Chang94598 Senior Member Joined: 24 Oct 2007 Posts: 210 Location: SF Bay Area
Posted Sun Sep 26, 2010, 12:37pm Subject: Re: coffee refractometer
Andy, I appreciate your reply. I feel you are quite passionate about this product. Are you in anyways involved with the development of this product?
I believe the refractometer can be used to measure brewed coffee or espresso TDS, but just like any product, most likely it will have some limitations. Probably because it is a commercial product, any limitations are minimized, and may not be disclosed. I am merely raising questions about these potential limitations. If all these questions are insignificant or inconsenquential, I just would like to know how they are solved, that is all.
When I read journals, be it cohort study, randomized controlled, retrospective, prospective, etc, the auhotors generally will discuss the limitations and shortcomings of any study or procedure. It appears from your replies that this product, when used to measure coffee TDS, has no limitation and no shortcomings whatsoever.
I did not state clearly the coffee lipid amount. The coffee beans contain 5% of lipid by weight, and when brewed, a small amount is present in the drink. Some of this lipid will be filtered and be adhesed to the filter paper, and some will still be present in the drink. It is great that you think the glycerol chain does not contribute to the refractive index, and the galactomannan has no bearing on taste or flavor.
Due to the high pressure preparation of espresso, small amount of solid particles will still be present after filtering, compared to brewed coffee. Are you saying these particles don't affect refractive index? Just centrifuge your filtered sample and see what is at the bottom, and compare the supernatant to your filtered sample.
I was a beta tester for the instrument and the software, and I had input into the methodology of the espresso application. But I have no financial interest in the company (VST) or the products.
Before getting involved with Mojo, Jim Schulman and I did a project where we measured extraction yield in espresso by oven-drying the spent pucks. Our method measured total solids yield (dissolved and undissolved). But the main thing about it was how tedious it was; the refractometer is a hundred times faster and seems to be far more accurate. From time to time, Jim still complains about how the puck-drying stunk up his kitchen!
It appears from your replies that this product, when used to measure coffee TDS, has no limitation and no shortcomings whatsoever.
You are correct. The refractometer and software are perfect and have no limitations whatsoever.
Seriously, of course there are limitations. The $359 refractometer is actually a tremendous value considering its development and manufacturing costs. It's accurate enough for continuous everyday use, although hardcore professionals (roasters, commercial brewer manufacturers, etc) might want to spend $1000 or $2000 or $10,000 for more accuracy and/or durability.
More limitations: you can't handle the instrument roughly and expect it to remain in calibration. You have to follow a fairly strict experimental protocol in order to get repeatable results. Your coffee has to be evenly extracted in order for the results to be meaningful. And the most accurate extraction yield measurement in the world is only a tool towards brewing great coffee -- only a set of skilled taste buds can ultimately decide what's good and what's merde.
The coffee beans contain 5% of lipid by weight, and when brewed, a small amount is present in the drink. Some of this lipid will be filtered and be adhesed to the filter paper, and some will still be present in the drink. It is great that you think the glycerol chain does not contribute to the refractive index, and the galactomannan has no bearing on taste or flavor....Due to the high pressure preparation of espresso, small amount of solid particles will still be present after filtering, compared to brewed coffee. Are you saying these particles don't affect refractive index?
I understand the theoretical objections behind your statements. But using the refractometer, the recommended filtration technique, and the Mojo conversion software, one can get results that are extremely close to the ideal coffee extraction measurement: dehydration of the beverage itself. So, your objections don't worry me. You, however, are welcome to continue worrying about them. :-)
As you stated in previous posts, the original development of coffee extraction yield theory was done decades ago. But using hydrometers and/or conductivity meters, accurate results are difficult to obtain, causing the SCAA's "Gold Cup" program to languish in obscurity for years. I am passionate about Extractmojo because finally quick, accurate results are possible and affordable.
I did a few searches in food science journals a few minutes ago. For the refractive studies, generally, the espresso drink is treated with hexane to remove the lipid, then the aliquot is centrifuged to remove solid particles, as I had suspected. I could not find a study that compares the accuracy of refractometry, conduction, and baked dry puck.
The data obtained by VST or ExtractMojo probably can be published in a peer reviewed food science journal, and I would highly recommend it. The Gold Cup I think suffered because the original testing protocol and data are not published except in a few advertisement pamphlets, and definitely not peer reviewed. I certainly could not find the original data on the conductive meter study.
Because my memory is failing every day, I also looked up the lipid content of espresso drinks. Espresso TDS typicall contains 5% lipid by weight, per Petraco.
Since your memory is failing, you undoubtedly forgot that Petracco, in the same chapter, said the following :
Refractive index is influenced mainly by the solute concentration, and the dispersed phases are less important as indicated by the minor differences found after filtering aliquots of espresso. Commercial refractometers specially graduated for coffee are available and correlate with total solids content.
Andy, you are still not answering the question. I want to know how the company determines its accuracy. There is no doubt that refractometry can correlate TDS, but the availability of a commercial product does not prove its accuracy as you stated, unless I can see some prove. Just because you can bang your head on the wall, does not mean you should.
If you should read the beginning paragraph of the same Illy book you quoted: "The refractive index is usually determined to evaluate solute concentration in TRANSPARENT liquids. Espresso's POLYPHASIC nature makes this method difficult to apply, and scarcely convenient to predict the body character. " For this reason, hexane and centrifugation are used to prepare the sample for refractometry studies. Even with filtering, minute solid matter still remains, including fractured cells, lipid, and cellulose that will pass through the filter. A filtered sample can be used, but then the accuracy decreases, and it may not be within 0.12%. Your comments showed you had no understanding of this method's limitations. Although my memory is failing, my original first recollection was correct. The typical espresso contains about 5% lipid by weight (TDS). Didn't you mention it was 0.25% in your post?
Although you mentioned you have no financial interest, did you get paid at all for your consultation/beta work? You did not even disclose your involvement with this product until I asked, which is disingenuous. Your posts to me borders on advertisement.
Hopefully we can continue productive discussions without hubristic attitude.
FYI, I pointed Vince to this thread and here is a statement that he sent to me last night:
I am the developer of the VST Coffee and Espresso Refractometers, and the founder and President of VST. You are asking Andy how the company (VST) determines their accuracy. Andy does not work for the company, although as he has stated openly, here and on other forums in the past, he has been a beta tester of some of VSTís products, along with several others, many who work within the industry, and who regularly test software and instruments under development. Further, Andy has personally done his own work conducting dehydration testing years ago here and on other forums, and shared his results. None of the beta testers are paid for their participation in testing and providing feedback, including Andy. Nor are beta testers privy to specific R&D information that would allow them to answer such a question. Andy has been open, honest and helpful in posting on this and other forums in an effort to help educate and answer routine questions and has shared his personal experiences, not only with this, but other products from other companies. Contrary to your assertions otherwise, I see nothing wrong with that, Mr. Chang.
We have no intention of disclosing (proving) to you how VST performs the correlation for Coffee TDS. We have vetted the accuracy, methodology and warrant the instruments accuracy as well as its suitability for use as stated in the specifications we publish. Further, we have had the instruments tested and measured at two independent labs, certified in AOAC methods for dehydration testing, and also within several facilities within the industry, such as the SCAA and SCAE, both of whom have conducted their own independent tests using facilities available to them. The SCAA performed their own tests using a CEM Dehydration oven, and found that two randomly acquired VST Coffee Refractometers purchased from different date codes in 2008 and 2009 measured within Ī 0.02% TDS of their $18,000 laboratory dehydration oven, after it had been calibrated to a NIST traceable standard by a factory technician. For $359, we happen to think that is a pretty good result, especially given that previous generation TDS instruments were reading as far off as Ī 0.35% when tested against the same standard. Additional testing has been performed by the Norwegian Coffee Association at the European Coffee Brewing Center in Oslo, specifically with espresso TDS analysis. All have found the VST Coffee and or Espresso refractometers to meet or exceed stated specifications for accuracy. This may not be good enough for you, which is fine. You are certainly under no obligation to purchase any of VSTís products.
Elsewhere in your post, you state we should provide full disclosure so that an industry peer journal can confirm our accuracy statements. I am very interested in knowing exactly which peer group w/in the coffee industry you would recommend? Unlike the sugar, and other industries that rely heavily on refractometry for process control and measurement standards, there is no International Congress for the Use and Measurements of Coffee Analysis. In the mean time we have more than a thousand users world wide who have been using the coffee and espresso refractometers with great success for almost two years. I see no other company within the coffee industry sharing their IP openly, as you suggest we should, and in general note that most such IP is closely held.
The VST Coffee and Espresso refractometer instruments are designed to provide users a method and tool to reproduce with reasonably high precision easily repeatable results. We state clearly what the specifications are, and within those specifications are clearly stated the limitations on accuracy, resolution, range of temperature correction, etc.... If one is interested in a higher accuracy and or precision instrument, we provide that also, albeit at a higher cost, there are several on our web site.
There is no need to use hexane nor centrifuge espresso coffee samples. Syringe filtration is more than adequate to remove coarse suspended solids. When measuring espresso, which contains primarily, dissolved solids, as well as suspended (undissolved) solids and lipids (also undissolved but emulsified), we provide a syringe filter kit with the refractometer. The Refractometer, in fact all refractometers, will ignore suspended particles greater than the source wavelength in size, they have no effect on refractive index measurement. We specify and provide a 0.45um filter, which filters the significant majority of suspended solids (and some emulsified lipids). What remains to pass through the 0.45um filter are negligible, in terms of percentage, and do not impact [i.e., are included in] the stated accuracy. In fact, the TDS in source water and CO2 infused into solution have more of an impact than any minute remaining cellulose, lipids or suspended solids smaller in size than 0.45um. If espresso was measured unfiltered, on the other hand, suspended particles would be significant enough in quantity to settle on the prism surface at the refractive boundary, and interfere with the detector and cause erratic/false readings, however, since the espresso is filtered, this is not a problem.
Almost anything less than 0.45 um in size will be treated as a dissolved solid by the refractometer, but as a matter of principle, these would also be part of the total dry substance as determined through oven dehydration methods (of a similarly filtered sample), and therefore should be considered as part of the solids measurement by refractometry.
Finally, accuracy required for espresso is about an order of magnitude less than that required for coffee, since coffee is so dilute, nearly at the refractive index of pure water, it must be resolved to 2E-05 in refractive index. Espresso is the easier problem of the two to solve with refractometry needing resolution and accuracy to only 2E-04. So, it is coffee that is the real challenge. Regards, Vince (at) mojotogo.us
I did not ask full disclosure of the trade secrets of your product. I am suggesting perhaps you consider submitting your study and data to reputable peer reviewed scientific journals for publication. The study can be comparison of refractometry, conductive meter, and bake oven measurements of coffee TDS in espresso or brewed coffee, and CO2 and the source water TDS effect on espresso TDS measurement. The proprietary algorithrm does not have to disclosed, but standard industry material, methods and results have to be described.
Several journals come to mind. For example, Analytical Chemistry, Journal of American Chemical Society, and Journal of Agricultural and Food Chemistry, etc. It may further increase the credibility of your product. Disclosure here. I am a member of the American Chemical Society, but not in the food science section, and have no financial/commercial interest to any previously mentioned publications.
Kafeman Senior Member Joined: 18 Feb 2012 Posts: 24 Location: FL Expertise: Just starting
Posted Sat Feb 18, 2012, 1:07am Subject: Re: coffee refractometer
Hello, Newbie here, seeing this spirited conversation just would like a question which seems to belong in this thread since it covers a lot of discussion regarding precision and accuracy of refractometer measurements of coffee.
Pardon if there's an obvious answer, I just haven't picked up on it if that's the case.
Q: When the refractometer measures TDS, and the TDS reported by the Associations, could one of you kindly explain to me how the water content of the roasted bean is taken into account and what the range of it might be? For example, if the coffee beans have 4% water, 2.4% would really be 2.5% if it is calculated using the absolute dry bean weight.
Maybe the question is not framed properly, but I am concerned about understand the acvcounting for the water content of the beans, if I get a TDS and want to say, TDS=A, total water=B, TDS*total water = weight of coffee beans gone into solution. Are the standards developed for 100% dehydrated coffee? What prevents the beans from being heavier when weighed in more humid places vs,. dryer ones. After all, a 4% water bean could confuse me into knowing how much my initial weight of coffee is and that error can be carried through.
Also and related is a little help interpreting the brewer's charts and understanding if the water is accounted for the same way, how this issue fits in with the onterpretation of them, just to know exactly what it is we are fingering for the measurement.
I should note: I am talking about understanding what is done with the refractive index itself in the calculations before it enters into any software assumptions.
Thanks kindly in advance for unrefracting my (lack of) understanding of what is being measured.
Posted Sat Feb 18, 2012, 7:41am Subject: Re: coffee refractometer
Q: When the refractometer measures TDS, and the TDS reported by the Associations, could one of you kindly explain to me how the water content of the roasted bean is taken into account and what the range of it might be?
It's a good question. When Jim Schulman and I first started measuring extraction yields, we used equations that included a factor for bean moisture content (MC). The problem was, we could only guess at the MC.
A standard method for calculating MC would be to grind a given weight of beans, dehydrate them in an oven, and then reweigh. But freshly roasted beans are said to contain CO2 and volatile compounds in quantities greater than the moisture. So in your oven you are driving off all three: water, CO2 and organic volatiles. How do you distinguish between them? Perhaps spectral analysis can do it, I don't know.
Early on in the refractometer beta testing I asked Vince about this issue. He wasn't aware that any of the researchers had ever attempted to compensate for bean MC. It kind of makes sense from the practical point of view, because according to Illy, typical MC for properly stored whole roasted beans is 2-3%. Variations in this range have a fairly small effect on computed yield (see chart below).
I seem to prefer my home espresso when the beans are 4-8 days old, stored at room temp in one way valve bags, and the measured extraction yield is about 19-20%. These storage times and conditions tend to minimize the amount of calculation-distorting moisture that can be reabsorbed from the air. I suppose bean MC might be more of an issue in humid summer months, and for beans that are stored for longer periods under poor conditions.
Since it is difficult for me (and most people) to reliably detect extraction yield variations of say, 0.5% or even 1%, I ignore the theoretical measurement error (typical target suggestions for extraction yield include a +/- factor of 1% or 2%).
If you get a chance to try correlating the measurements with taste yourself, I'd be interested to hear what you find.
Kafeman Senior Member Joined: 18 Feb 2012 Posts: 24 Location: FL Expertise: Just starting
Posted Sat Feb 18, 2012, 11:47am Subject: Re: coffee refractometer
"freshly roasted beans are said to contain CO2 and volatile compounds " (sorry I don't know how to quote like the forum regulars yet)
hello Andy and thank you for the insightful answer. The above quote from your answer really summarizes the heart of this confusion. And now that I am seeing it in writing, I am wondering if coffee beans continue to generate CO2 in the days after the roasting process or if it is mainly trapped insided them (disolved, like in a bottle of cola, can probably hold a larger weight than imagining it to be just gas bubbles) gererated only during roasting and then slowly escaping.
I looked at your calculation which is a practical approach and the easy way out. I mean that in a nice way, because if the easy way out doesn't work, the hard way seems a real can of worms. So I will post again here if I have more mental pains regarding the shifting weight of beans as they pick up water, and how much is actually left right after the moment of roasting (hey - and if you roast lightly, oh crap, what if a lot more water stays in, now it's more interesting - are the charts applicable to a very light roast as well as a very dark one; I think not as the roast itself will cause reactions in the bean which vary the weight- what is rthe standard roast used in the charts now I wonder).
I'm half convinced since I feel like my understanding is still on thin ice, just a slightly different way of looking at your same calculation, say substitution in it for 14 g of beans in 230 g of coffee, the three percent variation you illustrated gives uncertainty between 21.36 and 22.02% for my cups. Now, the only reason I put the hundreths percent digit is to see the number effect, not imagine it could be measured to that accuracy. But the brewing charts are written basically for knowing the percent to the tenth or there abouts, yet I can't imagine doing better than you say, that is to the half percent at best, but more likely to the percent.
Do you know at what point the brew chart, for example the American one, are made and what "they" are doing about controlling for water content?
What I can do is another experiment I can put here although it may be old hat for you guys. Maybe you already know this well, but I don't. My house temperature is constant and I do measure the humidity. I just got a new bag of coffee roasted three days ago. I can open it and sacrifice 10.0 grams of innocent delicious beans aside as is and then just weight them daily until I get bored, to see what happens to the weight. Of course, as you've already pointed out, I could be losing volatiles and gaining water, and just my luck that they cancel out for the measurement. I have a suspicion that's not far from the truth as coffee gets stale, so I won't have learned too much if the weight doesn't change. But if it increases significantly even after losing and volatiles, that would be interesting to know.
Thanks again and this is not the last of me in this thread I hope - I appreciate that you (Andy) could help revive this thread and get me more comfortable with the limitations in (any) measurement due to basically 'sample preparation' error. The problem with writing it off as withing the experimental error, though, is that it is a systemmatic error, not a random one, so I think you were right to ask about it early on and it would be very satisfying still, to get a better handle on it.
I'm wondering btw, what the VST coffee refractometer instrument accuracy is on the raw refractive index, just to compare that to mine, specifically in the range of (specific gravities) of 1.0050 to 1.0070, what is the smallest difference it can resolve before any software or multiple point averaging or whatever innovative tricks there are to improve upon this. (*note before anyone points out, for clarification: My readout is specific gravity, didn't mean to confuse the two, though I did, but the numbers (1.3330 nd and 1.0060 SG) are close enough w/significant digits that the question isn't too different with the significant digits, I think) Really, though I am having difficulty understanding the single reading accuracy of the minimum instrument separation of measured values that will be useful here: what is that spec for the VST detector before all the intelligence/analysis improvements are considered. Just to be clear also, I recognize that this is a proprietary product, however the proprietary part, I think, does not extend to the specification of the detector, since all refractometer manufacturers freely seem to consider this part of the specification provided to customers evaluating the purchase - but I may be wrong ;-) If I have made a wrong assumption on this, at least I would like to know for a fact, if the VST instrument's native sensitivity is a trade secret too. The only reason to ask about this is because SCAA seems to have chose it as a valid standard and I wonder how my refractometer compares. All this at 20 degrees btw, or whatever the room temperature standard is specified to. Didn't want to put this whole paragrah, but there are plenty of fine points here and for the price and claimed capabilities that I don't doubt, the VST is a far more economical solution than the one I have, except, I already have mine.
Symbols: = New Posts since your last visit = No New Posts since last visit = Newest post
Forum Rules: No profanity, illegal acts or personal attacks will be tolerated in these discussion boards. No commercial posting of any nature will be tolerated; only private sales by private individuals, in the "Buy and Sell" forum. No cross posting allowed - do not post your topic to more than one forum, nor repost a topic to the same forum. Who Can Read The Forum? Anyone can read posts in these discussion boards. Who Can Post New Topics? Any registered CoffeeGeek member can post new topics. Who Can Post Replies? Any registered CoffeeGeek member can post replies. Can Photos be posted? Anyone can post photos in their new topics or replies. Who can change or delete posts? Any CoffeeGeek member can edit their own posts. Only moderators can delete posts. Probationary Period: If you are a new signup for CoffeeGeek, you cannot promote, endorse, criticise or otherwise post an unsolicited endorsement for any company, product or service in your first five postings.