Decent Espresso Machines (DE1) - Any thoughts?

[Magic_Matt]

Will there be an easy way of renaming/grouping/tagging the logged settings file that's created with each shot setting? eg "a really great shot#1xxxdate". I realise this can be done with any file manager once save location is known, but within the app would be nice

+1, simple nickname and favourite functions would vastly improve the usefulness of this feature I reckon.

[>decentespresso]

+1, simple nickname and favourite functions would vastly improve the usefulness of this feature I reckon.

Agreed, as well as a scoring system, maybe even tasting notes. I've got a lot of feature requests on my plate, so for now we're just saving every shot and enabling you to bring it back, but over time I will add more to this section. Ideally, you'd be able to indicate when kind of beans, grind setting and dose weight to the tablet as well, so that you can track some of the more important variables needed to recreate a a great shot.

Re: GOD SHOT, yes that's exactly what the "background drawn reference shot" is. In lighter colors, your reference shot is charted, and ideally I'd like to give you a "how close to God were you?" score for each shot you pull. Where this is most helpful is in adjusting your grind to compensate for weather and bean aging.

re: Interns. Our interns are paid AU$8.50 per hour, and our internship program was certified/approved by a professor, so that the interns also earn school credit for their summer work. They also get money to buy lunch. We bought of them a new chromebook. We treat them well, not only because it's the right thing to do, but because we want a good relationship with the local engineering universities.

My goal is to roboticize much of the DE1 assembly over the next two years, and we'll need clever people to help with that.

[MrJack]

Here are a few visualisations I produced using data extracted from the images that John posted.

1.
The first shows plots of flowrate, pressure and a calculated volume parameter (i.e. cumulative sum of flowrate * timestep) for all three shots.
These profiles were quite different to what I've seen before, which were from lever machines and had volume profiles resembling exponential growth.



2.
The second is the same three shots, but this time I plotted pressure (which is equivalent to pressure drop) against flow. This is an indicator of the total resistance to flow (a function of viscosity, puck depth and grind size).

This plot clearly shows three distinct "phases" during the shot, where the slope of the curve changes. I suspect the phases are:
1) filling the piping / group prior to water reaching the puck (pressure drop driven by frictional losses in piping)
2) saturation of the puck (pressure drop increases relative to flowrate as water moves further into the puck)
3) extraction post-saturation (pressure drop decreases relative to flowrate as espresso becomes less concentrated / viscous.

Looking forward to seeing some video of the shots, to corroborate my assumptions.



3.
The third plot is an estimate of viscosity during the shot, based on Darcy's law, with several assumptions (i.e. 58mm basket, 20mm thick puck and a constant puck permeability of 10^-14 m^2, which is at the lower end of reported measured values. I was extremely surprised to see the estimated viscosity was within the ballpark of what I have seen reported in scientific papers (0.5 - 1.4 mPa.s)!


4.
Last (but not least) is an aggregate plot of all of the above, for the first shot only. There is additional information encoded in the colour of the scatterplot markers (as indicated on the individual plots).

[Shortyesquire]

re: Interns. Our interns are paid AU$8.50 per hour, and our internship program was certified/approved by a professor, so that the interns also earn school credit for their summer work. They also get money to buy lunch. We bought of them a new chromebook. We treat them well, not only because it's the right thing to do, but because we want a good relationship with the local engineering universities.

My goal is to roboticize much of the DE1 assembly over the next two years, and we'll need clever people to help with that.

[ATTACH=CONFIG]16210[/ATTACH]

Australian graduates would have wanted beard trimmers and moustache wax rather than a chromebook...

Seriously though, I'm super excited for you. Once I start lining up our new house there will be a de pro or cafe finding a new home.

[>decentespresso]

We've upgraded to a high end "pick and place" machine to make our own circuit boards for our espresso machine. We like the control it gives us, to know that every part on the board is the one we chose. It lets us make short runs and also create small circuits for specific tasks (such as for a testing station) . Even better, the machine pays for itself with the first 1000 espresso machines we make, bringing the per-espresso-machine board cost down from $90 to $50.

We're calling this machine "Peter", as in "Peter Piper picked a shed load of components almost 500 and made them into PCB boards".

Here's a picture of a Very Happy Jeffrey with his new toy. You can see the puny old machine we previously used, to its right.

[MrJack]

Following on from the images in my last post, I was thinking about the relationship between pressure drop and puck permeability this morning and thought I would plot the viscosity of water (at 90C) against the estimated viscosity based on Darcy's law (assuming constant permeability). Another surprising result!

[dlight]

So what does this mean?

[MrJack]

Viscosity tells us something about how much coffee is dissolved in the water in the puck. So, it potentially offers a window into the real-time extraction rate! There are a number of factors (temperature, CO2 evolution & changing permeability) which would have to be accounted for though.

[rheasman]

Following on from the images in my last post, I was thinking about the relationship between pressure drop and puck permeability this morning and thought I would plot the viscosity of water (at 90C) against the estimated viscosity based on Darcy's law (assuming constant permeability). Another surprising result!
[ATTACH=CONFIG]16254[/ATTACH]

I think perhaps the relationship is not a simple linear relationship. I think there may be two factors:

1: A square term relationship to flow rate due to fluid inertia. Perhaps Darcy–Forchheimer applies?

2: Whether there is a puck at all (coffee grinds swirling around and not yet settled, vs untamped but settled, vs tamped)?

[LeroyC]

Viscosity tells us something about how much coffee is dissolved in the water in the puck. So, it potentially offers a window into the real-time extraction rate! There are a number of factors (temperature, CO2 evolution & changing permeability) which would have to be accounted for though.

I think you'd need to account for a lot more than temp, CO2 and the change in physical conditions of the bed during extraction to be able to extrapolate any info about the extraction rate from viscosity. There are way too many other variables for this to be of any real use.

[rheasman]

My back of the envelope calculation puts the Reynold's number for fluid flow in the coffee bed at around 500 times the flow rate in g/s. Darcy's law is generally only considered appropriate at low Reynold's numbers (less than 10). I think "normal" flow rates range from 1 to 6 g/s when making espresso.

It is possible my calculation is wrong. Any attempts to double check me would be appreciated. :-)

[>decentespresso]

If it's not yet evident to those on this forum, Ray Heasman (who just joined this forum) is the internals engineer / inventor at Decent Espresso.

[>decentespresso]

budapest demo & manufacturing

Here's a short video Jasper Coffee took of yesterday's DE1+ demo with Scott Rao in Budapest. Best coffee shots so far on this machine, but that's because it was Scott's own Kenyan roast beans. Both the flow profile at 2.2 ml/s and a 9-bar-flat shot delivered good (but different) results.
https://www.facebook.com/jaspercoffe...0586847650013/

and meanwhile in Hong Kong, we're taking over 3000 sq ft (about 300 sq meters) of new space next door, to dedicate it to manufacturing, because our first space space (R&D, accessories, lunch!) is now crammed full.

[>decentespresso]

Large Ceramics are Difficult

We've been working for the past 4 months with a ceramics factory to make our water tank and drip tray out of porcelain. The sample they made for us were great, and the price was good (about $7 for both parts). Unfortunately, they've now tried twice to make the parts using the "mass production process" and both times, what we've received has been warped enough that they won't slide into the metal rails of our espresso machine.

The problem, and we've seen this before, is that samples are often made using a process for one-offs (pressure casting), but after you order, a different manufacturing process is used for quantity (slip casting). You don't necessarily get the same result.

We had a much more expensive bid from another company ($23.70 for both parts) that we'd passed on previously, but their quality was great, so we've gone back to them. We've decided to use them now, so that we don't add more delays, and keep working with "slip casting" ceramics companies to see if someone can make our vessels well and cheaply with that process.

The lead time of this new company is 60 days, which unfortunately means that the ceramics will arrive mid-to-late August. That's going to (sigh) put out our DE1 deliveries until then, since an espresso machine that has no water tank is of limited use.

With luck, this will be the worst delay we encounter, and everything will have arrived by then, and we can just slip the ceramic parts in.

Sorry for the bad news, guys. Manufacturing is not easy.

[>decentespresso]

This is the first photo I've posted of our new internal design as a real machine. We totally reconceived the insides after we found our November machine wasn't at a high enough evolutionary level.

In this photo, you can see (1) the machine opens up quite easily for complete access (2) the silicone insulated hot water and steam heaters (3) totally redesigned and much lighter group head [for faster temperature change responsiveness] and quicker warmup (4) much more internal space and reduced complexity.

We still have some work to do to organize the tubing better, to test and re-test everything again before we "pull the trigger" on the sheet metal order (which FYI takes 35 days to manufacture) and the mixing chambers (which have been refined to lower mould costs).