Are today's small batch roasters too inconsistent? - Page 3

Discuss roast levels and profiles for espresso, equipment for roasting coffee.
User avatar
another_jim
Team HB

Postby another_jim » Aug 29, 2018, 10:41 pm

Almico wrote:Simply following a BT roast profile will get you in the ball park on darker roasts, but I fear not with lighter roasts for the discerning palette. I don't see automation as the answer for consistently great coffee.


This gets to the here and now of the discussion -- third wave roasting and brewing often suck; and with some roasters and cafes, mostly suck. For the past five years or so, everyone has had the same solution: more technology, more quality control, more data, more theory. The result: the coffee sucks even more.

I'd like to make an analogy to souffles and potato gnocchi. According to most cookbooks, these are ultra unstable: soufffles will collapse, gnocchi will dissolve if you even look at them wrong. The recipes read like horror movie scripts. Guess what? It ain't so. Knead the gnocchi dough by hand, and in a few tries you know when it feels right; if it isn't, maybe because the potatoes are too young or slightly undercooked, add a hint more flour until it is right. Dip a finger in the egg white, and you'll know when it's right for a moist, non-collapsing souffle. This doesn't take a magic nose or finger, just experience and knowing what's important and what isn't

You want a good light roast; again, know what's important, sufficient development, and what isn't, everything else. Reduce the heat ahead of the first crack slowly and just right. After the first crack starts, wait at least three and half minutes. If the roast is too dark, tough; start reducing the heat a little earlier next time. If it is still light, check for bean expansion, sniff for grassy flavors. When the bean is expanded and smelling of sugars and caramels, you're set. Will this roast be consistent? No, sometimes it'll be too dark. Will it be the best ever? Obviously not. Will it not suck? Guaranteed, since you've given yourself a large sweet spot, one that's big enough to get good results without being super-roaster-robot.

So why all the crappy roasts? Too much tech. People want predetermined limits on a dozen or so theoretical variable -- BTs, time, ETs, and their differentials, at various portions in the roast The result is insanely exaggerated control inputs, and missing the forest for the trees. The roasts that suck are usually underdeveloped, heat damaged from raising the heat too fast, polymerized from dropping it too fast, and often all three. It's the result of using too much control to get all the irrelevant variables right by royally effing up the few that actually count.

What's the old saw about doing more of the same, and expecting a different result?
Jim Schulman
★ Helpful

User avatar
drgary
Team HB

Postby drgary » Aug 30, 2018, 2:35 am

another_jim wrote:You want a good light roast; again, know what's important, sufficient development, and what isn't, everything else. Reduce the heat ahead of the first crack slowly and just right. After the first crack starts, wait at least three and half minutes. If the roast is too dark, tough; start reducing the heat a little earlier next time. If it is still light, check for bean expansion, sniff for grassy flavors. When the bean is expanded and smelling of sugars and caramels, you're set. Will this roast be consistent? No, sometimes it'll be too dark. Will it be the best ever? Obviously not. Will it not suck? Guaranteed, since you've given yourself a large sweet spot, one that's big enough to get good results without being super-roaster-robot.


What? You mean this is cooking?! This is my favorite quote this year! :lol:
Gary
LMWDP#308

What I WOULD do for a good cup of coffee!

User avatar
Almico

Postby Almico » Aug 30, 2018, 7:28 am

We'll have to agree to disagree about most of this, Jim.

another_jim wrote:You want a good light roast; again, know what's important, sufficient development, and what isn't, everything else.


In my world, either everything matters, or nothing matters. We don't get a Chinese menu with some from column A and some from B. Development is highly significant, but so is everything that happens before. The way heat is applied during the dry sets up what happens during the ramp. What happens during the ramp sets the table for what compounds are available to develop. If Maillard is dragged out, development will be much different than if it's sped through.

another_jim wrote:Reduce the heat ahead of the first crack slowly and just right.


Without seeing an RoR curve, how do you know what is "just right". Different coffees behave differently at 1C. Some crash and need more heat than less. Some want to fly away and need a big heat reduction just prior to 1C. You're flying blind without an accurate measurement somewhere. You're correct in that a slowly declining RoR is best, especially at 1C, but how do you know how slowly its declining? I prefer not to guess. I do this for a living. I don't have time to guess. The bottom line is if the RoR drops too quickly, the coffee looses all "dynamics" and most of its sweetness. I've demonstrated this over and over again to my complete satisfaction.

another_jim wrote:After the first crack starts, wait at least three and half minutes.


Simply...no. Maybe on your roaster, but on mine I am well into 2C in 3:30min. I believe Rob Hoos is on to something with his theory that more early heat builds more pressure inside the bean. And things cook (develop) faster under pressure. I'm finding the less time most coffees spend in the roasters, the better. If I can fully develop a coffee in 10 minutes, so much the better than letting it stew for 14. But I need to know how fast I am adding this heat. I can't wait to see at what time 1C hits. It's too late by then. With accurate measurements, I can make adjustments on the fly and sculpt a roast to repeat a profile almost exactly.

I have a wonderful microlot Honduras. If I roast it past 2:30 it dies. No roast defects, no astringency, no nothing. It's like drinking the hole in the donut. But at 1:45-2:00 there is a marvelous juiciness of Kiwi and watermelon with a creamily satisfying body. I have not taken this coffee deep yet to see if there is a recovery after the death zone. But I will. I have a Yemen that sings around 1:30. No harshness, just bright red berry and bergamot notes.

Now would I drink these coffees as espresso? Sure, but with a bit of maple syrup to calm them down. As drip they are fantastic.

another_jim wrote:If the roast is too dark, tough; start reducing the heat a little earlier next time.


Again, no. I don't have time for guess work or play. The fun part of roasting for me is getting it right the first time. I set a goal of what I want to achieve with a particular coffee and hitting that mark is the fun part. Missing it, well, not so fun for me.

another_jim wrote:So why all the crappy roasts? Too much tech.


I think not. Maybe not knowing what to do with the tech is the problem. And maybe Mr. Hoffmann is too kind in giving most coffee roasters the benefit of the doubt. Maybe some are just pretenders that think if they follow the trend they are doing right by their coffee and their customers. Maybe ignorance is bliss and they'd rather believe their coffee is good, instead of actually making it good. The truth is, there is a lot of bad coffee out there. Too many tools is not the problem. Not knowing how to use them just might be.

For me, I'm trying to get the most I possibly can from any coffee. Once I achieve that profile, I want to be able to repeat it. My customers want this too.

I carry 10 coffees. 5 of them are blends. The other 3 are SOs that are part of my blends that also are wonderful as SO. 2 are SOs that rotate in and out with the season. I sell 3 bags of a blend to every bag of SO. Those customers want those coffees to taste the same every time. Being able to do this demonstrates competency in roasting control.

User avatar
Almico

Postby Almico » Aug 30, 2018, 8:30 am

I'll fill in my mistaken duplicate post with an example:

I play a little guitar. I can tune my guitar without any outside help whatsoever. I can hear an A440 dial tone in my head and tune my A string to it and then all the rest of the strings to that string. I get pretty close.

But if I was playing in a band with other instruments and being exactly on key meant a lot, having an electronic tuner that measured the frequency of each string precisely would be more than useful. It has nothing to do with the music I play. I can still play crap. But it's not the tuner's fault.

Ok, two examples:

I'm also a hobby photographer. In photography, exposure is akin to development in coffee. For a photo to be successful you must get the proper balance between film speed, shutter speed and aperture opening. Most modern cameras, including our phones, do all this for us. All we have to do is point and shoot. Auto profiling roasters I guess are supposed to do the same thing.

It took me a while to figure out that the Light Meter inside my high tech camera is not telling me the truth. Cameras measure the light reflecting off the subject, not the light shining on it. That is all well and good in most cases for average photos, but not all. Case in point, if you aim your camera at a black cat on a Snow White background the result will be gray snow and very dark cat. The meter in the camera is trying to average out the light it is seeing reflected off the snow. It turns the white snow to middle gray and reduces the overall exposure making the overall image too dark.

The solution is to turn all the auto features off on your camera and use it in pure manual mode. This is what we do with our manual drum roasters. The problem there is, now you are blind. How do you know how to set the aperture, shutter speed, and ISO settings? You can guess. I'm pretty good at it. I know the sunny 16 rule. Bu the correct answer, with a light meter. You can take measurements reflected off your subject, but that's no better than your camera. When you have one chance to get that perfect shot, it's far better to measure the incident light falling onto the subject than the light reflecting from it. You cannot fail that way. The snow will be white and the cat will be the same black that it is when you see it.

Our BT thermocouples are like measuring reflected light. We're measuring something, just not what we need to be measuring to ensure success. It gives us a little information, but at best not all the information and sometimes misleading information.

Mbb

Postby Mbb » Aug 30, 2018, 9:41 am

The machines are fine

Its the operators lack of ability to control all variables that leads to inconsistencies. Temp, humidity, bean moisture, heat input rate, etc.

Its art, not science.
Imperfections are tolerated, and make it better.
Like hand worked furniture......vs machine made.

User avatar
hankua

Postby hankua » Aug 30, 2018, 9:43 am

I have to agree with Alan, consistency is a huge factor in the food industry; which coffee is a part of. I can't imagine loading 30kg of green coffee first thing in the morning and "winging it". Datalogging at least gives the operator a heads up if the roast is going off track, and maybe enough time to make a course correction. Then there's automation, if it works well; why not?

Of course there's a place for the coffee shop or roastery, who follows a more artistic style of roasting; where each batch may be finished differently and people sit around a bar sipping samples.

Home roasters don't have all the challenges the pros face, there's always the next batch or new bean to try out. :D

User avatar
Almico

Postby Almico » Aug 30, 2018, 10:02 am

Mbb wrote:The machines are fine

Its the operators lack of ability to control all variables that leads to inconsistencies. Temp, humidity, bean moisture, heat input rate, etc.

Its art, not science.
Imperfections are tolerated, and make it better.
Like hand worked furniture......vs machine made.


Completely disagree. Obviously you don't roast coffee for a living. If you do, I don't want to buy your coffee.

Mbb

Postby Mbb » replying to Almico » Aug 30, 2018, 10:26 am

Nope

But as I said, the problem is the operator.

If you want mechanized short-cuts, repeatable computer profiling, to achieve repeatable quality, you are admitting you lack skill and are not an artist.

You are a simply a paid performer then. A drone.

Thats ok. Thats what pursuit of $ does.

User avatar
[creative nickname]

Postby [creative nickname] » Aug 30, 2018, 10:48 am

I am loath to get into another data vs. intuition argument about roasting. I think we have all been there, done that and can pretty much predict the rhetoric that either side will deploy.

Let me just say that there is some wisdom to be learned from both sides. Jim is absolutely right that you will make better coffee if you decide when to stop your roasts based on indicators that a data-logger cannot track. There is a good reason for this: different beans (even from the same farm, same year!) will not take up heat at exactly the same right, so development can vary batch to batch even with identical inputs and loggable outputs. But the smell in particular measures chemical changes holistically, so if you wait until the grassy sharpness has just faded away and the caramels are rising up, you will basically never drop before development is adequate. It takes practice but it will absolutely make your coffee more delicious.

On the other hand, there are other factors that influence how good your coffee tastes that your nose and eyes cannot tell you. I have done blinded experiments using my own roasts and was able to consistently taste "flicked" coffee, where the coffee starts taking up heat at an accelerating rate as first cracks are tapering off. This was for roasts finishing at the same temperature, at about the same overall time, with similar lengths through the proceeding stages of the roasts. This was a giant pain in the ass to design, but it convinced me that flicking, in and of itself, seems to produce an undersirable astringent note in the finish for coffees that my eyes and ears would tell me was perfectly good.

I do not claim to understand the chemistry that leads to that result, but it was measurable and convincing to me. Of course, if you eschew real-time data-logging you will never be able to repeat my experiment.

The takeaway is that both real-time BT measurement and tracking, and careful attention to smell, are helpful if you want to get the best flavor attainable out of a given lot of beans.

Would more accurate probes make a difference? I have no idea. The truth is that we are all trying to average away the noise of real time measurement in order to see the signal of average bean core temperature increase, which we cannot measure directly. I could imagine other modes of measurement that would reduce some error (like a temp laser measurement, which would not show fake values before the "turnaround"), but I don't really care that much about the data that early in a roast right now. And late in the roast, when I am paying attention to the data, I care about changes in its trend, not about its absolute value. But I remain open to seeing innovations in this space that prove themselves through better tasting roasts, which is what really matters. For now, that seems to be mostly a speculative notion.
LMWDP #435

User avatar
Almico

Postby Almico » Aug 30, 2018, 12:07 pm

Mbb wrote:Nope

But as I said, the problem is the operator.

If you want mechanized short-cuts, repeatable computer profiling, to achieve repeatable quality, you are admitting you lack skill and are not an artist.

You are a simply a paid performer then. A drone.

Thats ok. Thats what pursuit of $ does.


Obviously you haven't read my posts. I'm not talking about automation. I'm talking about better data gathering tools. Bean probes give very limited, if not bad data.