espresso non-dissolved solids might add 0.2-0.6% TDS (roughly, probably less), no idea for brew coffee. For most it's hard to justify the cost of syringe filter for each measurement. However, since the absolute number is not important, so just stay consistent and use that as guide (as it should be).

Thank you so much for your thorough and easy-to-understand explanation. I really appreciate your time and will change some of my measurement practices in order to account for your points. I am currently calibrating my Atago with distilled water with added fluoride. Do you think that's a potential problem too?

Thanks,
Victor

Thank you very much for your answers. Great stuff.

baldheadracing wrote:1.91 seems like a bad sample, e.g., drawn from the bottom of the cup instead of from just under the surface of a settled cup.

I always get them from the very top (surface), but usually right after I am done brewing, so temp might be too hot.

baldheadracing wrote: you have to expand your search to include non-English videos - I'd link, but I can't remember how to get YouTube to expand the search .)

I would love to see those videos if you happen to find them in other languages and can post here.

baldheadracing wrote: For the difference between the refractometer's temperature and the sample's temperature, Atago makes a formed weight called the "MAGIC" (yes, in caps) to equalize temps between a sample and sensor for the PAL-series refractometers

Would it be easier and cheaper to simply draw the hot sample from the surface with a dropper and give it a minute or so to cool down to ambient temp before dropping it on the Atago?

victorlinguist wrote:Would it be easier and cheaper to simply draw the hot sample from the surface with a dropper and give it a minute or so to cool down to ambient temp before dropping it on the Atago?

Do as Tim W. does to cool, see about 21 minutes in, and 45 minutes in, the "Manual Brewing" video (although the whole video is well worth watching): Tim Wendelboe on Roasting/Pourover/Espresso/Cupping [videos + index]