decent_espresso wrote:Apologies for slightly off topic, but I just had one of the better espresso of my life at "The Cupping Room" in Hong Kong, and well, I just feel a need to share!

Upon leaving, I noticed a sign that the barista that I had been congratulating is actually the HK 2013 Barista Champion, and 2013 runner-up World Champion! Whoops! As if this guy needs me to tell *him* his espresso is good! At least now I know I have a decent palette!


<image>

That guy was Kapo Chiu, he has two Cupping Room shops now. He is not only an amazing barista, he is a great guy and has really been a big part of helping to bring up the entire coffee culture in Hong Kong and Macau to amazing heights. 10 years ago there was hardly a decent coffee to be found in Hong Kong, now there seems to be an excellent third wave shop every few blocks.

+1 Decent Espresso

Bak Ta Lo wrote:That guy was Kapo Chiu, he has two Cupping Room shops now. He is not only an amazing barista, he is a great guy and has really been a big part of helping to bring up the entire coffee culture in Hong Kong and Macau to amazing heights. 10 years ago there was hardly a decent coffee to be found in Hong Kong, now there seems to be an excellent third wave shop every few blocks.

Bugs (my gf) and I have been going every morning. He is remarkably consistent at making the top shots I've ever had. Today was the first time he wasn't there, but the shot quality was unchanged.

However, we went to his other shop in Central today, also using a Black Eagle, and the shot was noticeably less good: less balanced, more acidity. Good, but not great.

I suspect at the main shop that the grind, temp, pressure are all dialed in, and need frequent adjustment, while the other shop doesn't have the Guru there to do this.

What I love about espresso is that you can "taste the talent" !

decent_espresso wrote:
What I love about espresso is that you can "taste the talent" !

Or lack there of! I was at a shop today that had good beans and high end equipment but no barista! The guy who made my "espresso" and "pour over " had no clue. He just dumped a bunch of water into a dripper full of a random amount of ground beans.

Then he made an "espresso" by grinding a random amount into the basket tamped as hard as he could 5 times while knocking the basket in between each time very hard.

I would have rather had Satrbucks! Seriously.

RyanJE wrote:Or lack there of! I was at a shop today that had good beans and high end equipment but no barista! The guy who made my "espresso" and "pour over " had no clue. He just dumped a bunch of water into a dripper full of a random amount of ground beans.

Then he made an "espresso" by grinding a random amount into the basket tamped as hard as he could 5 times while knocking the basket in between each time very hard.

I would have rather had Satrbucks! Seriously.

Crazy thing is that a "Starbucks" customer would have watched him and been like.

"Man look at that guy taking his time and doing all that tapping.... now that's a real barista, he must really know what he's doing back there!"

And he probably really thinks he's hot stuff and takes himself "very seriously."

Back on topic though. I love the new wave of espresso machines being linked to other technology. This is the "next wave"

What's new with the machine build? Any pics or shot vids for us?

They have been posting a fair amount on Facebook lately. They have had delays with factory location issues, etc it looks.

I'll wanted to give HBers an update on the state of our espresso machine project.

In October, we demonstrated a prototype of our espresso machine at a trade show in Portland, Oregon. With Ray (lead internals engineer) controlling the prototype with his laptop, tweaking parameters, it made good espresso, and people were happy.


ACHIEVING THE DESIRED WATER TEMPERATURE

The "trade show prototype" worked by carefully controlling a water heater, with room temperature water as the input, to output water at the set point of 92C. Heating was "on demand".

There were some downsides to this approach:
- the parameters needed to be careful tweaked, by hand, to get consistent results. We tweaked the parameters to make one shot after another, consistently.
- However, if the machine sat around unused for a few minutes, the water would be too hot, forcing a cold water flush of the heater, which took about one minute. You had to wait.
- We were nervous about how consistently we could achieve the desired setpoint, in the real world, with this approach.
- We couldn't make "Americano water" at (say) 60C without cooling the water heater off a lot, which would throw off the accuracy of the next espresso shot.
- When things worked well, we were within 0.8C of the set point, which is very good accuracy, but we didn't have high confidence in this approach working consistently "in the field".

In the weeks after the trade show, we prototyped a new approach, which involved overheating the water a little bit higher than the desired temperature, and then mixing in a bit of cold water to bring it down to the desired temperature. We were able to achieve a worst case of +1C over, and -3C under, the set point temperature, in varied circumstances, with no flushing time (ie, one shot after another, as well as with random pauses).

Those error numbers weren't good enough to ship with, but they're promising. The water mixing approach in our 2nd prototype was fighting the fact that we had plastic tubing which stretched to 2x its internal diameter under pressure, as well as a not very good flow meter. We've decided to go ahead and further test the water mixing technology with a more serious 3rd prototype, which we'd label as an "Alpha test".

We believe that if properly executed, the water mixing approach will give us everything we want (no delay, accurate espresso water temperature, and also accurate Americano water temperature).

We're currently:
- sourcing high quality sensors with an appropriate cost to be able to be used in manufacturing,
- replacing all the tubing with a molded, massive piece of Teflon, which should eliminate the "swelling" issues and simplify construction
- we're designing the circuit board which will provide the CPU power to achieve this mixing (no more Arduinos).
- we're writing the firmware to execute on this plan
- There is significant CAD, circuit board, and software work to be done, which is what we've been focussed on.
- We'll be able to test this approach in early February. If it works, we'll be fleshing out the firmware and tablet user interface through February.


PRESSURE PROFILING

For variable water pressure ("pressure profiling"), our trade show prototype used a DC vibratory pump, and we were able to extremely precisely control it via PID. That prototype pump cost us $400, despite being almost the same as an AC vibratory pump (typically $40 in single units), but we had assumed we'd be able to find a reasonably priced one in China for manufacturing.

Unfortunately, after contacting dozens of pump manufacturers, we were never able to find a DC pump appropriate to use in production under a cost of $80 (we need two of them), which would make it the most costly part of our espresso machine.

So, we're currently designing an AC control circuit for a standard AC "Ulka" vibratory pump. This pump is an "old faithful" pump that both Seattle espresso repair shops we talk to recommend we use, as they said they've both very reliable, and very easy to source in case they do break.

We think we have an approach that will achieve similar control with the AC pump that we had at the trade show with the DC pump, but we won't know for sure until we actually try it, which is what we're building toward now.


HEATED GROUP HEAD, WATER PATH

The big surprise for us was how well our designed-from-scratch group head worked. We really studied all the existing professional espresso machines we could find in the Seattle area, and innovated in a few important ways. We're electrically heating the group head, PID controlled, to the water set point. This has worked perfectly so far.

An innovation built into the trade show prototype was that when you hit "make espresso", water exits the heater at the desired (measured) temperature, and then is re-measured as it is about to go into the group head (a few inches further). If it has been cooled by the tubing and connectors, the water goes back to the tank.

This "last minute heating" of the entire water path to the group head takes about 20 seconds, but it provides absolutely on-the-button water temperature right to the group head. When the water temperature behind the group head is exactly right, it is then allowed to enter the group head, which is preheated to the water set point temperature, and thus suffers no change in temperature.

I believe that this combination of technologies (last minute path heating and PID controlled group head heating) is likely a key reason we were making good espresso at the trade show.

Pro espresso machines often use a "saturated group head" to try to achieve similar ends, but I do not think they measure and wait. My La Marzocco GS/3 has a 2.6C "adjustment" set at the factory for temperature loss from the boiler to the group head, to compensate for this effect. If the compensation never needs to be adjusted, that is a perfectly fine approach, but I prefer to measure.


STEAM

We haven't yet worked on creating steam, because we've been fixated on the making-good-espresso problem. However, this next round of circuit boards and testing (early February) will have our v1.0 steam approach, and we'll see how well we're doing with that problem. Our physics model shows us that we'll be able to heat a standard latte (200ml) of milk in about 45 seconds, which is consistent with what other 110V machines achieve. Under 220V we should be able to achieve higher steam pressure and volume.


MANUFACTURING AND HUMAN ISSUES

All our engineers have now moved from Seattle to Hong Kong. We're now housed, and as of January 1st, we'll be moving into a 3200 sq ft workshop where we can manufacture our machines. We've recovered from our ill-fated plan to be based in China (now: parts sourced from China, assembled by us in Hong Kong). We've decided to take control of our own manufacturing, and do it in house. We've made our first Hong Kong local hire, and we think we can scale up the needed labor, at a reasonable cost, to be able to pull off as building as many as 10,000 machines per year ourselves, if there is enough market demand for our machines to warrant it.



CONCLUSION

We are still in R&D and we plan to be through to the end of February. We plan to make a small number of beta units in March. Depending on how that goes, real manufacturing will start in April/May.

While we did make good espresso in October at the Portland trade show, we weren't at a point where we could convert that prototype into a machine that could be manufactured. Parts of our R&D are at a "we can built it" stage, but other parts need more work.

So there you go... a big update. Any questions?

Have you thought about using a thermoblock like a J-boiler to preheat the water instead of heating it in the reservoir? My brother is using that technique in a Gaggia Classic with very good temperature consistency. He uses PID for control of the output water but if you use both the input and output temperature it would be even better.

The reason I suggest this is to do you will reconsider heating the water in the reservoir. I had a plumbed in E61 espresso machine which kept water in the reservoir. I found that the reservoir came up to be quite warm. It ended up allowing some kind of gross stuff to grow. That really turned me off of a heated reservoir.

Just an FYI, the GS3 has an offset to account for the entire measurement system. You will also find this advantageous if you are using precision probes. Or you can go with an expensive 1% variable resistor on the PCB and test each board for accuracy and precision.

Check patent work on water Mixing for temperature in beverages before you go all out. You may need to search a bit inside a few other patents to find this in their claims. To get great results you will need a fast acting probe with a very very low time constant. After that its all software.


Bill