Why TDS/extraction yield measurements are not a good way to evaluate taste profile/quality?

#1: Post by **samuellaw178** » Mar 11, 2019, 10:48 pm

From the Niche Zero review (and corroborated with my past experience), I notice a trend/observation which I thought would be interesting to share/discuss. I have been noticing that although the extraction yield (EY) can be similar (within the deviation of my repeated runs), the taste profile or quality can be consistently different.

You can almost confidently say that different level of extraction yields will taste different but can you say the inverse is true (same extraction yield = same taste)?

In my opinion, the refractometry measurement can give you an inkling where the process sits on the extraction yield spectrum. But I argue that there is no substitute for tasting. As subjective as our palate may be, you will still need to taste the shot to evaluate the taste. For example, for what it's worth, my Helor 101 hand grinder can achieve similar extraction level with my big conicals, but they clearly don't taste the same to me.

So I am trying to rationalize why I don't see a statistically significant difference in TDS/EY yet the tasting suggests there may be subtle differences. I think there are three main reasons:

The nature of refractometry measurement: Refractometer works by passing light through a liquid phase and depending on how much the light is scatteredrefracted/bent, it predicts the amount of dissolved solids based on what it was calibrated on. There lies the shortcoming of refractometry measurements - assuming the refractometer was calibrated properly (using the right technique and right calibration solution), it can only tell you roughly how much bulk dissolved solids you have, but it has zero information on the chemical/flavor molecules composition. It's akin to telling you how many apples you have in the promotion bag you just purchased in the supermarket clearance last night, but doesn't let you know how many of them were bad apples.

Limit of detection : Human palate have a taste sensitivity of around 5-400 ppm (that's 0.005- 0.4 mg/ml or 0.0005-0.04% TDS)*. The accuracy of of VST TDS meter starts at ± 0.05% TDS (or 0.5 mg/ml). That is just the instrument error alone and it is already higher than the taste sensitivity. To draw on another analogy, this time related to coffee, this is like trying to measure the coffee particle size using a regular ruler.

Sure you can measure some of the larger particles, but you are definitely working at the detection limit and won't be very accurate.

*https://onlinelibrary.wiley.com/doi/pdf ... .tb03598.x

Human factor: One major factor, without a doubt, is the huge variation due to human factor.

I am not running scientific experiments (nor I pretend to) and my instruments are not lab-certified or calibrated on a regular basis. But I also don't know anyone who can repeat the process over 5 times and get <0.05% TDS variation!

With all these factors combined, even if I (the operator) am a perfect scientist with absolute precision, my limit of detection is still above the tasting threshold. I will also have the issue of not knowing what exactly have I detected/measured in my sample. In practice, >0.3 TDS % (or more) variation is normal for me even when I do everything in the same way (same coffee, similar extraction time/ratio etc). So all hope is lost if you want to put a taste label to the resulted extraction yield %!

So in my opinion, TLDR (or if it got too technical): Refractometer measurement is not sensitive enough to account for subtle taste difference, and the measurement doesn't account for the chemical composition. Most of us are not trained enough (if it is at all possible to achieve zero variation) to do the measurement with enough precision.

This is not to say refractometer is useless (far from it, they can definitely be useful in some scenarios), but I think it should not be the only guideline to evaluate anything, especially not without a proper context.

That's just my way of rationalizing what I am seeing and by no means tested or proven, so take it with a big grain of salt! But I am keen to hear what you think about it. Have you done EY measurements across different grinders/machines using the same coffee and notice the same observation?

#2: Post by **TheN5OfOntario** » Mar 11, 2019, 11:30 pm

I have a refractometer and have noticed the same thing you have. I could be wrong, but I think that if coffee contains more than one soluable compound, and the mixture/ratios of these compounds all combine to form a solution that bends light a certain amount (optical refractometry) then it's possible to have have a wide variety of different coffee compounds in different ratios that when mixed together form identical brix/tds values, and hence different flavor profiles. Refractometers only tell you metaphorically how many LEGO's are in the pail, not how many of each color there are.

#3: Post by **another_jim** » Mar 12, 2019, 12:25 am

I've measured extraction yields in an FDS way, using mass only, and it didn't correlate very repeatably with the refractometer TDS measures, varying by roast levels and coffee types. I made the measurements courtesy of VST and kept the results private, communicating them only with Vince Fedele. For my troubles, I suffered an epic pile on, having my private emails posted, being called a liar, and being threatened by lawyers. So, if you're OK with similar pile on and own a refractometer, here is the recipe/equation for doing a mass only, fundamental dimensional units style, extraction measurements. Compare them to your refractometer readings and draw your own conclusions:

Narrative: The procedure uses a fixed weight of coffee and water in a dripper or pour over. The apparatus including the filter paper have to be weighed. Once the coffee is brewed, and the apparatus is completely dry, the grounds can be weighed, and the extraction is simply the original coffee weight minus the grounds weight divided by the original coffee weight. But there is a catch. The brewed coffee remaining in the apparatus redeposits the dissolved solids into the grounds as the water evaporates, so that this basic extraction estimate is slightly low. However, if the wet apparatus is weighed as well, then the amount of evaporating water is known, and the true extraction yield can calculated.

Math

Mass of water => Mw. Mass of ground coffee => Mc, Mass of extracted coffee (the unknown) => Me, Extraction = Me/Mc, TDS = Me/Mw
Weight of apparatus => Wa. Weight of Apparatus, wet, with grounds, after brewing => Ww. Weigh of apparatus and grounds after drying => Wd
Net weight wet grounds => Nw = Ww -Wa (This should also equal the total water and coffee weight minus the weight in the cup)
Net weight dry grounds => Nd = Wd- Wa
Weight of evaporated water = Nw - Nd.
Weight of redeposited coffee in grounds = (Nw - Nd) * (Me/Mw) i.e. The evaporated water times the TDS

OK. Now here's the solution: Mc - Me, the true amount of unextracted coffee = Nd - (Nw-Nd)*(Me/Mw), the weight of the dry grounds minus the weigh of the redeposited extraction. Dividing through by Mc, to get Me/Mc, the extraction percentage, and totaling everything up gets you:

Me/Mc = (1-Nd/Mc) / (1 - (Nw -Nd)/Mw).

This will work for every brewing system where you can weigh both the wet, spent coffee and dried, spent coffee brew gear. Sadly, it doesn't work with espresso machines. If you have a very gently drying oven or dehydrator, you can speed up the measurement. otherwise leave the wet gear standing overnight before reweighing.

Obviously, this is a pile of work; and I'm in no mood to repeat it. My sense, having done it once, is that the tasted softness/angularity of the brew correlates well to true extraction, better than refractometer readings. However, it is always relative to the coffee being tasted, whereas the refractometer provides somewhat error prone measures across the board.

#4: Post by **jyl** » Mar 12, 2019, 12:44 am

Suppose you take a measured weight of brewed coffee, dehydrate it, and weigh what remains. Would that also give you TDS?

#5: Post by **another_jim** » Mar 12, 2019, 12:59 am

Yes; but not extraction yield, which is the goal of the measurement.

#6: Post by **Peppersass** » Mar 12, 2019, 1:37 am

I thought we had settled this question long ago.

After many years of using a high-quality refractometer, my conclusion is that extraction yield is not indicative of taste in the cup, except in a gross way as a relative indication of where the extraction lies on the sour-bitter spectrum. That's it.

Extraction yield won't tell you if you've nailed the extraction and pulled the very best cup from the beans. It can help you find the range where sour and bitter are in balance enough to unmask the more subtle flavors, but the range is pretty broad and the refractometer won't tell you exactly where in that range the optimum taste lies. And that point definitely varies from one bean/roast to another, and usually varies from one taster to another. Again, it's a relative measurement that's most relevant for one coffee and one taster.

But, and this is important for me, even that gross measurement is a useful confirmation of the balance I'm tasting, especially after the sensitivity of my tongue has been reduced by tasting multiple extractions in the same session.

%TDS is an even more rudimentary measurement. It basically tells you how strong (concentrated) the coffee is, but not with much precision. For my typical Kalita Wave pourover, I know that if the %TDS is 1.20%, the cup is going to be weak, and if it's 1.55% it's going to be stronger, and if it's somewhere in the 1.3%-1.4% it'll be about the strength I like. Sure, it varies by coffee, but the correlation of the measurement with what I perceive as strength is so gross that it doesn't matter. Still, it's a quick confirmation that tells me if I need to change the brew ratio, grind finer or coarser, etc.

I very much doubt that %TDS or %EY measured from weighing evaporated grounds tells you anything more about taste. It's possible that method may be more accurate for determining absolute %TDS and %EY, but who cares about that? Maybe lab people. Forget trying to share recipes based on %TDS or %EY with friends. Doesn't work.

Yes, I've use the refractometer to compare %EY of the same beans from two different grinders (Monolith Flat versus Compak K10.) I did a bunch of measurements, which were very repeatable within the precision rating of the instrument, and I found that the %EY of the two machines didn't vary by more than 0.5%, which is not enough to detect by taste -- at least by me. However, the cups from the two grinders, while very much in the same ballpark quality-wise, and similarly balanced, tasted different. The difference was subtle, but it was quite evident to me and I'm not the greatest taster. I think that illustrates the limitations of %EY measurements.

I mainly use my refractometer to measure %EY of brewed specialty coffee, with which I have far less tasting experience than espresso. It helps me to quickly determine if I need to change the grind setting or contact time to get the kind of balance I like. I can do that by taste, but not as well, especially after tasting multiple roasts from my Quest M3.

I've also used the refractometer to troubleshoot equipment and evaluate results of the flow and pressure profiling mods I've made to my espresso machine. It's been very helpful for that. And I've used it to compare the solubility of a particular coffee roasted to lighter and darker levels, and sometimes I use it to see how the solubility of a particular coffee changes in the days and weeks after roast.

The refractometer is a useful tool with enough precision for the purposes I've described, and a heck of a lot faster and easier than drying coffee grounds.

#7: Post by **samuellaw178** » Mar 12, 2019, 6:28 am

another_jim wrote: Obviously, this is a pile of work; and I'm in no mood to repeat it. My sense, having done it once, is that the tasted softness/angularity of the brew correlates well to true extraction, better than refractometer readings.

Agree, for a home (not lab) setting that is more than a pile of work!

Interesting comment regarding softness/angular versus extraction yield.

I did have an interesting 'outlier' experience that contradicts that correlation

(but then it's in a completely different context). I was experimenting with cone-shaped single-baskets a while ago and found that they tend to produce what I call 'truncated' extraction (EY% goes to about 14-15% and just refuse to go much higher even with a higher brew ratio). But they taste surprisingly soft and not sour as you'd expect for under-extraction. Just an interesting observation I thought.

Peppersass wrote:I thought we had settled this question long ago.

No and yes!

It was a long ago yes, but in my admittedly short-lived memory we had not concluded that similar grinders (both conical burrs) producing similar EY can taste differently using the same coffee/machine/water/flow rate.

Now the use of EY% is even more limited than I initially thought.

Peppersass wrote:Yes, I've use the refractometer to compare %EY of the same beans from two different grinders (Monolith Flat versus Compak K10.) I did a bunch of measurements, which were very repeatable within the precision rating of the instrument, and I found that the %EY of the two machines didn't vary by more than 0.5%, which is not enough to detect by taste -- at least by me. However, the cups from the two grinders, while very much in the same ballpark quality-wise, and similarly balanced, tasted different. The difference was subtle, but it was quite evident to me and I'm not the greatest taster. I think that illustrates the limitations of %EY measurements.

That is exactly the kind of information I was hoping for. I did remember your test and I remembered the EY% did not come out very differently. But I wasn't aware that the tasting notes were different for similar EY%. Time to read the thread more closely again!

#8: Post by **OldNuc** » Mar 12, 2019, 8:29 am

The refractive index is not related to the final taste. The attempt to bound all the variables to result in it being predictive is an advanced exercise in frustration. There is a term we used at work to describe this type of exercise but I can not use it here.

#9: Post by **another_jim** » Mar 12, 2019, 10:46 am

OldNuc wrote:There is a term we used at work to describe this type of exercise but I can not use it here.

Yep, we're definitely wonking here.

Peppersass wrote: ... After many years of using a high-quality refractometer, my conclusion is that extraction yield is not indicative of taste in the cup, except in a gross way as a relative indication of where the extraction lies on the sour-bitter spectrum. That's it. ...

I very much doubt that %TDS or %EY measured from weighing evaporated grounds tells you anything more about taste. It's possible that method may be more accurate for determining absolute %TDS and %EY, but who cares about that?

Yes, I've use the refractometer to compare %EY of the same beans from two different grinders (Monolith Flat versus Compak K10.) I did a bunch of measurements, which were very repeatable within the precision rating of the instrument, and I found that the %EY of the two machines didn't vary by more than 0.5%, which is not enough to detect by taste -- at least by me. However, the cups from the two grinders, while very much in the same ballpark quality-wise, and similarly balanced, tasted different.

samuellaw178 wrote:I did have an interesting 'outlier' experience that contradicts that correlation (but then it's in a completely different context). I was experimenting with cone-shaped single-baskets a while ago and found that they tend to produce what I call 'truncated' extraction (EY% goes to about 14-15% and just refuse to go much higher even with a higher brew ratio). But they taste surprisingly soft and not sour as you'd expect for under-extraction. Just an interesting observation I thought.

This is what has me scratching my head about what exactly a refractometer is measuring, especially for espresso. If you can very consistently take any grinder and get a softer shot by grinding finer, and presumably extracting more. But the results aren't consistent across grinders and baskets. Neither softness/angularity nor acidity bitterness correlate very precisely with refractometer readings; but they are easy to manipulate with grind, dose, timing, and temperature changes when working with a specific set up.

In other words, we can work the taste much more easily and predictably than we can measure it. So what gives?

#10: Post by **OldNuc** » Mar 12, 2019, 11:08 am

Taste in the cup is not a function of the refractive index. The refractive index is a measure of the concentration of the solubles and possibly small particulates extracted from the coffee. The idea was to come up with a direct relationship between taste and refractive index, this might work for some roast levels and coffees but has proved to be less than reliable.