The impact of filtering on refractometer TDS and extraction yield measurements - Page 4

Beginner and pro baristas share tips and tricks.
CwD

Postby CwD » Feb 15, 2019, 9:09 am

namelessone wrote:What I meant was, from your data (kudos for the experiment by the way - seems much better performed than Socratic's data) the *absolute* difference between centrifuged and VST filtered TDS would be significant at cupping ranges (i.e. difference of 0.05% to 0.1%). This could make QC difficult if you're comparing filtered results to centrifuged results for example.

I'm not saying that centrifuge wasn't consistent with other centrifuged results in itself.

My thoughts are that I see no reason to think absolute difference from VST will transfer over as a raw number of .03% and every reason to think it will scale with overall strength making the absolute deviation too small to pick up on a display that only goes to .01% on the samples that are 5-8 times weaker.

But I would like to at least do a brief test of 3-5 trials to confirm or debunk that thought before using the centrifuge in non espresso ranges. And then a longer test with as many trials as this last one eventually (unless the short test shows it to be entirely unviable in every trial). I could even be wrong and it might perform worse than with espresso as the lower amount of sediment can't condense into a solid layer in the tube as well or something. I'm curious how it will turn out, and I'll definitely share the results!

User avatar
AssafL

Postby AssafL » Feb 15, 2019, 2:20 pm

CwD wrote:...Then we can extrapolate from that how much of the dissolvable material in the puck made it to the coffee and how much was probably left behind. We don't know how much non dissolvable material made it, so we compare the coffees based on what we do know. I'll argue here that "extraction yield" is actually extremely appropriate here, because we are measuring what's extracted from the coffee. The sediment does contribute to experience, but it's not really extracted as much as just moved.


I agree - but Doesn't it stand to reason then, if you are measuring the extracted vs "moved" yield that it would be more accurate to use (as dose) only the extractable dose?

Wouldn't that be better to put into coffee tools than some arbitrarily (even if accurately measured in 0.1gr) "extractable coffee" + "moved coffee"?

Why does the "moved coffee" value not make the EY calculation wrong? (Apart from "that is what we have so be it" argument which is okay but doesn't mean the numbers make sense).
Scraping away (slowly) at the tyranny of biases and dogma.

jpender

Postby jpender » Feb 15, 2019, 3:31 pm

AssafL wrote:Why does the "moved coffee" value not make the EY calculation wrong? (Apart from "that is what we have so be it" argument which is okay but doesn't mean the numbers make sense).


It's a definition. You want to say extraction means everything and that's fine as an alternate definition but it's different than saying that extraction is only what has entered solution. Make a case that your definition is superior in a practical sense, not some vague aesthetic sense.

CwD

Postby CwD » Feb 15, 2019, 4:51 pm

AssafL wrote:I agree - but Doesn't it stand to reason then, if you are measuring the extracted vs "moved" yield that it would be more accurate to use (as dose) only the extractable dose?

Wouldn't that be better to put into coffee tools than some arbitrarily (even if accurately measured in 0.1gr) "extractable coffee" + "moved coffee"?

Why does the "moved coffee" value not make the EY calculation wrong? (Apart from "that is what we have so be it" argument which is okay but doesn't mean the numbers make sense).


No, because it's well understood a 100% extraction yield is impossible and we can't get above about 30% or so. We don't need to adjust the dose because we can already tell where we are in relationship to the cap.

Moved coffee being included brings that to a hypothetical 100%. And makes Turkish 100% even if you forgot to heat it up and extract anything. Moved coffee would be nice to know, as there are effects on the cup. But it's a different thing that isn't extraction and effectssthe cup in different ways than extracted compounds (drink Turkish coffee versus the same amount of grounds poured in cold water if you don't believe me, same total dissolved + undissolved solids, but in the Turkish much more have been dissolved) so shouldn't be in extraction yield even if known.

samuellaw178
Team HB

Postby samuellaw178 » Feb 15, 2019, 5:26 pm

The term 'extraction yield' makes sense if you look at the espresso extraction process as a solid-liquid extraction process, meaning the goal is to leech soluble solids into the aqueous/water phase. So the yield of the process is defined by how much soluble solids you've successfully transferred into the aqueous phase.

However the reality, as all things in life, is a little more complex than that.

Espresso is fundamentally an emulsion/suspension : it contains emulsified oil droplets, insoluble solids, and finally the soluble solids in aqueous phase.

How espresso really looks (under microscope):
Image
If you're interested, there are more pictures here Chasing coffee micro-fines (<10 µm) [Video & Pics & Updated]

As discussed by others, refractometer is only measuring the latter part - the soluble components in aqueous phase. Its accuracy will be affected negatively (to varying extent) by the other two components (oil droplets and insoluble solids).

Filtering the espresso will get rid of the insoluble solids, but more importantly, the oil droplets (which I suspect is the main culprit, in relative to the insoluble solids, for the variation in filtered vs non-filtered samples).

In comparison, this is how the espresso looked after 0.22um filtration:

Image

Oil and oil solubles definitely contain flavonoids, perhaps a significant portion of them. You will know/experience this if you're a fat eater like I am (i.e. the fatty part of bacons/meats is the best part, as it contains most of the flavour bombs). :lol:

So measuring the soluble solids is definitely useful to gain some insights into the espresso extraction. However, it is not the be-all-end-all measurement. But unfortunately that is currently the best we have and what we can measure relatively easily right now (before someone develops a pocket chromatography/HPLC that can also analyze the lipids profiles as well as the insoluble solids :lol: ).


From my own experience, the results I got from measuring unfiltered espresso is consistent enough to be useful, provided it is under one controlled environment (same coffee, same machine etc). I wouldn't trust it enough to compare my values directly against others' data carried out using different setup/environment. To draw conclusion from my own experiments using the same coffee? Sure!
★★ Quite Helpful

nuketopia

Postby nuketopia » Feb 15, 2019, 11:59 pm

The point of filtering is not "accuracy".

The point is repeatability. The refractometer can't measure particulates. But particulates can affect the instrument in different ways.

The point is to use the instrument to objectively measure the process and whether the process is consistent and whether changes in process are producing the inteneded changes in extraction.

Yes, we drink particulates, oils and other things in the beverage that the filter removes. But we want to remove things that may affect the instrument's usefulness as a process control. We really only want dissolved material in the sample because that's what the instrument measures by virtue of how that changes the liquid's refractive index.

We don't want stuff in there that may interfere with the measurement. That's what the filter (or perhaps centrifuge) accomplishes.

CwD

Postby CwD » Feb 16, 2019, 12:17 am

It is also accuracy, though. The filters or centrifuging don't just help to make measurements more repeatable, but to make them more accurately reflect the real tds. You can't inaccurately measure something you're not measuring in the first place. Refractometers do not measure suspended solids or lipids in the first place. If you only feed it the thing it's actually supposed to measure, it gives you a very accurate number for what that is. And we can see by the comparison to the centrifuge not only that centrifuging is equally effective, but also that the VST filters do not remove a substantial amount of what is actually supposed to be measured. They work as a check on each other to prove that filtering in no way makes a reading less accurate to reality when you realize that reality applies to what it's actually measuring, real total dissolved solids, and not total "stuff" in drink, which it makes no attempt to measure.

Furthermore, even if an instrument existed that could measure the total amount of stuff in coffee (which refractometers can't and using unfiltered samples does not do), tds alone is by far more useful than tds and suspended solids or tds and lipids or all three if it's only given in one number. Knowing all three separately is more valuable than tds alone, but tds tells you by far the most about the quality of extraction, and following from that the taste. Total everything also breaks down entirely for methods with no filtering or decanting away from grounds like Turkish where total everything is always 100% even if no extraction is done.

jpender

Postby jpender » Feb 16, 2019, 2:55 am

CwD wrote:Furthermore, even if an instrument existed that could measure the total amount of stuff in coffee (which refractometers can't and using unfiltered samples does not do)...


Even if? Of course such instruments exist.


CwD wrote:...tds alone is by far more useful than tds and suspended solids or tds and lipids or all three if it's only given in one number.


That's the standard dogma. Is it supported by evidence?

User avatar
AssafL

Postby AssafL » Feb 16, 2019, 3:51 am

This reminds me of audio. When HiFi fans were looking for a qualitative measure they adopted one measure (actually 2 measures: frequency response and THD).

Now this did not really reflect what audio engineers did (IR, impedance curves, spectral decays, horizontal and vertical curves, etc).

So then the goldenfish ears were like Frequency response isn't enough. True enough. But they couldn't be bothered with the math that loudspeaker design and amplifier design requires. So they ended up increasing transistor biases which increased distortion and sounded nice with chamber music... and dipping power cords in nitrogen and adding capacitors to RCA cables.

The moral of the story is that while I agree with Mitch that TDS is a perfect measurement for many coffees (I think double helix made the point of using instant coffee to calibrate a refractometers conversion - instant coffee being 100% dissolved with almost no lipids or "move coffee") as one moves to more esoteric extractions that relationship suffers.

I agree with Samuellaw that lipids are tasty. Just add a spoonful of butter to anything. Soups. Meats, pasta, cake, even tea as the Mongols prefer... and it is a rather big improvement in flavor.

So I'd prefer to think of it as if :
Espresso = (Instant Coffee) + ("whatever makes it espresso and not Nescafé")
So that VST and Atago (after filtering) measure the quality of the Nescafé. Not it's other organoleptic qualities. Which I personally can't discount because I prefer espresso to other coffees always. I like my flavors big fat and heady. And Nescafé doesn't quite cut it...
Scraping away (slowly) at the tyranny of biases and dogma.

CwD

Postby CwD » Feb 16, 2019, 4:32 am

jpender wrote:Even if? Of course such instruments exist.

If the instruments existed in a form outside of lab equipment that's a zero or two more expensive than the highest end grinders on here. I'm aware of trying ovens to measure TDS + suspended solids, potentially in two separate samples, one filtered and one unfiltered, so that you could know both things. Although I'm unaware of anything to measure the lipid content, but I'm sure there's some laboratory stuff out there.

jpender wrote:That's the standard dogma. Is it supported by evidence?

It is the only one of the elements that changes in necessarily track changes in extraction. And contribute the absolute bulk of what espresso is. You can easily try suspended solids only, lipids only, and dissolved solids only yourself.

Suspended solids are an absurdly nonsensical way to track extraction for taste, just try pouring a bunch of ground coffee in cold water. 100% extraction yield off of a measure including suspended solids, taste like gritty water. Probably provide some texture, which is important, but they don't have much ability to be a major player taste wise.

Can get some coffee oils by separating them with centrifuging, taste awful. They're not butter. And it's a very small percentage of the drink. They contribute something to the mix, but it's not the bulk of the coffee.

And can get very low oil espresso with a paper filter in the basket or by tasting the centrifuged espresso (don't taste VST filtered espresso, it has a bizarre chemical taste from the filter and can contain tiny shards of glass), it's a bit weird without the other stuff there, but it's the only component that's recognizable as espresso by itself. And the absolute bulk of the drink.

AssafL wrote:So I'd prefer to think of it as if :
Espresso = (Instant Coffee) + ("whatever makes it espresso and not Nescafé")
So that VST and Atago (after filtering) measure the quality of the Nescafé. Not it's other organoleptic qualities. Which I personally can't discount because I prefer espresso to other coffees always. I like my flavors big fat and heady. And Nescafé doesn't quite cut it...


There are several problems with this. First of all, this assumes that instant coffee is using equal quality coffee, brewing it perfectly, and there is zero loss of quality in any of the flavor compounds during the instantizing process (I don't think anyone says you can't absolutely ruin coffee after it's been extracted). And VST and Atago before filtering are still just measuring TDS, they're just getting it wrong. Neither of them is measuring the other parts of the drink. They fundamentally can't.

And espresso is going to be anywhere from ~4 to ~15 times the TDS of other coffees. It's dramatically different from other coffees even without the oils and sediment. Seriously, metal filter Aeropress and French Press have those too and they're not espresso. Turkish has those in far greater quantities than espresso (I've done centrifuging of Turkish and espresso, the espresso has a little dot of sediment at the bottom of the tube and Turkish it's like half the tube) and it's still very much not espresso. Those are an important part of espresso, but they don't make it espresso nearly as much as the high TDS does.