MrEd wrote: John Buckman just finished the 'advanced shot editor' that allows for features like pausing. Thus one could test adding enough water to saturate the puck and then waiting a predetermined time before moving on to extraction. This might more closely mimic that aspect of a lever machine.

Cool! I'll be anxious to hear what water volume saturates what puck volume, realizing this may also include headspace?

Tonefish wrote:Cool! I'll be anxious to hear what water volume saturates what puck volume, realizing this may also include headspace?

No doubt that this will include headspace, but the puck generally swells to consume the headspace, so this volume may not need much correction to accurately represent what is held in the puck. The water path between the primary manifold and the group head, along with the brass distribution chamber behind the shower screen is the only volume I would subtract from the measured volume.

Ben, any chance you can comment on an approximate volume required to fill these areas? I'm guessing it's in the 10-15ml range, but this could be way off.

Another approximation would be to measure the discharge from the exhaust valve after a shot. This would likely yield a pretty realistic "functional headspace" volume that could be subtracted from the flow recorded up to the wetting of the basket to get a feel for actual water retained in the puck. Another method would simply be to pull the portafilter after wetting the basket and weigh the darn thing! (Extra points for putting it back in and pulling the shot with detailed notes of the experience! )

- Jake

Jake_G wrote:Another method would simply be to pull the portafilter after wetting the basket and weigh the darn thing!

With the advanced shot editor, one could insert a lengthy pause to allow time for removing the portafilter and weighing.

Jake_G wrote:Ben, any chance you can comment on an approximate volume required to fill these areas? I'm guessing it's in the 10-15ml range, but this could be way off.)
- Jake

I can get you the volume from the grouphead valve to the top of the shower screen no problem. But this volume is not dry after the shot, the tubes etc will not drain completely.

I've been planning to put separate scales under the water tank and drip tray. This way we can see exactly how much water is used for a shot, how much enters the cup, how much goes through the flush to the drip tray and how much remains in the portafilter.

Also, I could give the total volume of an 18g basket minus the grouphead protrusion (so the volume of an 18g basket under the shower screen). From this, you could experiment with another vessel and work out the maximum preinfusion volume possible to fill the basket (puck + headspace).

Once I work out the volume of water required to fill the water path we can then work out the theoretical maximum pre-infusion required to completely saturate the puck. Giving us something to aim for when pulling shots.

---]The above text was somehow removed from the bottom of the previous post and appended here -Ed[---

Jake_G wrote: Did you happen to notice when first drops appeared using this method?

I ran a similar shot to the no preinfusion one this morning.
The profile was set to 6 ml/sec preinfusion, then 2 ml/sec extraction. 19g in an 18g basket. First drops at just under 2 bar pressure. And despite the traces looking good, the shot was pretty bitter. I suspect that is because there is almost 5 seconds of high flow rate extraction occurring before it hit the 2 ml/sec target.

MrEd wrote: I ran a similar shot to the no preinfusion one this morning.
The profile was set to 6 ml/sec preinfusion, then 2 ml/sec extraction. 19g in an 18g basket. First drops at just under 2 bar pressure. And despite the traces looking good, the shot was pretty bitter. I suspect that is because there is almost 5 seconds of high flow rate extraction occurring before it hit the 2 ml/sec target.

Still looks like about a 2x dose flow to first drops [(6ml/s x 7s)-(~3ml for the startup flow curve corner)]=39ml Really appreciate your sharing this data.

THis is very similar to what I got 37ml PI volume for 18g dose.

vit wrote:...these pumps have their max. capacity depending on the pressure. So for Ulka pump (here it's different manufacturer, but is similar in construction and working principle), it's something like 400-600 ml/min or about 6-10 ml/sec at 0 bar, decreasing with pressure. However, on your graph, it stays at 6 although pressure is rising + we have 2 pumps (although hot water pump doing most of the job I suppose), So I'm wandering is it maybe some software bug (or feature) preventing higher flow at beginning, which would make this ramp up time shorter (if desired - for simulation of a machine with fast ramp up time) although I think it's unlikely to make it in only a few seconds

Turns out the issue here is not the pumps, but rather the amount of power it takes to heat the water on demand to brewing temp. A flow rate of more than 6 ml/sec would draw too much power and blow the breakers. On the other hand, having on demand heating allows the machine to be at temp in 4 minutes from turn on. Everything is a trade-off.

MrEd wrote:A flow rate of more than 6 ml/sec would draw too much power and blow the breakers. On the other hand, having on demand heating allows the machine to be at temp in 4 minutes from turn on. Everything is a trade-off.

I had read that the DE1 originally would preheat the tank water to 50°C, so the change in temperature would only be ~45°C. It appears they removed this feature, though.

Edit:
I don't believe the heater can't keep up (except for 6mL/s), but likely the algorithms need tweaking. I'm sure Decent will work on adaptive models to minimize temperature drift even further. This is still a LOT better than I manage with an Arduino.

I believe they are budgeting for a 75°C temperature differential, i.e. from room temp, which would push things at 6 ml/sec. However my understanding is they've also found they can store quite a bit of heat in the aluminum casting surrounding the heater, which allows them to burst heat for up to 15 or so seconds at higher rates. John just told me that he thinks they might be able to push it to 8 ml/sec, so long as this wasn't maintained for too long. These values are all specific to the 120V machines. I think the 220V machines have slightly less heating capacity.

Thanks for clarification, so it is indeed software limit.

I think that ramp up time could be decreased (maybe to about the half) by modifying algorithm so that machine starts with say 10-20°C hotter thermoblock (depending on desired ramp up time), with higher flow through it and adding more cold water with other pump to achieve desired water temperature. This way, thermoblock temperature will be decreasing, which would be compensated by decreasing amount of added cold water, until desired pressure is reached

I was experimenting with preinfusion time on my manual espresso device before buying Flair, consisting of moka pot and strut pump, where, due to the small capacity of the pump, I also needed 7-8 s to achieve 9 bar. My conclusion back then (on limited "sample size") was that lighter roasts benefited from longer preinfusion, but darker / comfortable blends gave better results if it was kept as short as possible. So I think that a possibility of making this shorter might be useful for some coffees / owner's taste