Bring Me Coffee or Tea

extraction & roasting theory, cafe reviews, and other stuff about drinks that smell nice ~

[no-stream]

six months with the de1+ espresso machine

[ugly phone pic of the machine; I didn’t tidy up for this so there are fingerprints and coffee stains (gasp!)]

I've had my DE1+ around six months now, so I wanted to offer some thoughts. I originally started writing here to catalog the experience of getting used to an EK and later did the same with my Quest. So let's try to do the same with the DE1+. This isn't a review of the DE1+, but I am very happy with it and can't imagine going back to a traditional machine. (Though the L1 is beautiful and elegant and low-hassle and would be an option if I weren't a nerd.) I'm a bit sad that the DE forum is invite-only, since there's a really interesting ongoing dialogue, though I understand it would perhaps be a bit crowded if it were public and it might be bad optics to have the public see peoples' bug complaints. 

In no particular order:

- I'm pretty enamored with flow profiling. It's really finicky and sensitive to grind changes, but it's great when you nail it, and it's surprisingly good even when you don't. Even down to a 3-4 bar peak, the results can be pretty tasty, though mouthfeel suffers a bit at really low pressures and extraction yield drops slightly. Mouthfeel seems to benefit from pretty high pressures, though I've gotten "good" mouthfeel down to around 6 bars. (Note that I'm used to an EK, so I just want some mouthfeel, not a super thick, syrupy shot.) I have basically zero sink shots, though many of my shots are just "very good" and not "excellent." (I don't have the patience to stick with one coffee and dial it in most of the time.)

- Similar to the first point, pressure doesn't affect extraction yield as much as you might think. In fact, when I've screwed up and choked my machine and pulled 11 bar shots, extraction yield has been slightly lower (say by 0.5-1%, very roughly).

- Increasing flow toward the end of a shot - which is pretty inevitable if you pull with constant pressure - tends to create bright and even unripe flavors. I really don't care for increasing flow in a shot, though perhaps it could be viable with darker roasts that want more brightness. I'm not the first person to point this out, but I find it to be very true. This means that traditional flat pressure profile pump machines are pretty flawed, in my view, and levers come much closer to optimum.

- Lower flow rates than you might imagine can be viable with the DE1+. If you imagine a traditional third wave shot - 18 in, 30 out in 28 seconds, say, and with a dwell time of 8 seconds, you have a 1.5 mL/S flow rate.  I started around 2.2 mL/s (so a bit higher, and closer to what I used with my Londinium I; On the L1, I pulled pretty quick shots, which is something I've heard from other EK users, but that hasn't translated to the DE). I've worked my way down to around 1.5-1.8 mL/s. On the DE forums, many people are pulling <1mL/s, and these profiles actually taste pretty good! If I tried pulling these flow rates on any other machine I've worked with extensively (CC1, L1, BDB, LM Strada & Linea), I would've gotten a savory, bitter mess. So why is this the case? One pocket science guess was that the DE1+ manages overall puck temp, not just water temp, meaning it avoids the temp increase of other machines, but the BDB also does this and suffers from the low flow rate problems.

- Lower flow rates don't necessarily increase extraction. In fact, my shots pulled at very low flow rates (<1mL/s) trend toward lower extraction yields. I get my best extractions at more traditional flow rates, though the effect of flow rate is not massively significant and I haven't studied this systematically. - I'm pulling the longest shots I've pulled in a while since the DE/EK combo is so efficient at extraction. A typical shot - when I nail it, which is not always - is 21g dose in a VST 20, 58g out, 8.4 TDS, 24% EY. When I don't nail it - and I change espresso often, so this is common - extraction yield might drop to 23-23.5%, and strength accordingly. These are not thin, tea-like coffee shots. They're rich and sweet and fruited and have (some) mouthfeel. 

- Pause/bloom profile shots seem to offer better mouthfeel than Slayer-esque shots with deeper flavors (more rich, less bright). On balance, I probably prefer bloom shots to Slayer shots, but it varies depending on the coffee. Extraction yields are both excellent and similar. (Check out Rao's post on these kinds of profiles if you're curious - https://www.scottrao.com/blog/2018/6/27/flow-profiling-on-the-de1 ). (When I say Slayer shot, I mean slow pre-infusion at 1-1.5 mL/S up to around 6 bars followed by a slow extraction at around 1.2-1.5 mL/S.)

- Having a machine heat up passably (well enough to pull a shot without too much compromise) in 7 minutes or so is pretty great. No need for a scheduler or an automatic outlet or getting up and flicking on the machine and then returning after a shower. Just turn it on, throw on some pants, and come back and pull a shot.

[a shot I pulled this morning. This profile is needlessly complicated, but I’ll point out some features. Decreasing temp; PI at 3 mL/S to 3 bars with 20 second pause; 5s ramp to 6 bar to preserve puck (testing this idea out); pressure cap at 8 bars to prevent crazy pressures from the DE trying to reach my desired flow rate; flow at roughly 1.8 mL/s.]

[no-stream]

Hello! Do you think that an EK43 with SSP Red Speed burrs produce what you would still call as EKpresso? Because if I'm not wrong, they produce more fines than the old coffee burrs and are more similar to a Mythos One. Interested to hear your thoughts on this.

Hi! Sorry for the delayed the response. I have SSP Red Speed and have been very happy with them. Socratic Coffee showed their particle distribution to be more modal than the stock burrs and way more modal than a Mythos. (I’ve posted the graph before but sadly can’t find it immediately.) 

I get great extraction yields, better ability to choke flow / grind finer, and arguably cleaner flavor. Drip coffee seems not dramatically harmed in terms of flavor. 

Feel free to follow up or message me on HB. Cheers.

[no-stream]

week one with the Ikawa home roaster

[excuse the sloppy phone picture.]

I. Intro 

This is the first in what I hope will be a series of posts on the Ikawa Home roaster. Sadly, there's not that much written on the Ikawa Pro and even less on the Home model. (If you're curious about the pro, you can check out the Ikawa blog and the Royal Coffee blog.) The Pro is pretty inaccessible for hobbyists due to its cost. (You could almost buy two EK's with the money.)

Experimenting with roasting is fun, but actually doing the roasting can be a little tedious. Fortunately, the Ikawa lets you roast without the tedium - "set it and forget it," so to speak. There are a few downsides compared to alternatives like the Quest M3 - with the Pro model, you get inlet (ET or MET) and exhaust (pseudo-BT), but no true BT. You can't easily customize or add probes. And with the Home, you just get inlet, and you get a maximum of five points to control. (Though that should be adequate.) The Home also encourages you to charge at room temp, though this can be worked around. Both roasters accommodate a max of 50-60 grams - enough for 2-3 shots of espresso or 2-4 cups of coffee but more ideal for experimentation than producing large amounts of coffee for friends or commercial service. Of course, being able to program a profile and have it followed near-perfectly is a major benefit, and I'm preferring the Ikawa to my Quest. 

Because the Pro model is extremely expensive (around $5,000 US when I checked), I got the Home (less than a third of the price). This means that I have only inlet temperature, but since I typically control the roast using ET/MET, this is only a minor downside. Even with the Pro, true bean temp is not available, so the loss of the second probe is not the end of the world, and in some sense, I'd rather have "MET" than exhaust temp, since it's more similar to MET from my Quest and I'm familiar with adjusting it. 

II. Working Out a Baseline Profile

[photo of baseline debo profile]

I started with two profiles using the washed Hambela Hassan (not the dry-processed version of the same coffee that has been talked about on HomeBarista) from somewhere near Yirgacheffe in Ethiopia. (Ethiopian coffees are, no surprise, my favorites, and this was supposed to be a pretty good one.)

I'll be labeling profiles with numbers - e.g. (1), (2), (3).

1) Debo Kochere Ikawa Profile (shown above); I chose this since it was for an Ethiopian coffee and the curve shape made sense.

2) Inlet Gesha profile from Pro model; as with many aftermarket Pro profiles, this was very fast, around 5 minutes

1) was fine - a little chocolatey, caramelized, and bready for me, but pretty good. My intuition would be to charge hotter and move faster with less development. 

2) was overtly baked. I tried lengthening and shortening the roast without luck. I will later attempt to return to the problem of adapting Pro profiles to the Home model.

Next, I shortened the profile to a bit over six minutes (versus around seven). This helped and improved clarity and acidity. No picture is included since I seem to have lost the profile.

Then, I simplified the curve and shortened the roast further.

[shorter debo pic]

3) This was quite a bit better and reminded me of something from Heart - really sweet and caramelized. It had a lot of development time ( : , %), and it was extremely soluble (I typically taste roasts as v60-01 brews using a lightly modified version of Perger's recipe - 12g : 204g (1:17), 209 F charcoal filtered Oakland water (soft, ~45 PPM), total brew time ~2:30-3:00) (x % EY for this profile, vs. x % for the baseline profile). After this roast, I wanted potentially slightly more acidity and fruit at the expense of some sweetness, but this was definitely good as drip coffee and I would suspect also as espresso due to the high solubility. 

Next, I tried and adding more heat early - the idea here being to simulate the extensive heat application in a preheated roaster. (The Ikawa has very little thermal mass, so we can't just cruise through the drying phase using stored heat.)

[hot start pic]

4) By now, results were excellent. With the profile approaching the 5:00-5:30 range I see in most aftermarket Pro profiles (e.g. from Royal Coffee's excellent blog), fruit and acidity were pronounced. Unfortunately, there was a slight hint of baked flavor in this one, and one idea was that it was due to the strange heat application at the start - increasing, then decreasing.

[charge at 30s pic]

5) This is where I ended up, though there's always further room to tweak. You can trick the Ikawa into thinking you've added beans by completing the charging motion without actually dropping in the beans. Ideally, this will ensure that the beans see constantly decreasing amounts of heat applied, and thus decreasing bean temp rate-of-rise. (In my experience, having declining RoR during development is extremely important, but you can often get away with non-decreasing before first crack.) I added the beans at 0:30 when the roaster's thermocouple read 250C. The total roast time was 5:10, with around 1:00-1:10 of development time. This roast was pretty much exactly what I wanted - sweet, bright, and soluble, with no baked flavors or breadiness. Roast weight loss was x%, very light, and extraction yield of the v60 was x%, which is very respectable. I'll probably try pulling a shot with this, though I would develop slightly longer and possibly a few degrees hotter for an espresso-oriented profile.

III. Other Roasts & Future Work

I also tried a few profiles with less of a "ramp," meaning the inlet temp dropped less from the "charge" or initial temp spike to the ramp phase. These were sweet but tasted a bit heavy, woody, and spicy. I suspect they spent too long in the development phase. (Since the roast neared its max ET earlier, the BT was closer to the ET for longer; with time, BT approaches ET asymptotically.) Perhaps I could have shortened the roasts to get better results. However, this gave me another idea - to make the opposite adjustment to the ramp - still charge hot, but then ramp from a lower temp, potentially producing a brighter and more aromatic result - but we will see. (This approach may reduce solubility, but the hot charge will at least be working in our favor.)

I'm out of this coffee, so I'll be exploring these ideas on a washed Guatemalan coffee (SM's Rosma Lot 1) and a dry-processed Ethiopian coffee (the well-regarded Hambela Hassan DP; from the same area, but processed differently.) It's my experience that most coffees react well to the same basic approach - applying quite a bit of heat early (but avoiding scorching) and finishing the roast quickly to preserve acidity. Frequently, I've used very similar profiles for dry-processed Ethiopian as washed Ethiopian coffees and not had issues with scorching or tipping. The exact length and shape of the ramp curve affects aromatics, and the exact amount of development and finish temp determine the type of sweetness (fruited vs. candied vs. caramelly vs. heavy, brown sugary) and I typically adjust development time depending on the intended use (drip vs. espresso). But the overall curve looks relatively similar.

If you'd like any of my profiles, feel free to contact me and I'll post them to Ikawa - but they're under-development right now, and I will post them when they're closer to done. 

[no-stream]

power analysis and espresso experiments

I've seen a lot of really great and informative espresso experiments online - most notably from Socratic Coffee and Compound Coffee. If you have a coffee refractometer and filters, you can conduct experiments too. Since I'm planning on running some soon, I did something called a power analysis, and I wanted to share the results so you can decide how many shots you'll need to conduct a valid experiment. 

The TL;DR is you'll need around 10 shots per condition for an experiment intending to detect a .5-1% difference in extraction yield, but if you're really consistent and use lower pressure, you might be able to get away with slightly fewer. (A condition is just a variable you're changing; for instance, you can pre-infuse or not pre-infuse.) 

In some espresso experiments, I've seen people use as few as five shots per condition (e.g. low pressure and high pressure or low temp and high temp). Unfortunately, with small samples, it's hard to tell if your result is just chance. However, if we know about how variable your results are (i.e. how much your extraction yield varies from shot to shot) and how large an effect we expect (i.e. do you expect an increase of 0.5% EY? 1%? 2%?), then we can figure out how many shots we need. Realistically, it's a major pain to use more than 10 or so shots per condition. Even five is a bit tedious.  

Of course, we don't know how reliable your extraction yields will be, but we can extrapolate from past results. For instance, we can use https://compoundcoffee.com/experiments/12_Brew-Pressure-Impact-Sidamo-Guji and https://compoundcoffee.com/experiments/13_Pre-infusion-and-its-impact-on-espresso. We're looking for the standard deviation (SD) - a measure of the spread between subsequent espressos in terms of EY. (In simplified terms, you can think of standard deviation as the typical fluctuation between two measurements - in this case, two shots.) We see that SD is higher for 9 bar shots and lower for low-pressure shots and pre-infused shots. (This is enough to convince me that I'd run experiments at no more than 8 bars and with pre-infusion.) By lowering our SD, we'll be able to establish our results more easily. (The less noisy the measurement, the easier it is to see our differences.) 

For extractions at 9 bar, expected SD is around 1.2-1.5% EY. For lower-pressure extractions, it's on the order of 0.6-0.8%! So we can plug this value into a power calculator - http://www.sample-size.net/sample-size-means/. You can leave alpha and beta unchanged. (Try googling them for further info, but they describe how willing you are to have a result that's wrong for one reason or another.) q1 will remain at .5 since you're splitting shots evenly between conditions. You'll just change E, the expected effect, and S, the expected SD. Assuming you're using lower pressure and are as reliable as the baristas at Compound, you can enter around .70 for expected SD. So for instance, if you're experimenting at 8 bars and expecting EY to increase by 1%, you'd enter .7 for SD and 1 for E. 

Optimistically, you end up with as few as eight shots per group (1% difference in EY expected, .65 EY SD) or as bad as 28 shots per group! (0.5% difference, same SD). 

Since these are based on low-pressure extractions, you'll need even more shots if your technique isn't reliable or if you're pulling at 9 bars!  Just remember that if you're trying to find a pretty significant difference in EY (0.5% is very tasteable), you might still need to pull 40 or 60 shots in total! So if you pull 10 shots and don't get an effect, you also haven't proven that there's no effect! Alternatively, when reading experiments that use only 5-10 shots, don't take the lack of a significant finding to mean that there's really no difference; there just isn't one large enough for the experiment to detect. I hope this was somewhat comprehensible. As always, feel free to message me on HB if it isn't clear. (user NoStream)

[no-stream]

Notes from Perger's Lecture - Advanced Coffee Making Lecture at Assembly Coffee London

I like to take notes on cool videos and post them to (1) encourage you to watch and (2) help you skim through the content if you don't want to spend an hour or more watching. _my notes are in parentheses_ I'd love to include some slides but feel that would be iffy in terms of intellectual property, so just skip to the timestamps if you want the visuals. LINK: https://youtu.be/2lPGf1gM9nA idea from The Hog distribution tool - We have to grind finer to get around our uneven, clumpy bed and slow flow down; but our espresso actually extracts higher with a coarser grind (thus, lower pressure is better); we want to extract faster and with a coarser grind through non-clumped, even distribution - refractometer/vst talk is done now, we get it; less focus here - "strength" means measured TDS - EY is just coffee solvated / total coffee mass Low EY -> sour, vegetal, salty, nutty; "could be more" / "underwhelming" / "boring" High EY -> Dry, bitter, astringent, powdery (like baker's choc - nice); "empty" "correct" EY -> transparent, sweet, balanced, nuanced, aromatic... Definition of EY Unevenness of extraction - some people think uneven extractions are "complex" This is most espresso! This is also what we perceive as "espresso flavor" - the mix of over and under-extraction. (If you don't know what "espresso flavor" is, try putting an EK and Robur shot next to each other) ## (16:50) Sources of Unevenness - ### a) growth - regional (within the Cauca region, for example) - microlots concentrate the flavors of a particular terroir (more even) - moisture content affects Maillard reactions in roasting ### b) roasting - drum roasters kinda suck; we have to work with them - blending reduces transparency, of course - outer and inner agtron should be identical, ish! (some people still think they should be different) - (this means you can roast lighter, since your limitation is the lightest portion of the bean) - pre-blending (you should solubility match - Heart does this) - - > cool slide - check it out ### c) grinding - EY is a blend of pockets of lower and higher extraction - small particles extract instantaneously (ish) - large particles are really difficult to extract (cellulose takes some effort to penetrate) - EK produces small, even-ish particles - - > (31:30) _for a while, everyone cared about the height of the peak (the mode / local max), but what we care about now is the distance from our most-extracted to least-extracted, or the standard deviation of the particle sizes_ - for each particle, we potentially extract everything from the outer layers and next to nothing from the inside (if it's a large particle) - ideally, want a fine grind with very even particle sizes - the majority of the flavor you're tasting was extracted at approx 28% EY (maximum extraction); less was extracted at less than 28% - it's really hard to overextract; many "overextraction" flavors are bad roasting; it is possible to overextract, but only f you e.g. channel an espresso or boil an immersion for several minutes (remember when people used to say that grinding espresso too fine / bitter espresso meant overextraction? but those espressos were 15-18% EY, if that!) ### Experiment! - _(40:05) experiment: sift out fines, medium, and larger particles; same weight of each, cupping bowl, boiling water, stirring, check extraction over time; small particles plateau as expected (24% EY within a minute), but the large particles don't really catch up! (just over 19% EY at 10 minutes!); we can figure out how deep the water penetrates the particles based on these curves - water penetrates 100 +/- 10 microns!_ (here's another reminder that typical french press grinds are way too coarse and lead to a generic underextracted flavor) - this means we extract the outer 100 microns of every particle, assuming the worst case of sphericity, 200 microns is ideal, but brewer would clog; probably roasting can optimize for increase in volume and improve access to the inside (maybe I'm slow, but this is an interesting explanation of the roasting -> solubility relationship) - large particle EY slowly increases due to erosion of outside of particle granting access to inside over time - onyx coffee used above, light roast, less soluble; result would differ for a different coffee - (52:51) assuming we have decent evenness, grind to maximize extraction yield (go as fine as you can go without getting clumping, channeling, and clogging) - if increasing EY doesn't help, you're roasting poorly or extracting unevenly (again, true overextraction is really really hard to achieve) - huge industrial grinder at MK -> 26% EY! result was "otherworldly" ### Tips (56:53) - Brewed coffee: treat every grind the same, agitate/stir, wet immediately, don't choke it, sift boulders (>500 microns) - MP has betrayed the cone, now prefers flat bed  - manual brew: bloom 2x dose, keep water level consistent, don't toss water around messily but keep height constant - auto brew: constant delivery / water level; know that it doesn't deliver the amount of water it says; coarser grind than manual brew otherwise water just goes through paper - immersion: dump water in vigorously; stir but don't leave spinning; use two papers; inverted press - espresso: VST please (why would you not? they're cheap and will force you to get better), flat 58.5 tamper, full hopper b/c grinder designs suck; distribute in some way - hand taps do fine - MP is working on a fluffer / "sideways tamper" that sticks holes in a spro bed - up to 2% EY increase - as big as conical to flat or curved tamp to well-sized flat tamp - MP wants you to take his course; it’s $150 which isn’t too bad if you have already spent $1000+ on gear. ### Questions (abridged) - regrinding: generally not a good idea because you can stall out or overheat your grinder; but it can be good in e.g. roller mills - my ek flows too fast! align it. (also - you can re-zero them to start with after they've had 5kg+ put through them); on a well-aligned EK, spro should be at 2:15-2:45 if you imagine it as a clock  - ad-hoc research is more productive than academic, and the costs of false positives are not so bad - arrhenius eq suggests doubled rate of extraction per 10C increase; so increase your temp within reason and pull shots in less time - if you have chaff left after roasting, may be too light a roast or inadeqate airflow

[no-stream]

We’re back, for a bit at least

[the first of a series of pictures of a shot on the DE1+] Here's a post on a no-longer-defunct blog about coffee. Tumblr is even less of a thing now, but who cares; I'm not going to move this, yet. If you messaged me on Tumblr, I didn't see your message, so my bad for not replying.

I received a DE1+ espresso machine in the mail a few days ago, and my main goal is to write about my experiences with it / review it. I'll also be posting a bit about other things - some posts I have planned will deal with upgrading my EK43 (Titus burr carrier, SSP Red Speed burrs) and my current workflow, tampers (I picked up a PUSH and a Eazytamp), and less exciting stuff like coffee crawl reports, bean reviews, and so on. In terms of aspirational posts, I'm thinking about buying an Ikawa roaster and will write about it if I do. I'd also love to do a few coffee experiments - specifically, testing for effects of WDT and RDT on extraction yield. (I do the first religiously but not the second.)

I'm also looking to sell my heavily-upgraded Quest M3 (four EricS thermocouples, phidget, usb loggers, insulated) and Londinium I, so contact me if you're in the SF Bay Area and are interested.

[no-stream]

Hey, happy New Year! As a Quest owner, I just wanted to thank you for providing such extensive insight into the roasting process, including your experiments and thought pieces. There have been times where I think I'm off, or I can't understand what or why something is happening, but I've bookmarked your site to glean useful tidbits. The only question I have is why my (subjectively) high MET is like it is, usually between 500-600F to get the time I like for 150g loads.

Hey, glad you’ve enjoyed it. I don’t maintain this page or post anymore, but I’ve left it up for people like you. I don’t have an answer for you regarding MET. I know there are several revisions of the Quest, at least two drums, at least two fan circuits... I would look into that. Reddit and HomeBarista should have some threads discussing the M3′s revisions. 

Obviously, MET could be higher because of your probe location. MET can also be higher for a given level of roast if your roaster is less insulated (on the exterior, or is lighter, or has a thinner drum).

I’d also check out other M3 users’ graphs - I know TomC on HB used to post with one, although his is heavily modded. Mine is modded with insulation, but I think I clearly marked when I added that. I know I’ve seen graphs from users’ reaching 550-600 F MET who were happy with the product.

[no-stream]

Hi,I would like to ask, how do you keep your ET so high? From the chart, it seems you could always keep it above 400 degrees.But In my case, I pre-heat the drum to 480 degrees, and put my bean (150g) into the drum, then the BT went down to 200 and ET also went down to 300 degrees. I'm so curious how did you do that. Thanks!

Well, there are a few possibilities. The first that comes to mind is that you didn’t preheat enough. Keep your MET above charge temp by 20-30F for 20 minutes or so before you roast to keep the whole drum nice and hot. Then drop it to charge temp for 10 minutes or so before you charge (for your first roast of the day). 

Another one is that your fan is really high. Do the Rao lighter trick (Google it if you aren’t familiar with it - or just buy the book). You shouldn’t have to max out heat early to maintain MET. 

The last possibility that I can think of is that you have an ET probe and I have an MET probe, but I suspect we both have the same probe, by EricS, at the top of the drum. 

[no-stream]

I am considering buying a Quest M3 after reading your articles on Nordic style roasting. I'm still a bit apprehensive about the purchase though. Do you feel that the Quest is giving you comparable results to commercial roasters? And if you had to start again, would you get the Quest or something else? Sorry if this message double posted, I got asked if I was a robot by tumblr and my message vanished so I retyped it.

I honestly haven’t roasted in several months and am not really updating this blog regularly at this point, so take my words with a grain of salt. I’ll still be posting the occasional tea or coffee review.

I think the Quest can produce good roasts, but you need to be familiar with it (as with any roaster), and you may need to mod it. The fan speed control posted on reddit is a must if your fan is too strong at low settings. Insulation is nice but unnecessary. The drum drilling mods on home-barista look intriguing...

I got the Quest because I wanted something electric with a small batch capacity. I didn’t want to have to roast any more than 1/3 lb. at a time. The biggest competitor I know of is the Huky. I think it’ll give you a little more control of both fan and power. (Electrical elements have lag when you want to increase or decrease heat.) Obviously, a gas roaster is a bit more of a pain to use, so it’s up to you whether that makes sense. And you can go higher-end with something like one of the North roasters, but those are way larger and heavier. For what it’s worth, when I participated in a roasting competition, a friend of mine who roasts on a North has put out some pretty great roasts, and I was unable to replicate those on my Quest. Another person working on a Quest and a third on a HotTop were similarly unable to do so. 

[no-stream]

Hello! I was just curious how you rinse your vst filters? I know it's not the best idea, but if it can get me a ballpark number for casual testing of espressos, that would be awesome. Thanks!

You’re right that it’s not ideal. I actually went ahead and tested the effects of reuse and found that readings would gradually increase in terms of TDS with reuse - even with rinsing. Expect an effect on the order of +.1-.3 TDS. I’d much rather use off-brand filters - I recommended SEOH .45uM PES filter a while back - than rinse “proper” VST filters. Of course, I’d use VST filters if I enjoyed burning money.

Anyway, check out the official VST documentation, but I’ll give a brief overview of how I cleaned filters when I used to do so. Use the filter. Then remove the filter from the syringe and manually dab off any excess espresso with lintless paper towel - I bought a box of KimWipes for something like $5, but a paper towel should be OK too. Then syringe up a bit of water into the empty syringe, attach the filter, and slowly express a bit until the water is clear. Then you’re done. Make sure to try to dry the filter manually and let it air dry, otherwise the cleaning water will dilute your sample and bias your reading down.

Again, I’d recommend buying off-brand filters, doing a quick comparison with your VSTs to assure yourself that they’re essentially equivalent, and not reusing them. The cost is comparable to washing the official filters.

[no-stream]

quick question about airflow. i saw in your experiments that low-med-hi seemed to be the best option (similar to rao's thoughts), but then later on you decided that increasing the air led to baking. i've also heard a couple of well known roasters mention using the lowest possible airflow and not changing it. just wondering where you land on the issue currently.

Airflow’s pretty complicated and not something I’d claim to understand very well. For simplicity, I’ve resorted to keeping a steady level of airflow for each roast. Too low compromises development and causes savory flavors and too high strips aromatics and leaves a hollow, grainy flavor. The sweet spot is somewhere in the middle. Keep in mind that the airflow in my Quest isn’t really well-directed when compared to something like a Probat. Ideally, air flows from the heat source, through the drum, and out the exhaust. If you have a Quest and are curious about its limitations, I’d check out this h-b post: http://www.home-barista.com/home-roasting/quest-m3-mods-t25590.html

I did previously adjust airflow during the roast. Suffice it to say that even though it might produce marginally better results, any improvement was minimal and not enough to warrant the massive hassle of having two variables to manage (heat and airflow). I think steady air or increasing air are the two plausible ways to manage airflow in a roast, at least in the Quest. For what it’s worth, in my very minimal attempts to compare the two styles, I did get a bit higher acidity (pleasant acidity) from increasing airflow throughout the roast. 

As you said, there are plenty of well-known roasters who don’t adjust airflow throughout the roast, and I’m content to be in that camp for now.

[no-stream]

some thoughts on the starbucks reserve roastery

[exterior]

I spent a few days in Seattle a little while back and didn’t have any third-wave coffee, unless you count the Starbucks Reserve Roastery. (I also had an awesome Da Yu Ling at Teahouse Kuan Yin – please do check that shop out if you visit Seattle.) As silly of a place as it is, I’ve wanted to see its comical – and also quite appealing – design for a while. And of course, there’s the morbid curiosity of wondering just how good Starbucks coffee can be…

 The exterior is pleasant – tons of glass and metal, white stone. Walk in and you see a gigantic espresso bar with beans being tossed about by vacuum chutes. The copper and reclaimed-esque wood are a crafty aesthetic borrowed from various third-wave classic shops – I’m inclined to mention Sightglass or Four Barrel in SF. At this bar, you can just walk up and order in the traditional third-wave bar flow.

[main bar]

[Congo spro]

I ordered an SOE. It was, in fact, a “Congo Lake Kivu” because they didn’t have anything better. (I would’ve liked to try a coffee which had the potential to be really excellent – a Yirg or a Nyeri or a Kirinyaga or something along those lines.) In case you haven’t heard, the Reserve has some crazy equipment – Black Eagles, Mythos (I have no idea how to pluralize that word)… Anyway the shot came out pretty pleasant – it was a little funky and herbaceous (that’s the green coffee talking) with pleasant orangey-sweetness and creamy body. Really, it was very decent. As was later true of the brewed coffee, I was perfectly content with it. And I’ve been served much worse in third-wave shops. I’ve even been served worse in some pretty decent shops that I frequent regularly.

[the smaller Probat]

The roaster portion is very clearly indebted to third-wave norms. There are the requisite Probats, a smaller one for this location and a larger one for distribution. Thankfully, Starbucks didn’t go full-third-wave and find a vintage roaster. It’s my understanding that fetishism for vintage roasters means that the all-in cost of a fully-retrofitted Probat is higher than a new one. And, of course, Probat only makes classic-styled roasters because of the vintage-artisanal-mania of specialty coffee. (Though I readily admit that you really do want the thick walls and lovely construction of the classics. It’s just amusing to see Probat taking after Red Wing.)

[modbar pour-over]

We went downstairs and tried some brewed coffee. Rather, before that, we took a seat, a server came over and talked to us – she was super friendly and pleasant and really did offer some of the best service I’ve received in a coffee shop – and we ordered. I had a pour-over pf the same Congo from upstairs made on the Modbar (might as well have multi-thousand-dollar water dispensers, right?), which came with a really amusing but over-the-top card. To me, it tasted like black pepper, tobacco, burnt toast, and sweet orange. The coffee was clearly well-developed, well-extracted, and sweet – no vegetality or off-nut flavors. It was also a little smoky, but not obnoxiously so. As it cooled, it became slightly papery but not excessively so. Again, like the espresso, quite decent; not exceptional but very acceptable.

 Tipping at the Reserve is a little strange. There are tip bins, but they seem little-used. Mostly, they just have coins. Credit card prompts don’t offer an opportunity to tip. I really do hope the baristas are paid decently.

[records; or, uh, maybe the folks here would call them vinyls? and a crosley so you can promptly screw up the grooves]

In the end, visiting the Roastery is a bit of a surreal experience. It’s like being in a third-wave shop built by aliens who’ve done their homework.

 It’s interesting to think about how much coffee is caught up in trendsetting and nonconformity. Now that the third-wave notion has been entirely co-opted into a traditional commercial model, we see even stranger and contrarian ideas emerging – like Dashwood’s experiments with pod espresso and Freese’s instant coffee (Sudden Coffee). There’s still a possibility to make great third-wave coffee, but you’ll have to differentiate yourself through quality and spread info via word-of-mouth rather than grand architecture, design, and indie signifiers (e.g. records).

[no-stream]

some thoughts on the bonavita brewer & flat-bottom auto drip

I haven’t posted here in approximately an eon. I won’t be posting here unless I have something vaguely interesting to say, perhaps every month or three. Anyway, let’s begin...

[next to the previous brewing abomination]

Last holiday, I had a chance to brew coffees on Bonavita’s newer model of auto drip brewer, the bv1900ts alongside a Baratza Encore grinder, since my family had recently picked up the pairing and I was visiting. All in, it’s around a $300 setup and one which I previously would’ve readily recommended readily. (Spoiler: it’s still one I’d recommend.)

First, I thought the Encore was pretty excellent. Retention was a mere .1g with single dosing and grounds looked pretty consistent. The grinder responded well to adjustments in size. Build quality is more than adequate for the money, and I have no reason to expect one wouldn’t hold up for at least a few years.

My opinions on the Bonavita were somewhat more mixed. Before I offer any thoughts, let me mention briefly the parameters I used. I brewed several coffees, and the ones which I remembered to note down were a Honduras, Pinona Pacas, from Ritual, and Intelli’s Zirikana, Rwanda. For both, I tried a pretty wide range of brew ratios, from 1:15 to 1:17.5; doses, from 30 to 60 grams; and a wide range of grind settings, from bitter-overextracted to sour-underextracted. I had absolutely no trouble getting what seemed to be a good extraction - sweet and clean and pleasant. I could get a syrupy, heavy mouthfeel if I wanted to. 

[near-perfect post-extraction bed - better than most Fetcos I’ve seen]

But the flavors tended toward somewhat generically coffee-like. Now while I had tasted both of these coffees as v60 brews in keeping with my slightly-modified Perger technique, I didn’t have my equipment with me in Los Angeles to stage a direct comparison. But the thought that occurred to me was that this well-extracted but not aromatic “defect” was really reminiscent of what I’d experienced with the Kalita Wave. I used to be a believer in the Wave, and I still don’t think it’s a bad brewer, but once I learned to extract properly with my v60, I found the Wave’s aromatics lacking in comparison, even if it was more foolproof. In some sense, the two brewers remind me of each other - easy to produce good coffee, but in some sense not disposed to producing really excellent coffee. 

It’s perfectly possible that flat-bottom brewers extract “better” than cones - more evenly, higher with an equivalent amount of bitterness... but I’m not convinced that the result is necessarily better to my palate. Maybe it’s just a technical issue, and certainly I think I have better cups from a well-dialed Fetco than the Bonavita. (For one, the Bonavita’s far less programmable. Also, the Fetco’s dispersion screen achieves a much more even - but not perfect - distribution of water. Why the world lacks an auto drip machine that can distribute water evenly without producing divots in the coffee bed is beyond me.) 

As I have said before, I think we may actually be seeking near-perfection and not quite total perfection in our brewing practice. Some small elements of under- and/or over-extraction may produce a really pleasing, complex, aromatic cup. Don’t use “complexity” as an excuse to brew sloppily or to skip weighing or refracting anything, but remember than measurements are only useful to the extent that they correlate with ideal taste.

[no-stream]

Are you still using the insulation on your quest? I've been trying to get a declining R-O-R with my quest and I'm struggling, curious if the insulation changed anything for you.

Suffice it to say that if you aren’t achieving declining RoRs without insulation, adding it will not magically change anything. Insulation does make it a bit easier to achieve declining RoRs and more importantly smoothly declining RoRs (because you have more mass, and so your roaster’s less prone to rapid fluctuations). It also lets me pull off lighter roasts with good development; or in other words, it lets me achieve more solubility (%EY) at a given roast level (% weight lost). 

So it’s worthwhile but not necessary or a magic bullet. I always recommend the fan mod as a first step to new Quest owners. Just google “reddit quest fan mod” and you’ll see the thread. After that, just trial and error. You can look at my and other quest owners power and fan controls. The Quest works pretty well with steady fan - just enough or a little more than need to pass the Rao lighter test - and just controlling power. With only one variable (power/amps), there’s less to think about and your skill at roasting will improve more rapidly. Also keep in mind that the Quest still has some thermal mass, so you’ll want to move into development with some momentum and then ease of power throughout. That’s really the key to a nice clean RoR curve. Development is the hardest part of roasting and also the most important. 

[no-stream]

I recently bought a quest, and for the time being am using a thermocouple available from Coffee Shrub to measure bt. my tp is sitting at around 1:45 right now, which I'm not happy with at all. where would you recommend I look to find a more responsive thermocouple? are you pretty happy with the one supplied by EricS? also what are your thoughts on the stock placement on the m3?

First of all, I definitely recommend EricS’s probes. They’re a bit expensive, but you’re paying for Omega thermocouples, which are never going to be cheap and which are much less prone to random failure and noise than a lot of cheap eBay probes.

I have two BT probes, one in each of the EricS locations. Suffice it to say that I have been meaning to write about them but haven’t had the time / haven’t been roasting often enough to offer any concrete opinions. Anyway, EricS’s BT1 is at 3 o’clock and offers a pretty good view of bean temp, but it’s a bit thick and has a tendency to lag when the actual bean temp mass has stalled out in terms of temp. BT2 has less of a tendency to lag since it’s more fully immersed and is way thinner. BT2 also has a way quicker turning point time, so you get useful data sooner. The downside is that it’s a shorter thermocouple. In general though, I actually prefer BT2 because it helps me track stalls and flicks more precisely.

By the way, 1:45 sounds like a really weird location - more of an ET or MET than a BT. In general, you want to prioritize having a thin probe and having it deeply inserted in the bean mass.

[no-stream]

tea brew vessels & heat retention

There’s a whole lot of traditional wisdom in tea, and brew vessel selection is certainly subject to plenty of it. In my experience, tradition would dictate a glass or thin porcelain gaiwan for green and often white teas, a gaiwan - either thick or thin - for light oolongs, a thicker gaiwan or teapot for darker oolongs, and a teapot exclusively for black and pu-erh teas. 

I was a bit suspicious - does a gaiwan really lose that much more heat than a teapot? Does a thick gaiwan really brew hotter than a thin one? (On this question especially, I was very suspicious given that thick ceramic pour-over brewers are actually a very poor choice, given that they require an absurd amount of hot water rinsing to heat up properly. I highly recommend this reddit post if you’re curious about that topic.)

Anyway, I decided to investigate this with the three brew vessels shown above. From left to right, we have a very thin porcelain gaiwan, a very thick ceramic gaiwan, and a pretty thick unglazed “Yixing” teapot. All are between 120 and 150ml in capacity. If traditional wisdom is right, heat retention should increase from left to right, and if my suspicions are right, then the thick gaiwan should actually retain the least heat.

[here’s the super thin porcelain on the porcelain gaiwan; it’s so thin as to be translucent in the right light; I was drinking a lovely Ali Shan oolong at the time.]

So here’s the procedure:

I heated each vessel with water at 205F for just over a minute, pouring in a few segments as to avoid giving one a head-start. I then measured temps, and at this point the thick gaiwan was noticeably cooler (154F water) vs. the others (169, 170F for the yixing and ceramic). I then poured out the water and poured in more 205F water as if I were brewing a tea. Now the difference lessened dramatically. From left to right, I got 185, 180, 187F respectively. So the thick gaiwan was still the coolest, but by less of a margin. At one minute, the gap lessened once again (177, 174, 179), and at two minutes, all three were virtually identical (173, 173, 174). It’s plausible that the thick gaiwan would be slightly hotter than the other two in the final minute or two of a very long brew.

First of all, the differences between these devices were pretty small in absolute terms. It’s probably best to pick tea brew vessels based on aesthetic and kinesthetic-enjoyment terms rather than pretty arbitrary traditional wisdom. That said, it’s important to preheat, and if you do so, brew vessel selection becomes relatively unimportant. (It causes a 12-25F difference and reduces variations between different types of brew vessels!) Thicker vessels tend maintain cooler temperatures than thinner ones, at least for typical gong fu cha brew times (30 seconds to 2 minutes). (This is for the same reason that thin pour-over brew vessels suffer from less heat loss - less mass to heat up.) Teapots may offer slightly improved heat retention, but only by a small margin.

[yixing detail; it’s pretty thick for a yixing, but most yixings are actually pretty thin - maybe traditional artisans had the right idea, since thinner designs actually tend to retain heat better.]

[quickie excel graph of the data]

[no-stream]

a few coffee shops in... portland

[bonus filson pic outside Case Study. I’m not used to seasons being a thing.]

So, finally, I went to Portland. I’ll just go through shop by shop and offer opinions and recommendations. And since I’ve gotten out of professional coffee, I’ll be a bit more straight-forward about where you should and shouldn’t visit. And since it was a trip, I went full-tourist and brought my SLR, so pics abound.

First, Heart, over in Burnside. Heart’s a perpetual favorite of mine, and what better shop is there to visit first than the original Heart. (By the way, I’m always confused on whether I should or shouldn’t capitalize the “h” in “heart.” In this article, I will.) First of all, much to my surprise, the folks behind the counter were very mellow, friendly, and happy to talk. In fact, I was surprised by pretty much universally pleasant baristas in Portland - not service-training-fake-nice but just people doing their jobs contentedly.

So, I had their SOE, which was their Kiunyu, Kenya. Heart’s pushing boundaries as far as shot length; I consider myself to enjoy some seriously long shots, and yet this shot was long even by my standards - and served in a cap cup, no less. Flavor-wise, it tended toward tropical fruit with prominent guava character, light and creamy body, and again, was really light and non-espresso-like. In any case, it was a pleasant and unique shot. And props to Heart for being willing to serve a beverage so far divorced from Italian norms. (I did also try my friends’ milk drinks, which were spot on - sweet and caramelly.)

Then, it so happened that our lunch spot was right next to the Downtown Portland Heart, so, whatever, time for more coffee. Here I had their Yukro from a Fetco. It tasted pretty much like that raspberry lemonade that’s served at The Cheesecake Factory. (The memories of eating 2000 kcal meals from the place long ago flood back.) Again, it was really light - somewhere near 1.3 TDS. I typically brew around 1.35 TDS, but given that I’ve heard many Bay Area coffee pros claim that it’s just not feasible to serve coffee below ~1.45 TDS… evidently Heart manages. I think we shouldn’t be so afraid to challenge norms of what coffee is and should be. Heart’s a pretty and clean shop, but it’s not so wrapped up in its aesthetic as to forget about the coffee.

Next, Barista in Downtown(?) Portland. I had an espresso made with an El Salvadoran coffee from Coava. Given the dearth of exceptional coffees I’ve had from El Salvador and/or Coava, I should’ve gone with the Ethiopian “Epic SOE” -- God, that name is getting dated -- from the ever-excellent 49th Parallel -- but I was feeling a little tired of Ethiopian coffees. In any case, the El Sal was salty and papery - like old green coffee - plus some orange, smoke, and underextracted savory bean stew character. I don’t think this was necessarily Barista’s fault, since the coffee simply wasn’t well-developed, but they didn’t do it any favors by pulling it tight.  Luckily, my friend’s Jimate, Ethiopia filter coffee by 49th was excellent - immaculate and tasting like lemon and powdered sugar.

[Coava’s iconic interior

[a pretty impressive macc-setta at Coava]

Despite the mediocre experience with that El Sal, we went to Coava. It’s pretty much a landmark in Portland and much adored by the people, so we had to try it. I had their Meaza, which I’ve had before and thought to be pretty decent. It was brewed on Chemex-Kone, so oil made it through the metal filter, of course, and muddled up flavor clarity a bit. There were tons of berry flavors - reminiscent of a dry-processed coffee - without dry-process funk, since it was a washed coffee. The coffee’s sweetness was present but not prominent, suggesting adequate but imperfect development in the roaster. And some questionable flavors - herbaceous (oregano) and nutty (walnut) were also there. Anyway, it was a pretty good cup, just not great. And my friends’ milk drinks were quite well done - a macc and a cap.

Next, and finally, Case Study. I had a drip coffee - a dry-process Kochere Yirg run through an EK and Fetco. It was a dry process coffee, and a decent bit of processing character shone through. Sweetness was pretty unimpressive for an Ethiopian coffee, and flavors were unripe berries, eggshells, lemon peel, and caramel. My friend’s macchiato just tasted like a dry-process mess, but at least the cappuccino was pretty pleasant. I don’t think these flawed cups were the baristas’ faults - more so fault was on the poor green coffee (old, and not well-processed) and roasting (achieving poor development).

So that’s it. Making good coffee is pretty hard, and even in a coffee oasis like Portland, it’s tough to find greatness. In the Bay Area, we actually doing a pretty good job, even though norms tend to be a bit old-fashioned in terms of coffee and espresso strength and roast level. If you’re in Portland though, definitely prioritize Heart, and Barista would be a pretty close second.

I’ve been messing around with a lever at home - the Londinium I - so I’ll write up a brief review of that at some point. It’s a refreshing but challenging experience, and it honestly took me a couple dozen shots worth of experimentation to produce better coffee on it than on my CC1.

[bonus street-couch]