Tweet
Our knockbox design came in today, 50% larger than the last one, CNCed from two blocks of aluminum and then welded together. It's a bit of a Frankenstein, but it's useful for judging relative size and functionality. This is way more solid than the earlier model. The capacity is maybe a bit more than is typically needed. And on my first "knock" it I was surprised to see splatter so high up on the back wall of the knockbox. We'll play with this prototype for a week, and if we like it, it's off to manufacturing.
-
-
"Secondly, the plan also covers digital imports which currently evade GST. Netflix, for example, serves content from Australian soil but claims it is an offshore vendor immune to GST."Originally posted by gc View PostComment
-
LONG: why will it take so long to get my Decent Espresso Machine?
Yesterday, I received this question via PM:
I thought it was a very good question, that likely others were thinking, and so I asked the questioner if I could answer him publicly. Here goes. It's a long explanation. Sorry about that.I saw yesterday you mention that a realistic delivery date for regular release DE1's will be October now. Does the same apply for the DE1+? I ask because the website currently mentions a June 30 shipping timeframe for the DE1+. I was planning on pre-ordering one next week (when a paycheck arrives). It probably won't sway my decision, but a four month difference in shipping is something I'd like to plan for.
WHY DOES IT TAKE SO LONG TO BUILD AN ESPRESSO MACHINE?
In November, when we last built 8 machines, it took one person 5 full time days to build one Decent Espresso machine.
There are a few reasons why this was the case with that November design, :
- each building step was sequential, you had to do one, to do the next. That's slow.
- physical access to the machine wasn't easy as things got packed into a small space
- many steps were fiddly and required time to get right, such as mounting the steam wand through the faceplate in a way that was tight and not leaking when steaming
- there were a lot of steps and many weren't easy, some requiring quite a bit of dexterity and assembly experience.
For these reasons, and also because I felt the design of the internals was lacking the refinements that were needed, I sacked my mechanical engineer in January, hired a new one, and we've spent the past 4 months totally redoing the internal design to fix these issues.
With the new design:
- there are a half dozen sub-assembles, which are clusters of components that can be built ahead of time, and tested separately too.
- getting into the machine is much easier, because there are panels where things mount that are removed from the machine, and screwed in at the end
- hopefully, most of the fiddly, hard-to-get-right problems are fixed
We're estimating that each machine will take a full person's day to make.
WHEN WE TRANSITION TO BUILDING MACHINES:
Some facts to bear on our transition to manufacturing:
- we're currently a R&D company, and we have zero "factory workers" on staff who might build machines. To fix this, we're planning on hiring 8 Mechanical Engineering students as summer interns to build machines with us from June 1 to August 30. That gets us "smart people" to build the machines, and 3 months to transition to less highly qualified staff to be building them. I'm also hoping that these students will, by virtue of their studies, have insights for design improvements.
- our office/R&D space (3200 sq ft) is almost totally full at the moment. We have another 2500 sq ft of warehousing, but that's full now too with parts. Why so much space? Because when you order a part (say, tablets) you have to order the minimum order quantity (1000 of them) and they show up in boxes, stacked up on pallets. You have to store them, and they take a lot of space. In June, we're taking out a lease on another 3500 sq ft, directly opposite our front door (6 ft away) and if orders start coming in over the summer, we're also going to take a lease out on another 6000 sq ft (around the corner). We're pretty organised, but it will take a bit of time to fit out the new spaces.
- we don't yet have all the parts in stock, and some will certainly take longer to arrive than others. To work around this, we'll start building and testing sub-assemblies in quantity=200, so that when the parts all finally arrive, we're mostly done.
- each espresso machine will be tested for 3 days before it ships: 24 hours on a shipping simulator machine, and then 48 hours of constantly making espresso and steam. This should expose most problems.
WHAT TAKES SO LONG TO BUILD A DECENT ESPRESSO MACHINE?
Building an espresso machine is similar to building a Tower PC clone machine. You buy parts, you put them into a chassis. However, there is one big difference: with a PC clone, all the parts come ready to be screwed in. With our espresso machine, lots of parts come and require more work from us before we can use them.
AN EXAMPLE: THE HEATER SUB-ASSEMBLY
We have two heaters in each machine (espresso and steam) and here are the steps for getting one heater ready:
- put a temperature probe bead in the center hole, and back fill it with thermal putty
- screw in two thermal fuses for CE compliance
- screw in one thermostat for UL compliance
- use pliers to connect locking power cables to fuse 1, to fuse 2, to thermostat, to electrical in, and also to ground
- place entire heater (with all the fittings) suspended in a 3D printed mould of our own design
- fill the mould with quick drying liquid silicone. Wait 30 minutes to dry. Remove and use a knife to clean up stray silicone. Let it further dry another 2 hours.
- slide two two teflon tubes into the two water connectors (in/out) and use pliers to force in a tiny clip to hold them in place
- put a cable tie into a teflon tube and wrap it around the heater, and then tie it to a metal panel
- hook up the water lines to a water pressure line, and test under 15 bar for leaks
- hook up the electrical lines to power, flow water at a known rate, and test that water is being heated at the rate expected
- repeat again 399 times to make enough heaters for the 200 espresso machines that have already been ordered.
Lots of steps! However, this approach is a good one because:
- it's a major component and it can be built and tested totally independently from the rest of the espresso machine
- it can be done in quantity, alone, so the people doing it can do it faster.
- if we can guarantee quality, we can sub-contract out this assembly in the future. However, it has so many steps, and so many ways to screw up, and it is so vital a part, that for now we prefer to assemble it ourselves.
SO: WHY OCTOBER TO RECEIVE MY MACHINE?
Firstly: because there are 200 people ahead of you in the queue, and if we make 4 machines a day, it will take us 50 work days to build them, which is about 2 1/2 months on a 5 day work week. We're ordering parts now, but realistically of them won't arrive until the end of June, and no espresso machine can leave here until every last part is in stock.
Secondly: we don't yet know how many man hours it will take to build a machine. We estimate it around 1 man day at the moment. If we can reduce that to half a day, that doubles our production speed. I don't think we can reduce it much below that, however, unless we contract out parts of our subassembly to other factories (effectively adding manpower).
Thirdly: we need to hire a lot of people, train them, and have good processes for them to follow. If we do this too fast, we'll have bad staff, poorly trained, with bad processes.
Fourthly: the order backlog is growing each week, and each new order goes to the back of the line.
TO CONCLUDE
This summer, we'll be making espresso machines.
However, we can't go from "no machines per day with no factory workers" to "40 workers and 40 machines a day" in one step. Hiring and training and process-making will be incremental.
We can't afford to sign leases for 3200 sq ft + 2500 sq ft + 3500 sq ft + 6000 sq ft (4 leases) unless we have some idea of demand. Growing capacity too fast is a good way to go bust.
I expect that this summer, if our espresso machines are well reviewed, that orders will increase. We have a plan to cope with that demand, which you've read above. But, it will take a few months for us to deploy that plan.
If you've been reading this forum for a while, one fact should have popped out at you: EVERYTHING DEPENDS ON INITIAL REVIEWS. That's why we're moving slowly, to ensure that from the very beginning, the machines that people receive are good, and people are pleased.
In practical terms, what I expect is that if our espresso machines are well received, that they will be in short supply (like any popular new product) for the first 6 months, as we grow capacity to meet that demand.
I hope this (too long) answer was interesting, and I'm happy to answer any questions.Comment
-
USB charger for the tablet
We've found what I think what will work well for providing a USB plug on the front splash panel, right near where the tablet needs to plug in. These two models below are:
- minimally visible
- the one on the right is watertight, the one on the left is probably water tight enough
- USB-A, so we stay don't stray from the USB standard
- the aesthetics of the two vary slightly
We're ordering samples of these two today, and will likely settle on one of them.
Comment
-
One week to design freeze
The current design of our Decent Espresso machine.
Recent improvements:
- hugely increased the "twisting" strength (aka stiffness) of the case, with additional bent panels (in the middle, on the top, under the group head)
- hot water return is now as far away from the intake as it can be, so that calcium should precipitate during initial water heating and fall into the ceramic water tank where it causes no damage and can be rinsed out.
- we've found an excellent, high speed temperature sensor (incredibly, with 0.2mm thick stainless steel walls) and now can stop using our hand-made-by-us sensor design (which was expensive and very time consuming). This has allowed us to add two more temperature sensors to the main mixing chamber.
- however, the main mixing chamber now became too big to mould, so it's been broken down into two parts, which is why that part on the top (with all the bolts on it) has that odd shape.
- most of the water tubing is now in CAD
Still to be done:
- a USB charger plug needs to be installed on the front panel (to the left of the group head)
- our flush diffuser design needs to be finalized (it's too small at the moment)
- the "cool touch" group head cover needs to be mounted
- a panel separating the circuit boards from the machine needs to be installed
- the "over pressure safety valve" integration needs to be finalized (that's the part floating above the bright green part)
Looking at this CAD model now, I'm pleased that we took the extra months to rethink and/or refine everything. Nothing is crammed in, the wire paths are not hard to follow, and the design is fairly mature.
Comment
-
Insulating our water heaters in silicone.
What we're planning on doing--which I believe is a first in espresso machines--is to fully wire up our heaters and then cast the whole assembly in a silicone mould. The goal is to insulate the heaters incredibly well, as if you had the world's best fitting "oven mitt" around the heater. Some espresso machines do wrap their heaters with insulation (many don't) but that leaves air gaps and is messy to fit properly (you have to use a die cut sheet and tie it around).
This silicone casting idea was originally prototyped a month ago, with a mould made of legos (top of photo). This created an amusing pattern on the resulting dried silicone (bottom of photo).
Last week, Alex designed a proper mould for our heaters and it arrived today. The advantage of this mould is that it keeps the wall thickness constant at 6mm all around the heater, and holds it in the right place.
Here's what the first mould looks like, freshly CNCed from a block of solid aluminum:
- - - - - -
In other news, we posted an ad yesterday at a local engineering university for engineering students who might want to build espresso machines this summer. We received 12 applications in 24 hours! We interviewed two today, and they were impressive, with really solid CAD experience, as well as computer programming, and Arduino experience too. This is great, because it means we can put the students in charge of creating testing stations for each part.Comment
-
working toward even water flow in the group head
Ben and Ray have been working pretty intensely for a few weeks on new designs the group head. The goal is for water to be evenly distributed as it comes into the top of the chamber, and then again evenly distributed as it comes into contact with the puck. Even water flow is important for even espresso extraction.
Today, we received one sample each of Ray's design (using calibrated orifices) and Ben's design (using turbulence).
The "apple" shaped parts are for the top of the group head chamber (where water enters) and replace the common espresso design of a single hole. Ben has been using flow simulation software to test all these designs, and I've been posting the graphics here over the past few weeks. The turbulence approach performs a bit better in simulations.
Next, we'll be testing both of these designs in the real world, and choosing one for our first two hundred machines. We might still improve this design over time (or you might, starting from our shared CAD drawings) as Ben convinced us to make these parts modular, and thus inexpensive to CNC.
You might want to compare the design of our parts to these Rancilio parts, that provide a similar function. Our computer fluid simulations found that this sort of design (which we were using a few months ago) resulted in uneven water flow, especially during preinfusion. That's why we've been working on this so much, as it seems like a previously-neglected area of espresso machine design.
Comment
-
Testing our flush diffuser
We're trying to slow down the high pressure water after an espresso shot, so you don't get splashed when the water goes into the drip tray. Here's a slow motion video of our prototype idea, where water goes into a kind of cyclone, and can't exit until it slows down, because the hole is in the opposite direction of the cyclone. This is one of the last things on our "to-figure-out" list before our design freeze in a few days.
Comment
-
I worry things like this will eventually clog with gunk and be a maintenance issue. Any thoughts?Comment
-
clogging worries
It's definitely a possibility, yes. We're going to keep the plastic cover on it, easily removed, even for customers, so you can see if there's any clogging and easily clean it. You can get at this part by simply removing the top cover, no other disassembly needed.Originally posted by readeral View Post
However, group head flushes tend to be "grey water" and not so particulate heavy. In that video, that's real flush water you're seeing, from an espresso that just finished.
I didn't mention it, but that diffuser has been in constant use for about week, and it's still pristine.
In case you're curious, here's where the flush diffuser will sit, it's quite easily accessible.
The diffuser is located in such a way that if it really clogs constantly, and that just can't be helped, then a longer tube from the flush valve can be swapped in and send higher velocity water directly into the drip tray. If that's ever needed, we'll send you the part and you just need a pair of needle nose pliers to take the clip out, put a new tube in, and reclip it.
Comment
-
I have never been splashed by water when pulling a shot on my rocket gppComment
-
Connecting everything up
One of the really annoying things about building an espresso machine is that there is no standard way to connect the water tubes between things that do stuff to water. Pumps, heaters, flow meters, valves, steam wands: they all use different fittings.
There are lots of different "standards", most of which are fiddly, can leak if they're not installed perfectly, and they're expensive. How expensive? Well, each connector on our first prototype machine cost $9, and we had 15 of them. $135 in connectors in a machine that sells for $999 is not great. And, one of them completely broke during my two month press tour in November. Marvelous.
And then... there are these connectors that have no name.
You only find them when you open up some models of amazingly super-engineered Nespresso machines. We call them "clip connectors" and they are solid teflon tubing, with a metal sleeve inside, and another one outside. Once the tube is made, it simply pushes in with a rubber gasket, and clips in, and... they seem to never leak, they rotate, they're super easy to install and remove. They're amazing!
What's the catch?
The only hitch is that they are totally proprietary and not available off the shelf. Even the tool for making the tubes (mounting the outer metal ring) isn't available for purchase, and it requires exactitude and a fair amount of compression and force.
Sigh.
We did find one company that would make the tubes with the ends on them, and since nobody will sell us the tool to make it ourselves, we're using them to buy the tubes ready-made. A bit more expensive, but they're our only source at the moment.
For converting between other connector types (such as to the steam wand, or the water heater) to these clip connectors, we've CAD designed our own converters. Today, 50 samples of these two converters arrived (they're the shiny things) and you can see them mounted on the back of the steam wand (first photo) and on both ends of the heater (bottom photo). We're sending a dozen of them to our heater manufacturer, so they can give a price quote for welding these on for us.
Happily, it looks like our connector-adaptor designs are working, so now we can use these great water tubes virtually everywhere.
Comment
-
Just about there...
We've just about wrapped everything up we're now going to build one final "release candidate" machine, to make sure everything we've changed is ok. That will be ready in about 14 days. Everything that is certain to not change will continue to be ordered.
The only two things left to nail down are:
- the exact shape of the USB plug on the front panel (choosing between USB-A or Micro USB-A)
- the exact shape of the flush diffuser (the rectangle in the bottom right of the chassis).
Here is a render (in false colors) of the final design of the internals.
and here is where we're putting the USB plug, so that a very short cable can connect power to the tablet.
Comment
-
Help me explain these features better
We're proofing the translations of our web site into French and German, and in the process, finding confusion around some terms I use.
There are two features in particular, which are new to espresso, and I haven't found a good way to name them. They currently are.
" Basket temperature goal mode"
"Automatic channel healing"
Can you read along, and please suggest ways I could describe these concisely? I feel like both of these are really important new things to espresso, so it's pretty essential that we communicate them.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
"Basket temperature goal mode"
In every other espresso machine, you set the water temperature, and that's what the boiler heats to. Say, 90°C water. But, by the time the water leaves the boiler it's gone down about 6°C, sometimes more, sometimes less (depending on the machine). And then the espresso coffee grains themselves are at 20°C (room temperature) and when they make contact with the water, the slurry is a temperature which mixes the water and the coffee (and potentially the cooling effect of the portafilter if it wasn't preheated)
I've made this ugly chart to show what we're doing. Other machines have a constant water temperature, so they have a linear climb toward their goal. Our DE1+ machine can dynamically vary the water temperature, to get the puck temperature to the goal faster.
I like to think that this feature will arguably make better coffee, since different products are extracted from coffee at different temperatures and there really are different flavors at different temperatures. People really fixate on water temperature for pour-over coffee for this reason.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
"Automatic channel healing"
When espresso is being made, sometimes these channels open up (sometimes called "crevasses"), where water spurts out. This (a) tastes bad and (b) makes a mess. Here's a photo:
when we detect a jump in water flow, we back off the pressure immediately, which generally causes the coffee to fill in and for the channel to "heal". We can only do this because of our use of vibratory pumps, which are like pistons, because we can instantly stop pumping and start again 1/10th of a second later. Traditional pumps can't do this.Comment
-
(choosing between USB-A or Micro USB-A)
USB-A is tough, never seen a damaged one.
Micro USB-A is always damaged, stretched and sometimes stuffed after the plug is jammed in the wrong way.
Anyway up... USB-C is a good connector too.Comment
Comment