woensdag 14 juni 2017

The Four Daltons

COMPAK models R120, E8, E6 and E5 lined up next to LONDINIUM L-R
First impression, notes after a visit from Roemer Overdiep:

Roemer came over to taste espressos from all 4 grinders. We made sure to every time have 18g in the 18g VST basket, using the distribution tool, getting about 30g of espresso in about 30s. R120 was best, E8 ‘redspeed’ almost as good, the E6 and E5 were somewhat similar in that the espresso was excellent but in comparison with E8 / R120 it has more ‘sharpness’ — the R120/E8 were softer, more subtle. If someone has excellent beans, an excellent espresso can be prepared with the E5 and only if one also has the bigger E8 nearby would one notice there might be something extra in buying the bigger grinder. Difference between E5 and E6 seems minimal.

These tests were done with a first roast of Kenya beans. We poured the beans from the hopper of one grinder to the other while testing and gradually the hippers got emptier.

Data about these 4 different sizes grinders:

E5
550W
6.1 kg
58mm steel burrs (good for 250kg of grinding)
HxWxD in mm: 495x160x375 (800gr hopper)
Price indication € 800

E6
700W
11.58kg
64mm burrs (steel: 300kg, Red Speed Lucidate: 2.250kg)
HxWxD in mm: 635x215x400
Price indication € 1.150

E8
700W
13kg
83mm burrs (steel 800kg,  Red Speed Lucidate: 5.000kg)
HxWxD in mm: 635x215x400
Price indication € 1.450

R120
1000W
37kg
120mm burrs (12.000kg)
HxWxD in mm: 775x230x380
Price indication € 2.250

In order to have more consistency over a larger mass of beans, I added batches of these roasts of the same Kenya beans together, mixing them for 15 minutes in a cool drum of the FZ94 roaster:

Batch #3, Tonino 84, 615g roasted 9 June
Batch #4, Tonino 99, 630g roasted 9 June
Batch #5, Tonino 96, 1000g roasted 13 June

More about this setup soon.


Grinds, not redistributed
Grinds, distributed

Refractometering
Update 25 June A
Using the refractometer and the centrifuge, I tested some extractions from the E6 grinder:
17.6g grinds, 30g espresso in 28 seconds, TDS 10.53 -> 18.47% Extraction
17.9g grinds, 30g espresso in 29 seconds, TDS 10.64 -> 18.75% Extraction
17.9g grinds, 26.3g espresso in 29 seconds, TDS 9.48 -> 20.2% Extraction

Pulling a larger volume of espresso easily gets the EXT at the 20% mark but for the taste I prefer the slightly shorter pulls with a little higher TDS and more mouthfeel. So for testing I would prefer about 18g grinds, about 30g espresso in about 30 seconds.

Update 25 June B
Tije de Jong visited to do some blind tasting for me:

4 identical cups, marked on the bottom

Setup (no refractometering in this sesssion by the way)

Extractions from 4 different grinders, all grinds stirred with te Distribution Tool

Tije tasting and ordering cups along his preference, cat sniffing

The two on the left are Tije's favourites with slight preference for the far left, the ones on the right he found more acidic / sharp even though the differences were all very slight.
E5: 18g of grinds, 30g espresso in 30 seconds
E6: 18g of grinds, 30g espresso in 25 seconds
E8: 18g of grinds, 30g espresso in 34 seconds
R120: 18g of grinds, 31g espresso in 30 seconds

Preference:
1= E6
2= E5  (1 and 2 more mild and a wider spectrum of tastes)
3= R120
4= E8  (3 and 4 more acidic / sharp)

We repeated this test:
E5: 18g of grinds, 30g espresso in 36 seconds
E6: 18g of grinds, 30g espresso in 28 seconds
E8: 18g of grinds, 31g espresso in 30 seconds
R120: 18g of grinds, 31g espresso in 24 seconds

Preference:
1= E5
2= E8
3= R120
4= E6

It seems the most expensive grinder does not automatically guarantee the highest score and the E5 was on first and second place. Repeat tests must show if there is a pattern here.

Finally we pulled two espressos with 18g grinds out of the E5, one just shaken to level the grinds, one stirred with the distribution tool.
Non Distributed: 18g of grinds, 30g of espresso in 37 seconds
Distributed: 18g of grinds, 30g of espresso in 33 seconds (more even flow)

Preference:
1) Distributed
2) Non Distributed




Update 26 juni A
Jan van der Weel came over to do the first of a set of two blind tasting sessions.
E5: 18g in, 30g out in 33 seconds
E6: 18g in, 30g out in 29 seconds
E8: 18g in, 30g out in 26 seconds
R120: 18g in, 30g out in 35 seconds

Jan van der Weel tasting
Jan found the taste of the espressos very close together, but after much tasting, spitting, mouth rinsing with water and more tasting, his preference turned out to be:

1: R120
2: E8
3: E6
4: E5

We will do this again soon to see if this result will be repeated.

Next, we did 30 extractions from grinds from the E5, on and off using and not using the Londinium Distribution Tool. So #1 not stirred and only shaken, #2 stirred and shaken, etc. We noted the flow (even/uneven), the timing, and measured the Extraction using the centrifuge and VST refractometer.

Over time, the extractions went faster. This could be caused by the grinder warming up with intensive use or the grinder being close to the Londinium (has been there since it arrived). Also the beans ran out towards the end and we barely made the last shots with the last beans. We would have preferred a near full hopper all through the set.

Then, during the centrifuge session at 3500 rpm, 5 of the test tubes broke with the centrifugal force so those samples were lost. We threw out all similar tubes to prevent this happening again.

Jan has all the data of this set and he will join in with his results and thoughts.

Jan numbering the test tubes

Test tubes

Assorted test tubes

Syringes and pipettes

Ready for the centrifuge

Oops. That is why the centrifuge made such violent noise today.
Syringes (used sucking in/out coffee to mix before taking a sample to measure TDS) ready to be washed, dried and used again
On Youtube, Anthony Skrivan writes: "I recently purchased the Compak E5 to pair with my Linea Mini and I couldn't be happier with the set up. For its smallish size it's a beast and capable of extreme consistency as well. Love it."

Update 29 June: Grind speed
Currently the E8 and E5 are both equally fast, and the E6 almost as fast in delivering 18g of a grind setting that gets 30g espresso in 30s

4.8 seconds for E5 and E8 and the E6 just a little over 5s

I keep doing tiny adjustments after every shot so it’s a moving target but that is the trend.

Quite remarkable speed in the much smaller E5 and at no cost to quality because so far it turns out to be difficult to detect the difference in a blind tasting.

Update 3 July 2017
Roemer Overdiep (from culinary Instagram https://www.instagram.com/roemer.has.it/) and his wife Barbara visited for a blind tasting session today.

Roemer posted this photo on his Instagram:

For the tasting session we used Kenya beans roasted 26 June at Tonino 103, a little on the light side for me which allowed for a somewhat dry aftertaste with the full bodied spicy cup. The Londinium Distribution Tool was used on all doses.

I used 18.5g in the VST 18g basket for all 4 extractions.
E5 gave 30g espresso in 35s
E6 gave 31g espresso in 31s
E8 gave 31g espresso in 30.5s
R120 gave 30g espresso in 28s

If an extraction came out with a different volume/time, I ground another dose and tried once more until they were all 'in the ballpark'.

We shifted the 4 cups around until we had no clue which was which anymore before starting the tasting (labels of which grinder was used were stuck to the bottom of the identical cups). Each cup was on a piece of paper labeled A, B, C, D so we could each write down, without discussing it, which order we preferred. Then we shared these lists and looked under the cups to see the grinder that was used.

Again the differences were small, with the taste spectrum showing the spicy quality on all 4 cups with  a little more sweetness in one, a more mild balanced taste in the other and the dryness prevalent in the next, depending on who tasted which cup.

Order of preference:

Roemer: R120 - E8 - E6 - E5
Barbara: R120 - E6 - E8- E5
Frans: R120 - E8 - E5 - E6

So here the R120 ranked highest, obviously, followed by E8 and the E5 mostly on fourth place.

Then we shifted the cups again, also their paper 'coasters' and tasted once more.

Order of preference:

Roemer: R120 - (E8 and E5) - E6
Barbara: E5 - R120 - E8 - E6
Frans: R120 - E6 - E8 - E5

While R120 is mostly scoring best here, the E5 extraction was doing well on the more cooled down tasting with E6 and E8 alternating.

We plan to repeat this soon, then using a more average blend from a nearby specialty coffee place.



Update July 5: Jan's analysis

The full text with links is on Jan's blog:
http://www.byleew.nl/experiment/measuring-effect-londinium-distribution-tool/

Measuring the effect of the Londinium Distribution Tool

JULY 5, 2017

The recently launched Londinium Espresso Distribution Tool promises to reward you with a very high level of consistency to your espresso extractions. What is the effect of using the LDT? Does an espresso made with this tool taste better? Frans is doing a series of experiments of the tool in combination with different Compak Grinders. Blind tasting is a great way to measure the effect of the LDT but a downside of this method is that it’s very time consuming. One would need many samples tasted and compared. Also this method is not completely objective. The selection of the espresso tasters would influence the result of the experiment.

Frans and I wanted to get more objective data about the effect of the LDT. What does it do to the neatness of the flow (even/uneven), timing and extraction yield? To answer these questions we did 30 extractions from grinds from the Compak E5, on and off using and not using the LDT. So #1 not stirred and only shaken, #2 stirred and shaken, etc. We observed and noted the flow (even/uneven), the timing, and measured the Extraction using the centrifuge and VST refractometer.

All the espresso’s for this experiment were brewed using the Londinium R lever machine, an 18 grams VST Precision Filter Basket and the Londinium Dosing Funnel. For every shot we dosed exactly 18 grams to brew an espresso of 30 grams.

Results

Our experiment wasn’t without problems or errors. First problem was that the extractions towards the end of the experiment got increasingly shorter in time. An explanation could be that the grinder behaves differently when it heats up during use. This made our samples less uniform than intended. Also the high speed centrifuge (3500 rpm) did make more noise than usually. This turned out to have been caused by the breakage of 5 test tubes which were of a lesser quality than the standard glass tubes we also used. We need to use these learnings to improve in case we decide to do a next experiment.

Regarding extraction yield: there were some measured differences but statistically these may have been likely caused by chance.

The LDT did give us more consistent extraction percentage. This means that there was a narrower variation in extraction yield using the LDT

The LDT also gave a significantly higher number of smooth, even flowing extractions. Especially the latter is often an indicator of a better tasting cup

The more ‘sloppy’ extractions without LDT did show some higher extraction numbers but, as said above, not in a statistically relevant manner so we would need to test more if we’d want to pursue this further.


Below you can see the effect of the LDT on the neatness on the flow. Espressos made with the LDT had about 80 percent nice and balanced extractions. 27 Percent of the merely-shaken grinds had nice looking even extractions. That means that by using the Londinium Distribution tool we tripled ‘even and nice looking extractions’.



A statistical test showed us that this result is significant (p < 0.05). That means that it’s unlikely that a difference between the groups was caused by chance. Below you can see the result of the statistical test (Pearson’s Chi-squared) that was calculated with R Commander in RStudio.



Conclusion

The LDT will give you a significantly higher number of smooth, even flowing extractions. Especially the latter is often an indicator of a better tasting cup. Regarding coffee extraction yield we did not find significant results. We would need to test more if we’d want to pursue this further.


Update 5 July II:

Pulling a 40g shot of high extraction quality espresso using 25g of grinds from the small COMPAK E5 grinder, the VSTlabs 25g precision basket, the Londinium distribution tool and the Londinium L-R espresso machine. This new 25g basket may require a specially modified naked portafilter (mine was slightly widened by Tije de Jong in Amsterdam). It enables one to pull a large volume espresso with a tremendous mouthfeel. If you normally start the day with two shots, this one will send you off for the day by itself. After preparing one with the huge R120 grinder, I tried one with the much much smaller E5 grinder and it performed very well. A blast of taste, not due to under-extraction but very nice and balanced.

This E5 can really pull its weight, plus more.

Small grinder Mammoth shot from Frans Goddijn on Vimeo.


"Mammoth shot from small footprint grinder" II from Frans Goddijn on Vimeo.

Update 6 July 2017, a conversation on Youtube:

3 opmerkingen:

Unknown zei

Did you use the centrifuge to separate insolubles for more accurate pre-filtration tds measurements? Or were they used to remove the presence of microparticles for sensory examination post centrifugal separation? I think this is a more objective way to analyze certain aspects of flavor as these insolubles have a dramatic effect on how our tongues perceive flavors. Great post!

Frans zei

Thanks Ray! We did not drink/taste any of the 30 shots. On every extraction, a syringes was used to agitate/mix the extraction and then some ml's were taken out and put in a test tube. After centrifuge a pipette was used to take a little of the fluid below the surface to measure TDS on the VST refractometer. In an earlier blog or two about the centrifuge I posted more details about that procedure.

Unknown zei

very cool