Fedele article on VST filter baskets, dosing, etc.

Beginner and pro baristas share tips and tricks for making espresso.
User avatar
RapidCoffee
Team HB
Posts: 4994
Joined: 18 years ago

#1: Post by RapidCoffee »

AndyS wrote:People interested in the filter basket discussion should read Vince Fedele's article on page 70 of the April/May 2011 issue of Barista Magazine (or maybe page 74 -- the magazine page numbers and the online page numbers aren't the same). For the time being, the magazine can be read online here.
This is an important article, and further analysis is in danger of being buried in the LM double basket thread (here). So I'm starting a new thread for general discussion of the many claims in the article, focusing on filter basket geometry and holes, and how they impact optimal coffee dose and grind.

Some positive first impressions:
1) This is a huge step in the right direction. Like many people, I've observed tremendous variability and poor quality control in filter baskets.
2) The hypothesis that filter baskets should be matched with specific coffee doses for optimal results is clearly stated and deserves further exploration. The corollary about dosing by volume rather than weight (for a given coffee and basket) is also interesting.
3) Matching coffee dose (and hence grind setting) to hole size is another provocative hypothesis, although IMHO more speculative.
4) Use of the refractometer as a quantitative tool for monitoring extraction is a great idea. (No, it's not the same as taste testing. But TDS is much easier to measure.)

Regrettably, I must note the article's flaws and questionable statements:
1) Barista Magazine is not a scientific journal, and this is not a scientific article. Much of it reads like an advertisement for VST.
2) The reference list is composed of studies on coffee and cholesterol, not filter baskets. AFAIK the claim that properly designed filter baskets reduce diterpenes by half (p.75, first column, first paragraph) is unsubstantiated by these references. It is not supported by any evidence in this article.
3) The graphs in Fig.3-4 are presented as though they were actual data, when it's obvious that normal distributions have been artificially generated to illustrate a point. It's not a given that the variability in typical machined filter basket holes is accurately modeled by a normal distribution. It's certainly not true of particle size.
4) Statements about hole area being "unmeasurable" (p.76, second column, last paragraph) are demonstrably incorrect. Hole number and area are readily computed by well-known image processing operations. It's true that average hole diameter may not be meaningful when there is huge variability in hole size and shape, but that's a different claim entirely.
e) The implication that histograms were developed by VST (p.76, second column, first paragraph) is absurd.
f) Statements about basket shape and hole area are presented without supporting evidence. For example, it makes sense that there will be dead spots in the extraction when the holes do not cover the entire bottom of the basket. But it does not follow that slanted basket sides will cause dead spots. Non-anecdotal evidence for these claims is indicated.

Again, let me emphasize that I regard this as important, pioneering work. My thanks to Vince Fedele for his publication, and to Andy Schecter for bringing it to our attention. I hope it will be followed up by more careful studies that either back up or refute the claims in the article. Such studies may provide a major step forward in both the theory and practice of espresso extraction.

Note: page numbers refer to the online article.
John

User avatar
cannonfodder
Team HB
Posts: 10497
Joined: 19 years ago

#2: Post by cannonfodder »

I have a variety of baskets by a variety of makers. Each one is used for different things, or rather different doses. Each basket takes a different neutral dose (filled to rim and swept level) and it an easy way to modify your dose. My particular kit requires some head space so a fill and level technique does not work well but they still make hitting my target dose easier and help to keep an even headspace as I change blends. Beyond that I have never noticed any difference in the cup quality between basket brands.
Dave Stephens

User avatar
RapidCoffee (original poster)
Team HB
Posts: 4994
Joined: 18 years ago

#3: Post by RapidCoffee (original poster) »

cannonfodder wrote:Beyond that I have never noticed any difference in the cup quality between basket brands.
I too remain skeptical of claims that espresso nirvana may be achieved through better filter baskets. But I'm open to any new ideas that are backed by persuasive evidence. For example, if TDS consistency improves with certain filter baskets, I'd be in the market. Ditto for shower screens and dispersion blocks, tamper shape, grinder burrs, doser(less) designs, etc. It doesn't have to be taste testing, which (as we all know) is very difficult to do.
cannonfodder wrote:I have a variety of baskets by a variety of makers. Each one is used for different things, or rather different doses.
Yeah, once upon a time, when I had an espresso machine with the standard 58mm basket size, I could make that claim as well. Unfortunately, the Spaz 53mm double basket is available in only one flavor. "You can have any color you want, as long as it's black." :(
John

MoJo
Posts: 5
Joined: 13 years ago

#4: Post by MoJo »

Hi John, and thanks for your comments.
1) Barista Magazine is not a scientific journal, and this is not a scientific article.
I don't see where barista MAG claims to represent itself as a scientific journal, nor do I state anywhere in the article that it's intended to be a scientific one. In the limited space available, naturally, some supporting documentation could not be included. Including what you propose would have taken tens of thousands of words, charts, graphs and summary findings, but this article was limited to 3K-5K words, and intended to address the general barista and specialty coffee community.
2) The reference list included of studies on coffee and cholesterol, not filter baskets. AFAIK the claim that properly designed filter baskets reduce diterpenes by half (p.75, first column, first paragraph) is unsubstantiated by these references. It is certainly not supported by any evidence in this article.
May I suggest you please re-read that section, I believe you mis-read or mis-interpreted it. It does not say that diterpenes are reduced by half, rather, it says that sediment levels can be reduced by about half. From Hilbrich, Daniel A.: The compounds cafestol and kahweol are present in the fine particulate sediment, in the lipid fraction (coffee oils) (both which pass through metallic filtration), and as pointed out by some of the references, have been extensively studied by medical researchers and have been conclusively shown to substantially elevate serum, cholesterol, triglycerides, and liver function tests.

That said, using the references provided, you can readily find your own supporting evidence, by following these references to others, including medical references and various patented processes on filters that provide plenty of support for the implied corollary that if the sediment can be reduced to half, then some of the compounds cafestol and kahweol present in the fine particulate sediment are also reduced.

From there, the article describes clearly how you can further measure the difference in non-dissolved solids as a fraction of total brew solids using readily available methods. Finally, it states that you can see a significant spike in the measurable particulate component when using a filter that has some significant percentage of its holes that are over-sized versus a targeted average.

There are few to no references to filter baskets to quote from. The only other reference I'm aware of is M. Petracco of Illy, who mentions non-dissolved solids are typically 10% as a percentage of TBS (total brew solids = dissolved plus non-dissolved solids), which agrees with what we measured in some typical filters. It was not the intent to provide the entire method and supporting evidence, there simply was not room, but it is easily reproducible, using the method described.


3) The graphs in Fig.3-4 are presented as though they were actual data, when it's obvious that normal distributions have been artificially generated to illustrate a point.
Thank you for pointing out that "it's obvious the distributions have been generated to illustrate a point". You are 99% correct, that is exactly what was done, and the intent was for illustrative purposes, as you surmise. The original caption said "Illustration of..." versus "Example of..." but those captions were lost in the original formatting and layout. In the original presentation of materials those graphs were excerpted from an animation that illustrated how [mean] particle size moved as a function of grind setting, and stated exactly that point, unfortunately the word "Example" replaced "Illustration" in the caption. That said, it was obvious to you, and so I hope it's obvious to other readers as well, thank you for pointing it out. The graphs were not, however, artificially created, they were normalized from actual distribution data of particle and hole size distributions, emphasizing the mean. Another point from the illustration statement that was left out of the article was that the particle size [normalized] distribution ignored the smaller bi-modal component, usually around 20-80µm in size, showing the majority components usually ranging 200-800µm in size. That said, I hope you understood the intended point of the "illustration", as it provides important background as to why you can measure some filters that produce 4-5% non-dissolved fraction of total brew solids (sediment) while other filters produce 9-12% non-dissolved fraction of TBS, whose shots can taste vegetal, and provide an unpleasant, chalky mouth feel, and contain approximately double the amount of [undesirable] sediment. It was also meant to illustrate why in some filters, users frequently have a great deal of difficulty pulling shots consistently when particle size is smaller than or significantly overlaps the mean hole size distribution, and grind settings seem extremely sensitive.

4) Statements about hole area being "unmeasurable" (p.76, second column, last paragraph) are demonstrably incorrect. Hole number and area are readily computed by well-known image processing operations. It's true that average hole diameter may not be meaningful when there is huge variability in hole size and shape, but that's a different claim entirely.
Not sure what your objection is, John. VST didn't make that statement, the metrology lab made the statement, because the apparatus to make the measurement was not known or was otherwise unavailable to perform the required task. Calls to a number of other metrology labs in the U.S. resulted in similar dead ends.

The "demonstrably incorrect" comment, I assume means that it was demonstrated that certain image processing operations could be performed, was true only after VST developed a system and methodology to do so, and to come up with meaningful ways to actually use the data as applied specifically to espresso filters. If that pre-existied, no one I could find was aware of it either inside of the filter manufacturing industry or outside, where one would normally look to find it. Your comment seems to imply that the process as applied to espresso filters has been in wide spread practice for years, but to the best of my knowledge, that is not the case, or if it has been done, no one has published it or otherwise made such a system available.

Finally, we agree, it's true that average hole area (diameter is only valid if the hole is circular) may not be meaningful if there is huge variability in hole size and shape, but that's a different claim entirely - was exactly the point of applying a histogram analysis as part of the evaluation methodology.

e) The implication that histograms were developed by VST (p.76, second column, first paragraph) is absurd.
I agree, you're right, that is absurd, but once again, I never stated VST developed the histogram on pg 76, nor elsewhere. Nor was it meant to be implied, sorry you read it that way. As Dave Walsh says, "Read, absorb, read again". That said, I'm unaware of anyone else who has ever used the tool to evaluate espresso filters. What was stated was that the histogram was one of the tools applied to the evaluation process. The histogram, combined with other measurement data became part of a grading method that allows a single number to describe several important attributes of the quality factor of a filter, the topic of an upcoming article.
f) Statements about basket shape and hole area are presented without supporting evidence. For example, it makes sense that there will be dead spots in the extraction when the holes do not cover the entire bottom of the basket. But it does not follow that slanted basket sides will cause dead spots.
Fair criticism, John. Again, there was not room in the space allowed to provide all of the data, and some data were left out intentionally. The article simply stated our observations that significant shot-to-shot variations could be measured from such filters, which anyone can easily repeat using commonly available measurement tools. Similarly, the same measurements made using several different but uniformly fabricated filters with an open area under the entire coffee puck yielded far less shot-to-shot variances.
Again, let me emphasize that I regard this as important, pioneering work. My thanks to Vince Fedele for his publication, and to Andy Schecter for bringing it to our attention. I hope it will be followed up by more careful studies that either back up or refute the claims in the article. Such studies may provide a major step forward in both the theory and practice of espresso extraction.
Appreciate the kind words, John. While I'm sure this response will do little to satisfy your scientific appetite, I appreciate your comments and perspective, and hope you understand there was neither the space nor was it the intent to provide a "scientific article" about the subject matter. Rather, it was intended to address a problem that has plagued the industry for decades, and how some steps have been taken towards addressing those problems. Accessibility was another goal, and the style of presenting the findings was to appeal to the general barista community versus a strictly scientific paper.
I too remain skeptical of claims that espresso nirvana may be achieved through better filter baskets. But I'm open to any new ideas that are backed by persuasive evidence. For example, if TDS consistency improves with certain filter baskets, I'd be in the market.
No one said nor claimed "espresso nirvana", but I like that phrase. What I hope did come out of the article was that maintaining specific design parameters and manufacturing with precision could produce filters that performed not only identically to each other, but also extracted into a desired area of the brewing control chart predictably, and would do so consistently over time, thus eliminating a host of problems that have been troublesome for the industry for the past many decades. Kind regards-Vince

User avatar
RapidCoffee (original poster)
Team HB
Posts: 4994
Joined: 18 years ago

#5: Post by RapidCoffee (original poster) »

Vince, thank you for your prompt and comprehensive response.

The worst thing that can happen to a publication is that it is ignored. That means nobody cares enough to bother. I have seldom posted about Barista Magazine articles in the past, but I consider your work important enough to initiate a discussion on H-B. I hope that others will weigh in.

The reference list includes only four publications (two of the citations are incomplete, and do not include the journal or publication date). They all focus on coffee and cholesterol, and are referenced in the sentence immediately before the statement: "A properly designed and fabricated filter reduces these by approximately half." "These" could refer to non-dissolved solids, diterpenes, or both. As author, it is up to you to make this clear. If properly designed filter baskets do reduce diterpenes, then this is an important serendipitous effect of your work, and certainly deserves further exploration.

If it would help in any way, I will gladly outline a simple procedure for measuring hole area, number, and average diameter from an image of the filter basket. As long as the holes appear as lighter regions on a darker background, this is trivial. It is only slightly more complicated to do hole sizing (basically, granulometry) and generate a histogram of hole sizes.

As a general comment, I look forward to seeing data to back up the claims in the article. For example: plots of measured hole diameter and particle size, together with TDS measurements, for a sample of faulty (or nonoptimal) vs. new VST filter baskets.

The curse of being a natural proofreader: :roll:
Either the column headings or the numbering scheme in the table immediately preceding Figure 13 are reversed for ristretto and normale extractions.
John

User avatar
erics
Supporter ★
Posts: 6302
Joined: 19 years ago

#6: Post by erics »

Well, I've got a great idea (unfortunately, its almost obvious). Vince, why not publish the technical paper on your website in pdf form, consistent with the generous amount of support documentation you already have there? http://vstapps.com/documentation/

Should you need any "hep" in sprucing up any of the images, that I can do, and will do gratis, of course.
Skål,

Eric S.
http://users.rcn.com/erics/
E-mail: erics at rcn dot com

User avatar
RapidCoffee (original poster)
Team HB
Posts: 4994
Joined: 18 years ago

#7: Post by RapidCoffee (original poster) »

I would also be glad to offer my services: brainstorming, proofreading, even running more tests (if you want to add a 53mm group to the mix).
John

jc69
Posts: 25
Joined: 14 years ago

#8: Post by jc69 »

Hi, I've read the article now several times and there's one point I simply don't get.

The claim is made that the new filters should be used in a well defined way only, i.e. the dose in the filter is to determined by the volume of the coffee grounds in the filter before tamping. Then a statement is made that, using this method, you more or less reach the same spot in the brewing control chart independently of the net capacity of the filter (by stopping at the target brew weight given by the brew formula).

That is great for consistency, but you'll obviously limit yourself to a line in the chart, at least for a given coffee.

So, we are left with two possibilities:
1.) You neglect a full taste dimension in the chart (along the lines of constant brew ratio)
2.) You don't neglect a taste dimension since it does not exist (i.e. is not discernible). This might be possible if there are other parameters (i.e. temp or pressure, but not dose) which one may use to adjust.

I can't believe in point 2.).

Regards, Jan

User avatar
RapidCoffee (original poster)
Team HB
Posts: 4994
Joined: 18 years ago

#9: Post by RapidCoffee (original poster) »

As I understand it (perhaps Vince will be kind enough to correct me if I'm wrong):

For a given diameter basket, a smaller dose needs to be ground finer to produce correct flow, and a larger dose needs to be ground coarser. For optimal results, you need smaller diameter holes for the smaller dose, and larger diameter holes for the larger dose. Too few/too small holes will cause underextraction, and too many/too large holes will cause overextraction and sediment in the cup. Thus the basket geometry, hole size, and open area must be tuned to the coffee dose. Or so the article claims.

The brewing control chart (Figure 13) shows that you can reliably produce the same extraction yield and TDS using low, medium, and high dose baskets, for normale, ristretto and lungo pours. And the taste is "virtually indistinguishable" when the extraction parameters are matched. I think that's pretty cool.

But your question is a good one: what happens if the dose is varied more than +/- 1 gram for a given basket? Can you still match the extraction yield and TDS for a given brew ratio? The article suggests not (without supporting data). And if you cannot match the extraction parameters, what happens to the taste?
John

chang00
Posts: 638
Joined: 16 years ago

#10: Post by chang00 »

Vince, thank you for the reply to the questions John raised.

What I noticed in the past few months of using weight to measure both espresso and drip coffee is, the elixir continues to lose weight on the scale within seconds. The fresher the coffee, subjectively the faster the loss. Presumably it is due to the fast loss of the carbon dioxide and various other gases, in addition to water vapor. How is the loss accounted for when TDS/TBS is calculated? For brewed coffee, sometimes I can observe a 6g loss every 200g.

It is also interesting to read the VST instruction on filter and espresso sample preparation for refraction reading: "Change filters when flow becomes slow or resistance becomes excessive. Generally, one filter is required for each measurement." Is there a comparison of accuracy of filtrate #1 vs filtrate #10?

As previously mentioned, publishing scientifically may not be your goal, and like John, who is a university professor, if you ever need another proof reader, in the medical aspect of the articles you mentioned, perhaps I can be of service. It is very rare to have several disciplines of science to converge on coffee. There was only one article I know of, published by the orthopedics department and a roaster in Washington. Guess the orthopedics residents needed a lot of coffee.

Post Reply