screen 2017-11-02 at 5.45.15 PM.jpg
We're beginning to insert all the parts in our Decent Espresso Machines into our shopping system. This will enable anyone to buy the parts to repair/replace anything in our espresso machines.
A goal is to reassure buyers of our espresso machines that they won't have any trouble getting parts from us. There are no forms to fill out, no approval needed. Everything is transparent. You can even buy parts ahead of time if you're the worrying type.
We will be posting annotated CAD files to http://onshape.com to make it even easier to see how everything fits together. We'll also be posting our own assembly instructions (videos, tips, and notes) to make disassembly/repair/re-assembly easier.
Next, we've made the URL to create a shopping cart intentionally very simple. The goal is to help people build software that automatically orders things from us. The URL is composed of pairs of "part number" and "quantity". Here is an example:
For companies that have automated purchasing systems, it's quite easy to integrate our products into your process and a few companies already work with us in that manner.
As with other things we sell, there is free shipping at 3 items, and price discounting starting at 5 items (10%).
We're back from the Korea Coffee Show. Our man Shin and his coffee-trainer/engineer friend ran the demos and made 1400 espressos in 4 days. This was a good test of DE1+ reliability under stress.
Indian National Champion Paras Bindra came by with 2kg his El Salvadorian Geisha (!) and was grinning ear-to-ear an hour later as we'd tuned the DE1+ to extract it at it's best (17s preinfusion at 3.5 ml/s, 8 second hold at 9 bar then decline to 7 bar). Sort of a Slayer-start with a Synesso end. "Wow, that's good coffee," said another SCAA judge, upon tasting it.
Super fan Chihyun Ahn is in the photo below, along with Joe McTaggart of Brew Brothers in Germany, representing Comandante hand grinders. I was super impressed by their grinder, though it took 60s to hand grind the Geisha, made a really an excellent Geisha espresso.
A big hello to our new friends Doug and Barb from Orphan Espresso, who we shared coffee, pastries and manufacturing war stories with.
I got to spend some quality time with Marco Feliziani of Nuova Simonelli who walked me through the incremental improvements to their Mythos 2 (my favorite coffee grinder).
And we were super excited to see my friend Kapo Chiu of HK's The Cupping Room - Roastery make it to the finals and take away 3rd place in the World Barista Championship.
Wow, that's an amazing effort.
That a couple of coffees a day for two years testing, a great stress test and I'm sure a lot of other machines would have struggled at that volume.
I hear the Korean shows are huge with a great coffee scene. I'll get there one day hopefully.
While the show was great fun, and I've had my fill of Korean BBQ now, I can't say I really recommend it to non-Koreans, as the vast majority of trade show stands had no English speaking salespeople. There were many products that I wanted a demo of, but nobody to speak to. As far as visitors go, maybe 1 visitor in 50 to our booth spoke English.
We had fully translated the DE1+ GUI into Korean, and I had two Korean native people working for me. Bugs spent all her time cleaning up after the espresso making (mostly cleaning and prepping portafilters) and I filled in during my Korean employee's breaks.
One upside to this, however, is that a lot of National Barista Champions came to our booth, as we were one of the only places they could hang out at an interesting booth and speak English.
And yes, the Korean coffee scene is really serious. 2/3rds of the Certified Q Graders worldwide are in Korea. We were able to have intensely technical discussions about coffee, with no attitude, and that was very gratifying. Nobody went "huh?" when I explained that this shot started like a Slayer and ended like a Synesso.
Last edited by decentespresso; 14th November 2017 at 07:25 PM.
- the biggest problem was that the "catering kit" for the DE1+, which is an external pump that refills our water tank, was accidentally pushed off the back of the table. As it fell, it ripped out the RJ45 phone connector on the PC board that it was tethered to. This "whoops" is likely to occur in the real world, so we're going to heavily reinforce the RJ45/PC board connection.
- tilting the DE1+, traveling in a cab to a cafe, and immediately righting the machine left the water level sensor confused, likely because water had dribbled up into the sensor tube. The next day, the water in the tube had dried and all was fine, but the water level sensor was confused and caused spurious "out of water" software reports at the cafe demo. We're going to work around this by optionally ignoring the water level sensor when it does this, as we can believe we can detect this (mis)state on power up. There is a more sophisticated solution possible, which involves using a semi permeable membrane, but in the past year we've not been able to reliably mount it without tearing or crinkling it, as it's very delicate stuff.
- a screw holding the box around a heater came out. Some "locktite" should fix that.
- two rubber feet came out because the holes holding them hadn't been drilled by our fabricator to our tolerances.
Two reasons I do these trade shows are:
1) to create deadlines
2) to real-life test the current state of the machine.
Last edited by decentespresso; 14th November 2017 at 07:27 PM.
It's great the machine is constantly being improved with bugs ironed out. Re the cable and reinforcing the board connection, wouldn't a magnetic link/connector or similar in the cable be better so that it won't stress any component if a mishap occurs?
Magnetic connectors were (I believe) designed to avoid people tripping on a cable causing a laptop to fly off the table. The fact that the RJ45 locks into place, and once clicked, won't come out, is important for a cafe. A magnetic connector that could be easily disengaged would annoy a cafe owner.
The "smart cafe" from Dalla Corte uses a round DIN connector (locking, I think, but didn't confirm). Those are also used on musician's microphones, and they're very strong. They're very big and heavy, but they are not uncommon (my Lyn Weber EG1 grinder uses one too). Here's an example of a DIN connector:
However, there are a lot of different phone connectors available, and in our engineering meeting, we think that switching to an RJ45 with reinforcements that get soldered to the board will provide the strength we need.
If not, we have the option to switch to a panel mounted RJ45 connector, like the one below, and that would totally separate mechanical stress from the PC board.
I guess the question is: "what would have happened if the board/connector didn't break?"
If the end result is that the machine is pulled off the table and breaks on the floor I don't think that's a better outcome. Ideally you'd want the breaking point to not exist (magnetic) or to be cheap and user replaceable (off the shelf cable).
We received samples this week of our espresso machine packing system. We're using 3 different pieces of foam to different effect. Rather than using styrofoam, I decided to use the same packing approach for our suitcase as for our cardboard box.
People who buy an espresso machine from us without a suitcase (the majority) will have this foam holding their espresso machine in a cardboard box, and that box will, in turn, be packed in another box, with an air gap between the boxes.
To hold parts in place, we're using the same molded EVA foam (spandex covered) technique that's in our Barista Kit. This time we've added a ton of ribbing for extra strength. The center layers are two symmetrically die cut PU foam pieces. Under the molded EVA foam pieces you'll find simple rectangular pieces
Two years ago, the foam I had made was high density, all glued together. The big problems with that were (a) it smelled terrible, very perfumey (b) the gluing effect was ugly, and (c) the foam weighed 5kg. The total espresso machine weight was 21.3kg, just slightly over most airplane limits, causing me to haggle at the airport every time I flew.
The new foam comes in at 1.6kg, bringing the "flight weight" of the suitcase and all espresso parts, to just under 18kg, safely well under the airplane limit.
The one mistake we made with this prototype is to require you to take the layers apart to get the tablet out. We're correcting that design assumption (with a cutout for the tablet stand) so that you can leave all the layers of foam in place as you insert/remove parts.
IMG_7453.jpg IMG_7456.jpg IMG_7462.jpg IMG_7461.jpg IMG_7452.jpg
This new feature helps you save everything about your best espressos so that you can try to keep making them as good as that amazing one.
Because the DE1+ interface has tabs, it's always taken two taps to start a function. For example, tap the "STEAM" tab, then tap "START".
On classic espresso machines, there is a dedicated button for each function, and so you only need to touch one thing to start it. That's much faster for a production setting, such as a café.
Today, I added a "one-tap mode" option.
If you turn this on, now when you tap a tab in the DE1+ GUI, that feature starts automatically. No more two-taps to start something, it's now one tap.
This is an optional feature, because it's not how "tabbed user interfaces" usually work, and might take a few minutes for you to get used to.
However, I think this is going to be a popular feature for cafés that use our machines, as well as for most people, once they become experts with making drinks on their DE1+.
You might also notice a new "stress test" option at beginning of the attached movie. This causes the feature you enable to repeat forever until you tap the STOP button for it. We're going to use this feature to "burn in" our espresso machines before shipping them to clients. However, I left this feature in the GUI in case you might find it a useful option.
IMG_7519 2.jpg IMG_7514.jpg IMG_7518 2.jpg IMG_7517 2.jpg
I'm often looking at a portafilter and wondering "what capacity size is this basket?"
An idea I had was to laser etch the number, LARGE, in the bottom of the basket. The laser slightly pits the stainless steel surface, causing the number to be legible. It shouldn't affect the basket's functionality.
However, something interesting happens when the laser intersects a hole. A black carbon "explosion" appears on the underside of the basket. I *think* this is caused by some of the very thin stainless steel of the edge vaporizing. Or something else?
At any rate, the black carbon does start to clean up with a rough sponge, but it's really hard to get into these nooks and crannies and clean it all out. A mechanical polishing might be better but probably wouldn't get all of it. Waterpik? Fine sandblasting?
I put the basket on the light table to see if I could discern any hole size difference, and I couldn't. However, the black marks are pretty clear to see and unsightly.
While the black marks probably won't affect the functionality of the basket, they're likely would make customers nervous. Before I go ahead with this labeling idea, I'd want to fix the black mark problem.
I'd love to hear your thoughts on this.
I agree, it would be better if marks weren't visible. No option to chose a font/shape that doesn't intersect with the holes but is still readable/recognizable to the human eye (a bit like a captcha)?
Mark it on the side rather than the bottom.
Java "Sideways thinking" phile
Toys! I must have new toys!!!
blast the reverse image into the bottom of the basket so both sides are homogeneous.
or you could mark the lip. a tiny number, or a colour-code/pattern.
Thanks for the tips. I think VST might use ink rather than a laser to label the basket. I'm going to talk to the basket engineer and see what options are open to us.
I've now saved my "best" shot on the shared office DE1+ as a "god shot". You can see how this feature plays out in real use. This afternoon's espresso (just made it) had slightly less pressure, a bit faster flow into the cup, but similarly accurate temperature, to my reference.
Result: this shot tasted good (nothing really defective with it) but not as good as the really good shot I made two days ago, which I saved as my GOD SHOT.
This makes me wonder, is there any sort of room for machine learning here? God shots would tend to be repeated with the same coffee, dose, etc wouldn't they? Could you help the machine to learn to better match the recorded data next time? This also makes me wonder if this could be achieved with other saved shots by tying into the record of beans and grinders used etc. Probably DE-CAF territory if manageable.
This machine learning ability lets us get the puck to your destination temperature much faster.
Below are two graphics, showing before vs after. The cooling effect of the portafilter and room-temperature beans is, I believe, more dramatic than people realize, and the machine learning feature has had a dramatic effect on how quickly the puck slurry reaches +/- 1ºC of goal.
Is it only for the temperature? Is it long term learning that adjusts over time and keeps learning? Is it across the board learning so using different beans might "pull" it in different directions?
Pressure sensing is extremely low latency, so not likely to improve with machine learning.
However, measuring flow with low latency is really difficult, because flow meters that operate on espresso-levels of water speed (1 to 10 ml/s, say) give only a few pulses per second on the low end of those flow rates, and giving too much meaning to a pulse can lead to erratic flow readings.
Here's a screen picture below of the kind of problems with flow meters when reading them over short time frames. I suspect this might be why so few espresso machines actually show you instant flow readings. Or else they smooth the data like mad, averaging a lot of data samples. That gives you a nice average number, but at the cost of latency.
The flow-measuring model we're just about finished moving to is having a physics engine count pump strokes, and track how much water we think was moved for each stroke at that given pressure (more pressure=less water per stroke). We then use the flowmeter to regularly calibrate the physics engine.
Machine learning kicks in to help us get lower and lower latency flow measurements, so that we hope we can show you sub-second channels appearing and healing.
The goal of all this tech, at the end, of the day, is always for me about making a better shot.
Very interesting but how's the production schedule going?
We're down to the details now, such as what color coding scheme to use on the water tubes. Each tube has its own two-color scheme, so it's easy to look inside the machine and find the start and end of any tube. We're using colored shrink tubing to label them.
Sounds good. Wife is getting antsy as it will be her machine. I must keep her happy!
We're working on many fronts at the moment, trying to perfect the 200 or so parts that are arriving in quantity=300 to make our first batch of espresso machines.
If you'd rather not "get into the details" know that things are proceeding well, with the so far the excuses and delays from suppliers only being in the one-to-two week scale. Everything still looks like it will arrive in December, but now it's late-December rather than first-week-of-December.
UPCOMING SAFETY CERTIFICATION AUDIT
For UL (USA) safety certification, we have two more steps:
1) a two-week detailed audit by the testing company, to ensure that nothing appears wrong in what we're doing
2) thereafter, we build 8 identical machines that are "destructively tested"
We're preparing for that audit now, and in double-checking everything, we found 3 small things that need fixing.
The biggest was that the "green light" on/off switch we use, has to be swapped for one without a light. We're using that switch to turn off both line voltage (110V/220V) and a 24V power supply, and it turns out that the with-a-light version of this switch wasn't approved for such use. A few hours ago, the light-free switch was wired into the machine sitting on my desk.
The other two safety certification issues are what anti-fire certification level we need around the insulating foam in the group head and around the circuit board compartment. Intertek's engineers are researching the question.
Once we pass UL certification, we'll go through the EU certification process. There's no sense in doing both in parallel, since the EU standard is mostly a superset of the UL standard.
CAST DRIP TRAY COVER
Next, we received the drip tray covers that are now cast (instead of being carved out by a computer). The good news is that the fears we had (caused by other suppliers not wanting to take the job) were unwarranted and there aren't any significant problems.
However, where the two molds meet, you can see a "parting line" where a thin amount of metal seeps through. This needs to be reduced, and then manually removed with a bit of sanding.
Also, the grade of aluminum used for this sample was not what we asked for, and the finish is correspondingly dull and tin-like.
So, we're asking for another sample to be cast, with the correct metal and with the parting lines manually removed.
FITTING THE CASE ON
Finally, the rounded cover that goes around the case, doesn't fit all that well on the two samples we've received. A flat piece of sheet metal is first cut and then bent, to make the chassis. Rounded corners are apparently difficult to get just right, so we're adding a bit more space to them to help the cover go on smoothly. The several-millimeter gap potentially caused will be hidden the reflective splash panel (on the front) and the back plate, so this solution should be fine.
Thanks for the update, I always appreciate the TLDR versions.
Here's a photo, below of the back of the DE1, and I've put arrows on the rounded corners that are causing the problem. They're coming manufactured to us, just a little bit too large, which makes the cover really hard to screw down. This ugly back panel will be covered by a dark-plastic, semi-translucent cover (you can see the standoffs for it mounting it, in the photo).
However, your general question (vibration causing noise) is a concern, and our usual solution about areas we worry about, is to use double-sided sticky tape covered 3mm foam, and then screw things down between them, tight but not too tight. At the moment, nothing is vibrating, but it's something for all of us to watch and to come up with repairs if it pops up as an issue.
I previously mentioned that we were working to fix a problem with the cover not fitting well on our espresso machine chassis. To make 100% sure that the problem is solved, we’re now paying the sheet metal company to completely assemble our chassis, so that they can confirm (before shipping to us) that everything fits. We shipped them last week a completely assembled espresso machine that we put together, they could copy from.
We’ll need to disassemble the case a bit here in Hong Kong, in order to put all the espresso machine components in. However, I think the slightly wasted effort is worth it to know that all the metal parts really, truly fit together. Not just once, but for every complete chassis.
Attached are some photos we received this morning from our sheet metal manufacturer in Shenzhen, showing what they’ve done. They’re sending it to us today, so we can take it apart and make sure they assembled it right, and that all the little problems we spotted in the past few weeks have truly been resolved.
WechatIMG1190.jpeg WechatIMG1191.jpeg WechatIMG1192.jpeg WechatIMG1193.jpeg
Glad to see diligent QC on this one and hope they do get the cover fit just right and tight/vibration free.
If there is a fit or ratting problem, we have a solution, which is to put a piece of slotted silicone rubber tube around the corner, like a "gasket" but for a corner. We have 50M of it in stock just in case.
Two thousand thermometer probes arrived today. We use 6 of these in each machine (they're very useful). That can cause a problem: so many probes and wires that look identical, how to make sure you plugged the probe into the right place? That problem is compounded by the fact that the probes all plug into a single circuit board, all in a line.
That's why we're putting a color-coded fabric mesh tube around each cable, with each cable getting its own color combination. This stretchy tube expands to 2x its size and then stays put on the cable. We can make colored drawings that show how everything should plug in together. Someone nontechnical can then be assigned to check that everything was done correctly.
In the photos, you can see the thermometer probes, two photos showing where they plug in, and the colored tubes we're looking at using.
600 water heaters arrived today (in each espresso machine: one for hot water, one for steam). We were waiting to receive them before we ordered the parts to make the insulation box around each. We were worried that some dimensions might have changed between "our receiving the sample" and "our receiving the final product".
The good news is that "nothing changed" and so now we'll have the boxes in about a week, and we can start assembling them.
Just wondering when we can expect these machines to be available to Australia? I'm in the market for a new machine soon and the de1 cafe has caught my eye(want to be able to steam whilst pulling a shot). Will this machine compete with the likes of a gs3MP?
However, a bit of bad news on the DE1CAFE. We had planned on having two plugs, one dedicated for steam, and one for espresso. I've just been informed a few days ago from Intertek that this design is not acceptable under EU rules. So... we're going to split the steam out into a separate "slim line" module, but that likely won't be out until Summer 2018 (as it's a new product, with all that entails).
The DE1CAFE coming out this spring will have a much higher power steam, from a custom heater we've designed (and being 220V only) however it will not steam during brew. The reason is that there isn't enough electricity for us to steam on demand during an espresso, and give you consistent steam (steam quality would fluctuate). We don't want to make a "steam boiler" machine, for many reasons.
1) spring 2018 for a DE1CAFE but high powered steam after brew
2) summer 2018 for a dedicated steam module
3) summer 2018 for a DE1CAFE with no steam (optional)
A good year ago, we all had a lively discussion about how to charge the tablet that sits on our espresso machine. What kind of connector? Waterproofing? Standards? Aesthetics? Permanent or removable?
In the end, we went with the standard "USB-A to USB-Micro" cable that virtually all Android phones use. We found a rubberized "panel mount" water resistant plug for the front panel of the espresso machine.
For the USB cable itself, I wanted something as invisible as possible, and just the right length. We only needed a 6cm long cable, whereas the shortest off-the-shelf cables were 14cm (photo shows ours vs the standard cable). The too-long cables looked really sloppy. I also didn't like the design "features", with logos, multiple colors and shapes, strain relief circles. Give me Simple, please.
So, we designed our own USB cable.
It uses a minimalist "box" shape for the side the plugs into the tablet (USB micro), that you can easily grip because of its size. For the USB-A side, a simple shape, no logo, is all we needed.
It wasn't easy finding a company that would make our cable because this is so inexpensive a part (about USD$1 each) that minimum order quantities are usually 10,000 cables.
Because we went with a common standard, you can (a) use your own cable if you prefer, such as to have a longer cable, or (b) use your own USB charger, if you prefer.
Or, just use our cable, which is an appropriate length for its intended use, and of a fairly invisible design, that it fits in.
1500 of these cables arrived today.
I've been looking at your web site and notice that the DE1 comes with a naked portafilter. Is there an option for a 2 cup portafilter, or must you split the shot yourself, or make separate shots?
To save John (DE) the trouble and because I think that people who cannot be bothered to read the whole thread to get info do not deserve to waste his time at this point of final production.
1) Yes, John has already posted that you can have a "splitter portafilter instead".
2) Why would anyone who wants a quality cuppa bounce their crema through an extra set of bumps? Crazy IMO.