what i dont get is dehydrating the pucks...

what happens if you make a puck but dont brew the espresso then you put that puck through the same dehydration process as the spent pucks.. does that un-brewed puck still weigh the same as it did before the dehydration? how much weight does it lose? i guess what i am tryin to get my head around is the fact that the coffe has some weight to lose even if it hasn't been saturated with water. i think weighing the espresso and comparing that to the weight of the coffee used makes more sense to me.

jon

Um ...

The espresso is a combination of water and coffee. When you weigh the espresso, you weigh both. The coffee in the espresso comes from the puck. So the difference in dry weight of the puck before and after brewing is the amount of coffee in the cup. The alternative method is to take the espresso shot and bake it until it is dried out and weigh that. Andy's data compares both methods to that of the refractometer. Unsurprisingly, the two weighing methods correlate more closely, but all three methods are close.

The "percent solids extraction" is the weight of coffee in the cup divided by the weight of coffee in the puck at the outset. It should be around 20% for brewed coffee and from 20% to 25% (according to Illy) for espresso. Less is underextracted, more is overextracted. Different compounds in coffee dissolve at different rates. So under and over extracted coffees taste subpar in different ways -- thin, overly sour and astringent/cutting for under extracted; dull and dark/fuzzy for over extracted

The dry weight of coffee is nearly the same as its gross weight, it loses about 1.2% of its weight during baking. When I was baking pucks, I baked one sample of the coffee being used and applied the weight change as a correction factor for the pucks used in the shots.

gscace wrote:Barry said that the refractometer didn't measure non-soluble components and that non-soluble components account for a large part of the espresso experience. Since they are obtained from the grinds, the correlation to extraction ratio isn't all that good for espresso,

MIGHT not be good for espresso.

barry wrote:MIGHT not be good for espresso.

Yo Barry! How goes it man? Eaten any fudge lately? Figured out how to iron the creases out of that tie yet?

Please say hi to June and Maddie for me!

another_jim wrote:I'm not ready to concede my main two points, which are:

-- The consistency of correlation of refractometer readings to solids concentration and extraction is not known....I'm more concerned with systematic errors due to different conversion constants for different coffees, grinders, baskets, etc.

-- The conversion that Terroir uses is unpublished. Until it is, it would be better to communicate with the most widely available units, e.g. brix for refractometers and CaCO3 concentration for conductivity meters.

Regarding your first point:

I think it's good that you're skeptical. At this point I'm a lot less skeptical than you, although I want to see more data before I'm completely won over.

The table I posted is hard to read (thanks to Dan's 700 px limit), but it contains trials using Gimme's Leftist blend (dark) and Intelly's Black Cat (medium), and includes doses of 13-17 grams. It would be good to torture test the method to try and get as wide a range of non-soluble percentages as possible.

Vince from Terroir is lining up some fairly ambitious testing sessions; I'm not sure exactly what he'll publish, but it will be interesting data, I'm sure.

I encourage you to do the math for yourself, but as stated in a previous post, I believe that the likely limits of non-soluble percentages will result in a calculated Total Brew Solids error of only plus or minus 0.5%.

Regarding your second point:
Forget about the conductivity meters. Although they won't tell you as much, the manufacturers of the meters decline to guarantee a reasonable accuracy for coffee because their own engineers know they stink in this application.

As far as the brix meters go, the analog handheld ones are cheap but take a lot of practice (and a lot of imagination!) before one can use them well. Since you already have one, for testing purposes you could multiply the brix reading by .95 or by 1.0 to get TBS and hope for the best!

Also, as you say, the coolest thing about this method is that it's so quick and easy. An easy method that's reasonably accurate is going to be a lot more useful than a super accurate but tedious method that stinks up your apartment.

another_jim wrote:-- The conversion that Terroir uses is unpublished. Until it is, it would be better to communicate with the most widely available units, e.g. brix for refractometers and CaCO3 concentration for conductivity meters.

Use what the instruments measure: index of refraction (optical angle) for refractometers and Siemens/meter (electrical conductivity) for conductivity meters. Brix and CaCO3 concentration are calibration scales based upon data correlation, and are just as "arbitrary" as the Terroir units (as well as the Agtron scale). If the correlation holds within the ranges required for testing brewed coffee and/or espresso, then those units ought to be acceptable.

AndyS wrote:Please say hi to June and Maddie for me!

facebook!


We missed you in Atlanta.

AndyS wrote:I encourage you to do the math for yourself, but as stated in a previous post, I believe that the likely limits of non-soluble percentages will result in a calculated Total Brew Solids error of only plus or minus 0.5%.

I'm interested in what sort of device folks are using for massing the samples. When we dessicated samples in chem, we used Mettler balances which were good to .001g, iirc. My guess is folks are using gram scales which read out in .1g, and are perhaps only accurate to +/- .1g. Even without errors induced by non-soluble solids (or the device itself), the TBS data is subject to around a +/- 0.5% error (I'm being less specific because I only ran one set of numbers). Given that air currents in a room can affect a .1g scale, some care in measurement needs to be taken with these samples.

Personally, I think it would be great to have an easy measurement system for espresso extraction, so I'm not throwing stones at the concept.

barry wrote:I'm interested in what sort of device folks are using for massing the samples. When we dessicated samples in chem, we used Mettler balances which were good to .001g, iirc. My guess is folks are using gram scales which read out in .1g, and are perhaps only accurate to +/- .1g. Even without errors induced by non-soluble solids (or the device itself), the TBS data is subject to around a +/- 0.5% error (I'm being less specific because I only ran one set of numbers). Given that air currents in a room can affect a .1g scale, some care in measurement needs to be taken with these samples.

Personally, I think it would be great to have an easy measurement system for espresso extraction, so I'm not throwing stones at the concept.

I'm using a .01g scale. I believe Terroir does work at .001g or .0001g.

One has to work quickly. Even with the .01g scale, a hot cup of espresso scrolls through the numbers losing weight through evaporation!

In terms of measurement accuracy on puck baking with a 0.1 gram scale: Worst case one measure is 1- (12 - 0.1)/(15 + 0.1) is 21.2% and the other is with the errors the other way around, at 18.8. Even this is likely to be better than the errors introduced by collecting wet grounds from the group head.

I of course am very much interested in using a brix meter for its simplicity. I'm just wondering if Terroir is selling water by the river, particularly your and maybe Alan the Aerobie guy's river, given that you were doing brix readings on coffee a while back.

It was difficult to get any sort of overall correlation between dosing and extraction using the baked puck method. Each grinder, basket, machine, and coffee combo produced a different slope for dose, timing, volume etc and extraction. A fast reading brix meter could be invaluable for mapping the prep variables into extraction.

That reminds me of the days when we had to mass out dry ice in analytic chemistry.

another_jim wrote:I of course am very much interested in using a brix meter for its simplicity. I'm just wondering if Terroir is selling water by the river, particularly your and maybe Alan the Aerobie guy's river, given that you were doing brix readings on coffee a while back.

It was difficult to get any sort of overall correlation between dosing and extraction using the baked puck method. Each grinder, basket, machine, and coffee combo produced a different slope for dose, timing, volume etc and extraction. A fast reading brix meter could be invaluable for mapping the prep variables into extraction.

The $30 ebay brix refract is pretty difficult to read. But with a lot of practice, the results aren't bad. You have to be really motivated, though -- many people would probably get frustrated and give up.

Alan uses the $300 Atago digital. It gives an unequivocal digital reading, which inspires confidence, even if it's not accurate! But you'd be crazy to buy the Atago now instead of waiting a little while longer for the more accurate Terroir refractometer.

There's a lot of work to be done here, and it will be difficult. But in the end I think we're going to learn a LOT.

AndyS wrote:I know there are a lot of experienced coffee people who are skeptical of this sort of numerical approach to coffee and espresso. "Screw the numbers," they snort. "I go by my taste buds."

Obviously, the numbers can't possibly replace taste buds. They can only augment them, aid in quality control, and help in the diagnosis of problems. But I do know this: I have had many mediocre coffees at shops run by these snorting, experienced skeptics. And once their "golden taste buds" walk out of the shop, employees with less developed taste buds are left in charge. The results aren't always so golden. With the proper use of objective measurements, quality control at these shops could be greatly enhanced.

I'm in full-agreement with everyone that this sort of tool can be super-useful.

My main concern with people employing quantitative analyses on espresso extractions (and often with brewed coffee) is that measuring a quantity of solubles is a far cry from measuring the quality of those solubles. In other words, nailing a 20.0% extraction only tells you how much was squeezed out of the coffee grounds. It doesn't tell you whether those solubles were the (predominantly) kind that you want, vs. the kind you don't want. In most quantitatively-measured situations, you can have a severely unbalanced extraction and still nail the numbers.

Just trying to further the discussion. What says you, Schecter? Just how augmented ARE your tastebuds?

AndyS wrote:I'm using a .01g scale. I believe Terroir does work at .001g or .0001g.

One has to work quickly. Even with the .01g scale, a hot cup of espresso scrolls through the numbers losing weight through evaporation!

What if you put on a tared cover at shot completion?

Nick wrote:measuring a quantity of solubles is a far cry from measuring the quality of those solubles. In other words, nailing a 20.0% extraction only tells you how much was squeezed out of the coffee grounds. It doesn't tell you whether those solubles were the (predominantly) kind that you want, vs. the kind you don't want. In most quantitatively-measured situations, you can have a severely unbalanced extraction and still nail the numbers.

I'm not sure that any of that is really true, Nick. The way extractions often seem to work is that the sour flavors come out early, then sweeter flavors, and then eventually bitter flavors. That's not an ironclad rule, but it seems to be a general tendency having to do with the relative solubility of the various components.

So if you don't get a high enough extraction numerically, you're probably getting a cup that's pretty sour. If you hit it right, you're getting a nice balance of sweet and sour. And if you run too far, you get dilution and a tendency towards bitterness.

If one worked at it deliberately (channeled puck, temperature way off, grind uneven, etc.), it would undoubtedly be possible to produce a bad tasting cup that measured "right." But using good technique and reasonable brew parameters, I think a fairly narrow band of extraction percentage really does correlate with flavors that most people consider nicely balanced.

cafeIKE wrote:What if you put on a tared cover at shot completion?

I'm doing that but the cover has to come off to remove samples for refractometer measurements and of course, to taste!

AndyS wrote:I'm not sure that any of that is really true, Nick.

AndyS wrote:If one worked at it deliberately (channeled puck, temperature way off, grind uneven, etc.), it would undoubtedly be possible to produce a bad tasting cup that measured "right."

A little contradiction, no?

Deliberate or not, that's sort of my point. Question is, what are the relevant thresholds? Just how "bad" can that "measured-right-cup" taste?

I'm just hypothesizing though. One of these days, I'll plunk down for the system and play with it myself.

Nick wrote: In most quantitatively-measured situations, you can have a severely unbalanced extraction and still nail the numbers.
...Just how "bad" can that "measured-right-cup" taste?

Instead of trying to dream up weird theoretical situations where the measurements fail, you need to get some actual experience with this system and note how it succeeds. IMHO.

Nick wrote:Question is, what are the relevant thresholds? Just how "bad" can that "measured-right-cup" taste?

At least as far as drip coffee goes, you could talk to Carl Staub, since he used to play with these ideas a few years ago. He'd take a batch of underextracted coffee and mix it with a batch of overextracted coffee to come up with "correct" coffee that tasted bad.

As for espresso, I think you could taste the results in a mildly channeled shot, where part is overextracted and part underextracted.

AndyS wrote:Instead of trying to dream up weird theoretical situations where the measurements fail, you need to get some actual experience with this system and note how it succeeds. IMHO.

I don't think talking about uneven extraction and its impact on a measurement system really qualifies as "weird theoretical situations". The question that I think Nick is getting to is "how does the system hold up in the real world?" instead of how it has performed under controlled testing. If Barista Jane pulls a 25-second 30ml shot which has suffered transient channeling, how will that compare in measurement to a 25-second 30ml shot which doesn't channel, recognizing that there will probably be subtle changes in dose, tamp, and/or grind between shots?