zaterdag 28 november 2015

Switch for line pressure booster

My local line pressure is about 2 bar but during pre-infusion, a few seconds of 3 bar line pressure is more ideal so my plumber installed a Grundfos device that quietly adds nearly one bar of line pressure to the system (water filter, pressure/flow regulator, LONDINIUM L1-P espresso machine. Tije made me a nice push button switch to temporarily power up the Grundfos. Releasing the button stops the device, so it can't be accidentally powered continuously and/or overheat.

Vimeo version:

vrijdag 27 november 2015

Designing smooth roast curves in Artisan

Smooth talk
This morning, Jan van der Weel and I were on the phone, discussing ways to ensure that the several roast profiles we want to try out are executed smoothly so we can evaluate the effect of the small differences in tasting. The toolbox offered by Artisan is very helpful here.

Rate of Rise almost touching zero at 7:00 and also barely at 8:00
First Crack Turbulence
Often, you can see a sudden change in the heat development around the onset of First Crack. At that time, the pressure inside the million or so cells of the bean has increased so much that a lot of things begin to happen. Steam, carbon monoxide and carbon dioxide begin to escape out of the weak spots in the enlarging bean cells and hundreds of compounds that will be decisive for the taste of the coffee are composed, decomposed and in some cases recomposed, each in their own pace.

The temperature probe sitting inside the mass of beans that are rolling and bouncing against it passes on as truthfully as possible what the temperature is, but in these turbulent moments around the start of First Crack, it is up to us to interpret the sudden ups and downs in degrees Celsius.

In my perception, the bean body is not getting cold all of a sudden with the splash of hot steam and gas that it begins to emanate. Just like people breaking out in a sweat in the hot air sauna are not suddenly cooling down but merely regulating their inner balance with a new process.

Rate of Rise
Still, one would want to have enough manual control or a clever enough PID system to cruise calmly through these turbulent moments.

I particularly try to keep the bean mass rising in temperature at all times, albeit at a steadily declining rate. Therefore I need to avoid touching a value of zero Rate of Rise.

The targeted Rate of Rise in the phase starting with the onset of First Crack is important then. In my observations, the Rate of Rise can momentarily drop 3ºC and if at that moment the targeted Rate of Rise is 3ºC or less, the bean mass will temporarily not heat up but stay at a level temperature or even cool down for a few seconds before the bean body temp is rising again.

That would mean that for a brief period in time, the driving force of all the development inside the bean is taken from the heat energy of the bean and not from the roaster.

The books and other sources I have available so far indicate that this should be avoided and it makes sense. First Crack is not like fireworks where you ignite the process and let things happen, stepping back.

Designer roast
So even as I try to design a roast profile that gradually diminishes the Rate of Rise toward the end, I want to have the Rate of Rise around the onset of First Crack as low as possible but high enough to remain safely above zero when the 2-3ºC are momentarily cut off of it.

Designer to the rescue: tweaking the profile plan
The Designer module of the Artisan roast program turns out to be an excellent tool to figure out what profile would best be an alternative for one that has already been tried.  You just load the profile that you wish to improve and start the Designer from the menu. It tells you that it will save the main points from the profile, in my case the Start, the Dry End, First Crack Start and Drop.

With the mouse one can freely move any of these points around and observe what effect this will have on the rest of the profile. I know from the previous roast at what temperature my Yirgacheffe beans go into First Crack (202ºC). I also know that the roast color was Tonino #99, so I can afford to roast a little darker and I can therefore raise the Drop temperature two degrees, pulling up that Drop marker on the profile. I then shift the Dry End marker to allow a smooth curve without any bumps or sagging trends and as I quit the Designer, the profile is shown as if it was an actual roast, including the projected Development Time.
Dress Rehearsal for roasting: checking the phases
Using the mouse, I then read the exact temp/time markers and enter these data points in the list of commands that Artisan will pass on for me to the PID.
Entering the wish list for the PID, Setting target Value for temperature in ºC in minutes:seconds
Then it's time to execute and there is very little for me to do but observe the "autopilot" running the planned trajectory. The red line shows the plan for the actual moment, the brown line in the background shows the complete design and the dark brown line logs the actual bean temp as measured by the robe. I only need to do a series of very small manual adjustments to the air intake of the roaster, because in the beginning, the small heavy beans need all the flow there is to get moving and as the roast procedure goes on, the beans get drier and bigger and gradually less air flow is needed. Too much airflow would blow away too much energy of the heater and the bean mass won't reach the target temperature. Also, too much airflow can diminish the aroma of the coffee.

Actual roast. Gradual decline of RoR works well, lowest is 2.5ºC at 7:30, 2.1ºC at the Drop 
Once the roast is well under way, above 170ºC, it seems I should leave the air flow as it is.

Rate of Rise as planned
In the logged roast profile above, all went very well. 250g green Yirgacheffe beans in, 216g roasted beans out at 213ºC, weight loss 13.6% and a 27.6% development time. The First Crack turbulence slowed down the Rate of Rise a little but it was 2.5ºC at the minimum there, never touching zero.
More to taste soon!

donderdag 26 november 2015

Roasting Yirgacheffe in the night

Roasting, flashlight aimed at beans in the roaster

Roast finished, spot at beans on cooling tray, fresh night air

The roast profile, weight loss, roast color, development time / percentage, Rate of Rise

woensdag 25 november 2015

Reading and Roasting: Jansen

One of the books about roasting that many roasters are hunting for is "Coffee Roasting -- Magic, Art, Science, Physical changes and chemical reactions" by Gerhard A. Jansen who was heading the R&D of Probat for a number of years at the conclusion of a long career at the company. Probat-Werke in Emmerich published the booklet in 2006.
A typical coffee bean has a million cells
Jan van der Weel owns a copy and he kindly let me have it for a few days. The seventy-one pages of the book are crammed with information so it takes more than one reading to grasp enough and to understand it all one would need a degree in food chemistry.

A coffee bean has a million cells, I learned. During the roasting process, the rising bean temperature  also heats up the water inside the bean. At normal atmospheric pressure, this water would boil at 100ºC but inside the bean, the cells more or less keep the water closed in and the cell gets in a state like a pressure cooker: the increasing pressure from the steam prevents boiling and pressure builds up further. At around 150ºC the beans that were hard like glass in their green, room temperature state, transform into a rubber-like condition. Cell walls become thinner, cells grow in size and steam and a gas mixture (carbon dioxide, carbon monoxide and others) escapes through week spots in the bean. A cracking sound can be heard.

In the many examples that Jansen discusses, the roasting time is 10 minutes maximum and the temperature is always rising, the beans keep growing darker until the end of the roast. Except for some super fast test roasts, temperatures in his examples have a steady rate of rise (unlike the roast profiles I aim for these days).

One of the details that Jansen explains peaked my interest.
"Some volatile compounds increase in quantity with an increased roasting degree, others in turn increase to a certain roasting degree and subsequently begin to decompose when the roasting degree begins to rise. Over and above this, there are volatile aromatic compounds that after a tendency to increase and decompose, are recomposed in the continued course of roasting." (page 55)
One such volatile compound, Jansen explains, is Furfural (C5H4O2), a compound of furans which produce a pleasant taste which is caramel-like to woody.
"The maximum furfural concentration is already reached at a very light roasting degree. The continuance of roasting, after a value held more or less constant, brings about a gradual decomposition of this substance."

In his illustration, one can see how furfural is composed gradually from the beginning of the roast, with a max production for a while after first crack sets in but the sweetness deteriorates fast as the temperature keeps rising and the beans grow darker than a light "filter" roast.

I am wondering if this pleasant caramel-like taste could be kept from decomposing during an extended roast after First Crack with very little Rate of Rise in the bean temperature. It could maybe keep this compound and other 'sweet' compounds intact, or allow more of these to be composed, while the acidity of other compounds is diminishing.

Jansen does not discuss this option, but what he does write keeps my hopes up.

In his conclusion, Jansen states that there is still much that we don't know about coffee.
"There are evidently still unidentified aromatic substances in the coffee bean which, although present in hardly measurable concentrations, still have a very effective say in determining the coffee aroma."
Inspired by this exposé, I again tried out the idea of a declining Rate of Rise, as advised by Scott Rao, and a somewhat longer development time with a very small ever declining RoR, much like the Natural Roast discussed by Marko Luther of the Artisan software.

I loaded 250g of Burundi beans (moisture 10.1% as measured by the Wile instrument for coffee) and I passed to Artisan these targets for my PID to manage on the roaster:

070.0ºC at 1 minute
109.0ºC at 2 minutes
142.0ºC at 3 minutes
166.0ºC at 4 minutes
184.0ºC at 5 minutes
197.0ºC at 6 minutes
206.0ºC at 7 minutes
210.5ºC at 8 minutes
215.0ºC at 9.25 minutes

The last stretch is longer than the others because I had noticed that the PID knows better how to cruise along tight margins when I don't interfere with temp points around First Crack, since that's mostly a little turbulent part of the roast anyway and while it is managing a balance there, I'd best not be shouting commands at it. In the graph below, you can see how the First Crack (between 7:00 and 7:30) does cause some waves but the Bean Temperature remains largely on coarse.

The beans smell very nice and the roast color is relatively light while still excellent for espresso. I wish I had someone near who is good at cupping tomorrow, but I will have to let the beans rest for a week, then grind on the Compak R120, brew on the Londinium L1-P and taste the espresso!

Thanks again to Jan van der Weel for letting me read this very rare and much sought after book.

I started the PID program a few seconds after starting Artisan, hence a delay 

zondag 22 november 2015

Roast Profile & Preference

Learning about coffee roasting is very adventurous. It's an old craft but much knowledge about it is anecdotal, some legendary handbooks have gone out of print ages ago while the available handbooks can be highly technical without giving any concrete pointers and the more "experts" you hear, the more it seems to become obvious that everyone is just groping in the dark, doing the best they can amidst elusive facts.

I watched an interesting video (see at the bottom of this blog). In it, Morton Münchow presents the results of a scientific research into preferred taste differences between several roast profiles using the same beans. A panel of 92 persons tasted samples (by cupping, I presume) from the coffee in a completely randomized manner. Each coffee was tasted three times but the members of the panel could never tell from any number or code what they could expect. There as an average trend of preference and there were members who agreed to that line remarkably closely, others deviated more.

I took a few screen snapshots to show what were the highlights for me.

One "reference roast" had a fairly normal roast color, a total roast time of 12 minutes, with 3 minutes development time after First Crack set in. Then there was a "light" roast with a very short development time, a "scorched" roast rushed to the same roast color as the "normal" one in just 6.5 minutes, a "dark" roast with a longer development time and higher end temp, an "underdeveloped" roast with 3 minutes development time and the same roast color as the "normal" one but a very long time before First Crack started and a "baked" variation with the same roast color and a development time of 6 minutes instead of 3 minutes, practically flatlining without any "Rate of Rise" in temperature towards the end. All done in a drum roaster (as indicated by the "dip" and "turning point" at the start of the roast, when the cool bean bass is charged into the hot roaster).

The test panel on average preferred the "normal" roast profile and to my surprise, the panel chose the "baked" one as second best with the ultra-fast roast as close third, the "underdeveloped" roast not very far behind and the "dark" and "light" ones trailing behind with the "light" one significantly failing.

To replicate the ultra-fast "scorched" profile, one needs a powerful fluid bed roaster or a small drum roaster with enough mass to accumulate lots of heat and the ability to transfer all that energy to the beans without too much harm, so I won't be looking at that much myself.

It seems to me that a "successful profile", a trajectory of time and temperature in a mass of beans that's being gently agitated, with a steady airflow blowing through the mass, could be scaled to happen in say 35% less time or even 35% more time than a fairly successful roast and still be acceptable/okay/fine. That would create a wide scope to search for the "sweet spot", the "best possible" roast with sweetness and a pleasant complexity of taste and aroma in beans that will keep up to a month after roasting.

In my view, the results of the test described in the video below suggest that looking to improve an already pleasant roast, one can safely risk "baking" by stretching the development phase while not raising the end temperature. This would comply with the advice given by Scott Rao to always have a declining Rate of Rise.

See also: -- A blog by Marko Luther about "Natural roasts."

Morton Münchow himself will probably not agree with my interpretation of his results. In his conclusion, he suggests that the "baked" profile should not have been that successful and he explains that most of his colleagues, better schooled at tasting, would agree with him that "baked" coffee yields a distinct and unwanted taste. He would prefer to "calibrate the panel" in a new test, so primarily members who agree with his preference would take part in further research.

Together with Jan van der Weel, I will look further into a declining Rate of Rise and a development phase stretching up to 25% and a little beyond that of the total roast time, in the little PID controlled fluid bead roaster that I use.

Below is a recent profile. I would be looking to extend the last phase, not raising the end temp because I aso want a slightly lighter color (Tonino 95):

Also see Jan van der Weel's blog about Joanna Alm's presentation on roast profiles.

Below is the YouTube video I referred to:

dinsdag 17 november 2015

Entree Magazine Best Barista competition -- some moments

Thirteen excellent baristas entered the competition. After a theory round, each in turn prepared an order of five different drinks for the jury to taste. In the last round, five finalists prepared and presented their coffee based 'signature' drink. Three winners were announced in the end and who of these are 3, 2 and the winner #1 will be revealed at the end of the month during a gala dinner event.

Jury were Moniek Smit, Esther Maasdam, Vincent Zwaan, Jan van der Weel and Frans Goddijn. Head jury, coordinator and organizing host was Rahwa Gebremeskel.  Technical jury: Bina Anbar and Ilithya Merien

Pictures below were the few I could take when there was time between taking notes, comparing notes, discussion. It's not a "report" about all that happened nor an implication of results.

Top view, picture by Jan van der Weel

vrijdag 6 november 2015

Line / pre-infusion pressure boost

The HX / brew head of my L1-P espresso machine is connected to the water line so the pre-infusion pressure is the same as the line pressure.

In my home, line pressure after water passes the Bestmax filter was nearly 2 bar, but the advised 'ideal' line pressure is around 3 bar.

This morning, the plumber paid me a visit and installed a line pressure booster.

It has three settings. OFF is pretty obvious. AUTO means that it starts when a flow speed of 1.5 liter per min is detected and MANUAL means that it's on permanently.

I set the switch on MANUAL, then connected the power lug to a cable with a switch on the right leg of the trolley that my Londinium espresso machine sits on, so I can switch on the booster for a few seconds during pre-infusion.

Now, the pre-infusion pressure is higher and when releasing the lever, it catches at an earlier stage in a gradual manner. I can grind a little finer as the initial brew spring pressure is higher now as well.

And it works! Nearly 3 bar (earlier it would approach 2)