Decent Espresso Machines (DE1) - Any thoughts?

[TampIt]

Every scale I've seen has a noise filter, displaying 0g of weight until some threshold is passed, otherwise ambient vibrations would cause the scale display to blink annoyingly between 0g and 0.1g. Ours is no exception, in the same way that Acaia and Brewista do.

However, at the bluetooth level, our scale's firmware is sending the weight 10x per second, without a noise filter. The first drop doesn't show up as a weight, but rather as a shock wave, a very quick transient spike on the load sensor. Since the scale is being read by our tablet over bluetooth, we can detect that noise, even if the display still reads 0g.

So, yes: we can detect the first drop.

Now, I know this for sure because I've written a flow meter for an earlier iteration of the scale, (video at https://www.youtube.com/watch?v=-arEs5jq09A) and a later version of that app (not in the video) had charts on it. In that software, I had two lines: the 10x a second weight, and a moving-one-second-average line.

What was really interesting in those lines is that channeling appears to show up as increased noise on the load cell, because the spurts of water shoot out with force. I haven't figured out the maths yet, but I think there's a way for scale software to understand the "evenness of an extraction", ie an absence of noise.

In theory, a less noisy espresso shot should taste better.

I wasn't go to comment on these thoughts, however it may save you some time / grief or at least exclude another possibility.

The idea that channeling could potentially be identified as "shot noise" is a (welcome) revelation to me. About time someone stated how two similar shots on the same gear can be so radically different in taste.

(taken from the DE Grinder thread)

- If your grinder is junk, you won't be able to make good espresso, even with the DE1. Thus, I want my customers to be able to buy an affordable grinder that I know will allow them to make good espresso.

- if you're a beginner and you don't weigh your grounds, you're going to have trouble achieving consistently good espresso

- the Baratza Sette: there are things I like about the Sette, and for the size, styling and price, I think it's unbeatable. However: our use of the Sette with the DE1+ finds that if you dose directly into the portafilter with the Sette, you will get bad channelling. Dosing into the Sette's plastic hopper, shaking, and then tipping that into the portafilter, produces quite good shots on the DE1+. This is fairly objective, and repeatable, as we can see the channeling as loss of pressure throughout the shot.

- if you are willing to dose into the Sette's plastic hopper, then the Sette is a fantastic choice, and I would not want to offer a product to compete with it. The grind-by-weight feature, in my experience, is mostly accurate to within 0.2g, which is impressive.

I was taught in the late '70's to set gear up by putting about 2mm of "coffee fluff" in the ("home made naked") portafilter and lightly (2 to 3Kg downforce only) tamp. Care should be taken that the first tamp should be absolutely level, with no signs of breakup in the puck.

Repeat every 2mm of "coffee fluff" / lightly tamp until the dose is correct. Redistribute if there is any sign of crevices in the puck.

Polish the final one and pull the shot. Whether polishing actually does anything I will leave to others to prove / disprove. Given decent gear the shots always taste brilliant. Oh, the shots channel slightly about once every blue moon.

Then Schomer came out with his 30lb / 13Kg single tamp. It never really worked for me. Sprayed everywhere too often for my comfort and tasted a lot worse than my usual tamping method.

Then there was an experiment with a 300lb / 135Kg tamper which showed that even at such an extreme pressure it was only the top 1/4" of the puck that was compressed. That made sense, as the lower 50% (roughly) of the grounds were a free for all which could channel freely.

Then LM came out with the swift grinder - which (more or less) tamps in 1mm stages at 8lbs / 3 Kgs force. The output from a swift channels slightly … about once every blue moon.

So, John, I suggest you try my tamping method and see whether it removes the “shot noise” you detected.

Hope this helps (or at least gives you something else to play with).


TampIt

[>decentespresso]

De1 progress report

Hi guys, we're basically in "tying up all the details" mode here, with lots of things that still need to be finalized:
- vibration dampening technique for the pumps
- rubber foot design
- final material of the drip tray & water tank (ceramic is really, really heavy, like 2kg, which is terrible for us since we're air-shipping these : it's +$20 in shipping for those, +$18 to make them. Too expensive.)
- testing new mixing chamber design that has integrated almost all connectors into it, testing new "ultem" material
- testing new approach of "clip connectors" everywhere for the water path, no more compression fitting or expensive (and failure prone) quick connects
- new design for thermometer probes so they look clip connector compatible
- on the firmware side, lots of hard work around implementing a file system and a firmware update system, and some crypto to prevent chinese copying of the firmware. Almost done with that, needs to be tied into the GUI.
- on my end, finalizing the gui, which involves increased polish, support for multiple DE1s in a location, support for load/save of profiles, translating it, and skin support. I also have a few things to do on the DE1+ gui based on what I have learned about our espresso making workflow with our gui in the past 4 months of making shots daily.

Below is the DE1 settings gui, for changing the pressure profile, and for a variety of other settings. This week I should make good progress on load/save of profiles.

[roburu]

That's good to hear. Any rough estimates on how long it will take to get the first non beta version out?

Also one quick observation on the timer, a bit more flexibility would be useful (please ignore if it's already planned). I used to have my machine on a timer and I had 2 timers for week days: 6 am to 10 am and 4 pm to 10 pm. For weekends I had just one timer from 7 am to 9 pm.

[>decentespresso]

That's good to hear. Any rough estimates on how long it will take to get the first non beta version out?

We're still looking at June for these machines to be manufactured and released in non-beta.

Also one quick observation on the timer, a bit more flexibility would be useful (please ignore if it's already planned). I used to have my machine on a timer and I had 2 timers for week days: 6 am to 10 am and 4 pm to 10 pm. For weekends I had just one timer from 7 am to 9 pm.

I had an earlier version of this with a kind of grid, where you could set on/off at any interval during the day (and multiple intervals) and I can see how that would be useful. For now, though, I need to cut back on features and complete what we do have. The kind of timer I just described (as well as your request) should be doable as a plugin, and that'll motivate me to get the plugin architecture done.

[>decentespresso]

Here's a short movie of us testing the button interface to our upcoming Decent Scale: https://decentespresso.com/scale

The idea is for one button to control the timer (on the right) and another button to control the tare weight feature.

There are a few button presses too many on the timer, and the "weight currently changing" LED should stay on for a second longer (since espresso can drip in slowly), but otherwise this is close to OK for me.

Any comments?

[>decentespresso]

one quick observation on the timer, a bit more flexibility would be useful (please ignore if it's already planned). I used to have my machine on a timer and I had 2 timers for week days: 6 am to 10 am and 4 pm to 10 pm. For weekends I had just one timer from 7 am to 9 pm.

As it stands now, there's one interval. I've had 3 other people ask for a weekday/weekend scheduler. At the same time, I'm trying to keep the initial feature set down as much as possible, to essentials, so that we can ship this machine. The tablet software will have successive versions, where I can add things that are 'nice to have' but not essential. With a 5 minute warmup time, I was thinking that a power on/off timer was already kind of inessential.

For the eventual power on/off feature, I had already designed about a year ago something which let you have multiple on/off times in one day, and a different schedule for each day even.

Below is the design mockup.

You can see that it's a lot more powerful, but it's also more work for me to program. So.. this is my intended (maybe version 3) scheduler.

[MrJack]

Here's a short movie of us testing the button interface to our upcoming Decent Scale: https://decentespresso.com/scale

The idea is for one button to control the timer (on the right) and another button to control the tare weight feature.

There are a few button presses too many on the timer, and the "weight currently changing" LED should stay on for a second longer (since espresso can drip in slowly), but otherwise this is close to OK for me.

Any comments?

It looks like:

• 1st press: Start
• 2nd press: Stop
• 3rd press: Reset

Would it be less complicated to use a press for start/stop (i.e. pause/unpause) and press+hold for reset?

[>decentespresso]

It looks like:

• 1st press: Start
• 2nd press: Stop
• 3rd press: Reset

Would it be less complicated to use a press for start/stop (i.e. pause/unpause) and press+hold for reset?

I'd rather not require people to learn to hold a button down for a while to have a different action. In my ideal world, one button does one thing.

My plan, though, is to change the 3rd press into a reset. The workflow then becomes:

1st press: timer LED turns on and starts counting time
2nd press: timer stops, LED stays on, final time is displayed
3rd press: timer is reset, and starts counting time (ie, step 1)

In other words, the timer button is a START-AT-ZERO/STOP button. There is no feature to unpause, since I think this is not a very used feature.

[>decentespresso]

designing a decent knockbox

About two weeks ago, we finalized the design for our Decent Knockbox https://decentespresso.com/knockbox and sent it off to a CNC service to make us one.

This one prototype costs about $120 to make. It's expensive, because it's carved out of a solid block of aluminum with a computerized drill. However, it's invaluable as part of the design process, because it lets us hold and use something that is very close to what the final factory-manufactured good will be like.

Instead of plastic, we opted for 2.5mm thick aluminum, with a black anodized finish. The goal is for it to be sturdy, really dishwasher safe (which plastic often isn't) and to not have that cheap "thonk" sound when used. Anodizing provides a similar surface to teflon, though it's more durable and less slippery.

Design wise, the goal was the deeply carve out the front so that a portafilter could be held level as it knocks. Some knockboxes don't have this feature and require you to tilt the portafilter forward or risk banging your knuckles. The back of the knockbox is scooped up to capture "splash". The bar is pushed forward because the spent pucks pile up in the back, and when I used knockboxes with bars in the middle I found that I filled the knockbox up when it was only half full, because the pucks didn't spread out in the container.

We'll be testing this prototype this week and next with different types and thicknesses of foam glued onto the bottom. The "knock bar" on this prototype is borrowed from another knockbox, as we don't have a way to make a prototype of that part without making a silicone mould.

Once we're happy with this, we'll send it the final specs to our factory, and in about 80 days we'll have them in stock.

Below are two photos of the real object, along with a computer-generated render. It's always interesting to compare the idealized perfect version off a computer to the real thing. In this case, they're not too different.

[>decentespresso]

New drip tray design

The new design for our drip tray cover came in today from our CNC manufacturer. This version is a hybrid of types "1 and 2" in that from above it behaves like round metal wires, but viewed from the bottom you can see that it's flat. There is no coating on this, nor has it been polished, so it's likely more water retention-friendly than a final product might be.

I poured water all over the top, to see what droplets would stick the top.

The ability of this design to repel from the top is fairly good:



but viewed from the bottom, the water droplets do like to cling:



however, this is still much better than our previous design, which I photographed today:



My feeling is that the performance of this design is likely good enough, and if we were to spray the bottom with a water-repelling coating, that behavior might be improved too.

A big caveat, though, is that we don't yet know if this design can actually be manufactured at a reasonable cost. This CNC prototype cost us $80 to make. We're going to talk to factories over the next few weeks, and we might end up with a related design, of welded 3mm stainless wires inside a metal frame. The appearance of a drip tray with that approach will be similar to this CNC prototype, from the top, but from the bottom we would get less droplet retention if the wires were round on the bottom too.

WHAT DO YOU THINK OF THIS DRIP TRAY DESIGN?

[gc]

About two weeks ago, we finalized the design for our Decent Knockbox

Below are two photos of the real object, along with a computer-generated render. It's always interesting to compare the idealized perfect version off a computer to the real thing. In this case, they're not too different.

[ATTACH=CONFIG]15278[/ATTACH]

[ATTACH=CONFIG]15279[/ATTACH]

[ATTACH=CONFIG]15277[/ATTACH]

I like the computer generated one.

Not a lot of scope for originality with knockbox shape I suppose, but that one would look fine next to my machine.

With such a narrow base is it stable on the bench?

[>decentespresso]

Not a lot of scope for originality with knockbox shape I suppose, but that one would look fine next to my machine.

We searched far and wide for inspiration, but it wasn't easy, because at its root a knockbox is a rubbish bin with a bar across it.

When I started looking at Champagne coolers I found designers trying to communicate the sexiness of champagne in their forms, and I was really floored by this design for Moet Chandon by Jean-March Gady: Jean Marc Gady for Moet & Chandon (NOTCOT)

Our design uses many of same design tropes that Gady's does, though implemented differently (we don't need a lifting handle, for instance).

With such a narrow base is it stable on the bench?

That was a very real concern with this design, but looking at the physics, we thought we'd be fine, since the force of the "thump" channels toward the center with this shape. If the bar were further back (ie, in the center) that could have caused a problem. In use, so far today, the knockbox hasn't "rocked" from use, but I'm definitely watching for that as a problem.

[roburu]

We searched far and wide for inspiration, but it wasn't easy, because at its root a knockbox is a rubbish bin with a bar across it.

Looks quite similar to a knockbox I used to have, can't remember the brand, however I found a photo of a similar one: http://www.espressoworkshop.com/imag...300.jpgWM.jpeg . Your rounded design looks nicer though.

In use, so far today, the knockbox hasn't "rocked" from use, but I'm definitely watching for that as a problem.

I found the wide base on mine was useful as without it I'm pretty certain I would have spilled it a couple of times. Spilling a full or nearly full knockbox is a pretty big thing especially if you have carpet near the coffee preparation area as I (unfortunately) do.

[roburu]

WHAT DO YOU THINK OF THIS DRIP TRAY DESIGN?

[ATTACH=CONFIG]15284[/ATTACH]

It's pretty decent

[>decentespresso]

HEATED GROUP HEAD and the question of a PLASTIC VS METAL GROUP HEAD COVER

Jeremy from Blossom coffee, who is consulting for us, expressed concern about the long-term reliability of our using Calrod tube Calrod Heater - Heater Elements - Wattco.com to heat our group head.

This is what a Calrod looks like:


Probably, we're ok, because they only really heat up when the DE1 powers up, but the problem is that it is hard to shape the heating tube to exactly fit into the group head's moulded channel, so that a lot of heat transfer paste is needed to mate them, and that if this isn't done right, or the tolerances are too far out, the Calrod tube will slowly fail over the course of a few years.

However, besides the "probably ok" there's the inefficiency of the heat transfer that's been bugging me for a while. We're heating the group heat, to indirectly heat the water path.

So, instead, I decided that we should spend two weeks on switching to directly heating the water path, with a much smaller heater that will be mounted directly on the brass "shower assembly" where the water actually ends up, right above the espresso puck.

Here is what a ceramic heater looks like:



A few benefits from this approach:
- we can use less electricity (200W vs 800W) to heat the DE1 to ready
- we should be able to heat up faster (less mass to heat up now)
- we will have much less waste heat (less radiating off the group head)

Which means that we no longer need to use plastic to make the group head cover an insulator around the group head. There won't be enough heat radiating off it to burn yourself on.

So, four things going on this week:
- some group head internals redesign to move off of a calrod heater, to a ceramic heater (mechanical engineering, David)
- electrical calculation changes and some software programming to support the change (electrical engineering, Ray)
- design thinking into what a stamped metal group head cover should look like. (industrial design, Joao)
- getting pricing to manufacture this part (Alex and Johnny)

From a designer's standpoint, I've never been happy about having a big blob of simulated-metallic-finish plastic on the group head, right where everyone can see it, especially since we spent so much time and money using metal and wood everywhere else, to produce a high end machine.

So, in our next rev this will change to metal, and in the next 2 weeks I'll be able to post here a design for that group head cover, which we can make in metal.



As an aside, another little sourcing detail for us has been how to tidy up that USB charging cable on the tablet, and where should the cable go.

This week Jeffrey found a supplier for flush mount charge-only USB cables. They're virtually invisible, and look like this:



We need to have a custom cable made for us, and either snake it through a slot in the front panel, or through a screw hole. That'll be much cleaner looking.