Robo-Pavoni ; learning to make espresso machine from scratch

[matadero210]

Hi All,

This thread is a continuation of my work with the PID Pavoni: http://www.home-barista.com/forum...-begins-t3767.html

It is also a break from my second thread, http://www.home-barista.com/forums/home-made-grinder-advice-t3880.html, while I finish the drawings.

In the PID Pavoni thread, I added a PID controller and brew-pressure gauge to my Europiccola. The main defect in that work was the need for an expansion tank in the water circuit. When the lever is raised, about 45ml is displaced from the head back into the boiler. This is, in turn, displaced back into the expansion tank, where hot and cold water may mix. In the robo-Pavoni design, the boiler has enough dead airspace to eliminate the expansion tank. Thus, no mixing and temperature instability.

The other design feature is stability. The Pavoni is not really up to the rigors of 9 bar. It flexes and tries to move. A second hand is nearly always used to hold the machine in place. This design seeks to allow 1-handed operation.

The design itself: take a Pavoni head, build an adaptor to connect to normal plumbing. Add a machined plate with water inlet (this is a plumb-in system), thermometer, and heater. Mount all in a brazed steel base.


This picture shows (left to right) the SSR, the Pavoni head with brew pressure gauge (this is a different head, the PID Pavoni is in service), the PID controller (from auberins.com, no affiliation), the chassis and boiler. Note that the colors are all messed up: the boiler is copper, the base plate brass, and the legs steel.


This picture shows the plate with mounting holes (4), recesses for brazing the boiler, and three threaded holes for water inlet (1/8npt), thermometer (1/8npt, PT100 RTD from auberins.com), and 1kw heater (1/2npt, mcmaster.com). The PID controller is 1/32--a real tiny one. Note how badly I suck at brazing. One of the purposes of the project is to sharpen my fabrication skills for the grinder project above and for building my dream machine from scratch.


This picture shows the adapter for mounting the Pavoni head on the 1.5" k-type copper pipe boiler. Note that the final through-hole will be drilled only after the adapter is soldered or brazed into place. The Pavoni head will screw into the adapter just like on a normal machine, but the adapter will be soldered onto the boiler just above the 2x1.5" flare, about 1/3 of the way up the boiler. The steel chassis is only partly brazed together--I ran out of oxygen over the long weekend and had to stop.


This picture shows the bottom of the plate and a nice view of the brew-pressure gauge.

[matadero210]

For those who are interested, I'm using qcad, a nice open-source auto-cad-like program. The adaptor flange and base plate drawings:

[hbuchtel]

Cool, I'm looking forward to seeing the end result!

A couple questions-

How are you going to control the amount of airspace in the kettle?

Why did you decide to go with a tall and thin kettle?

Where along the body of the kettle will you mount the group?

Will you try to control the direction of the cold water flow as it enters the kettle?

Have you thought about setting up a pavoni-head-gauge-adding service? People could send you their pavoni groups, you could send it back with a gauge attached...

Henry

[matadero210]

Hi Henry,

On to the questions:

1. I'll continue to use my regulated cold water supply, but without expansion tank. Its usually set for 7-10psi, or 0.7 bar. In that case, under half the boiler will be filled with water. As long as the water level is above the group attachment point, both the Pavoni piston and backfill ports will be immersed in hot water. The head itself will also have a good thermal connection. The airspace can be controlled by varying the cold water pressure regulator. I'd rather not add a schraeder valve and bike pump....

2. I'd rather have the kettle thicker, but off-the-shelf copper pipe doesn't get much bigger. Also, water level might get rather close to the ground (leaving less space for the cup). I might consider using 4" cast iron instead, but I wanted to keep everything metalurgically compatible. If I went with cast iron, I'd need SS fittings and a dielectric between the head and boiler. 4" brass or copper pipe is $$$$.

3. About 1/3rd up from the bottom is where the group will mount.

4. No. What I learned from my previous machine is obvious in retrospect. Once the machine fills and heats, operating the lever just short of opening the piston fill port, what I called 'dry pumping', allows circulation of hot water behind the piston, heating the group. When 'dry pumping', the air gap expands and contracts and the pressure varies, as 45ml is pumped in and out. The air gap is designed to be at least 4x45ml. Then, when ready to pull a shot, the lever is raised all the way up. That displaces 45ml to the boiler, which then returns to fill the piston. Only when the lever is depressed (and the piston disconnected from the boiler) does cold water enter the boiler. So--is a check valve needed at the cold water port? Maybe, but the regulator on the cold water port seals up when back pressure is exerted, so only a tiny amount of convective mixing will happen. I'll use tubing rated to 90C. I am slightly concerned that my thermometer is very short and near the inlet. I have longer, t-type thermocouples, but I wanted to experiment with PT100, which Suyi and auberins.com claims is the most accurate/sensitive for this application.

5. I have. This one I nearly spoiled, but caught just in the nick of time. So it would have to be on a best-effort, if-I-screw-up-you-are-hosed basis. The second one took about 1/2 the time as the first (30min compared to over an hour)--I'll bet I can get it down to 15 min with practice.

raj

[espressme]

Thanks for going where no man has gone before!
sincerely
richard

[matadero210]

Hi All,

The new system is up and running. Firsts shots are excellent, although there is at least one variable yet to dial in.





A couple of construction details:

1. Originally I planned to solder the flange, but many thought it would be too weak, so I brazed it instead.
2. You can't really see them, but fiberglass washers are used (and SS bolts) to thermally isolate the hot brass and copper from the cool steel frame.
3. The PT100 thermister does give +-0.2C after the inital auto-tune. I hope for slightly better after I calibrate the 100C to boiling and tweak the PID parameters by hand.
4. The brew-pressure gauge is the same design as my previous PID Pavoni system.
5. There is no thermosiphon. The brew water is in direct contact with the (back of the) head, and even without dry-pumping the head sits seemingly at brew temperature.
6. There is enough room (~130ml) of expansion space in the top of the boiler to allow use without an explicit expansion tank. Note in the photo the regulator is now by itself. The system is connected with a sanitary quick-disconnect between boiler and regulator. I'll put an OPV in there too when it comes from mcmaster.
7. The anodized aluminum controller box houses an SSR, the PID controller, and the power switch. Its pretty tight in there, but I should add a fuse. I always use GFI outlet for espresso machines.

In operation, I noticed the same thing as in the previous system: if the regulator is set too low, the pre-infusion is fine but the shot is short. I think it takes a modest pressure to force the air down and out before the puck seals. This is a Pavoni design flaw. In my next machine the water inlet will be at the bottom of the piston, not the top.

Obviously I need to tend to some cosmetic details, but I must say that I'm very pleased with the system. The major design goal, stability, is fully achieved. You can pull a 10bar shot WITH ONE HAND!

raj

[hbuchtel]

That is fantastic!

How about a video?

I think you are mistaken about the air getting trapped in the piston chamber. If you think about it, the water will go straight to the bottom of the chamber, right? Once it does so there is no way for the air to get out ... unless it is good at diving I can only guess that the considerable amount of air which is trapped in there just gets compressed... which might be part of the reason you are seeing a difference in volume with the different preinfusion pressures.

Regards,

Henry

[matadero210]

Hi Henry,

I think you are exactly right--my point is that Pavonis always work that way, both in my system and out-of-the-box. I've never understood why or how it works in either case, just that I observe that more pressure yields 45ml shots, less pressure yields 30ml shots (1 pull).

Will work on the video,

r

[hbuchtel]

Hi Henry,

I think you are exactly right--my point is that Pavonis always work that way, both in my system and out-of-the-box. I've never understood why or how it works in either case, just that I observe that more pressure yields 45ml shots, less pressure yields 30ml shots (1 pull).

Will work on the video,

r

Sorry, I misread your original post, re-reading it I see your comment "before the puck seals." I'm curious though, why do you think having the inlet at the bottom of the piston chamber would help?

Maybe you could drill another hole for air release like you did with your pressure gauge? One with a little valve that would seal at a pre-infusion pressure...

Henry

[matadero210]

Hi Henry,
In my ideal machine, the piston inlet and outlet would both be at the bottom of the cylinder, both with check valves. That way, raising the piston would suck water into the piston (no air), and lowering the piston would pump water out. The grouphead check valve could be a flapper, if the piston is in the grouphead, or a normal check valve if the piston is isolated from the group. This would only work for an unpressurized system (otherwise, the positive pressure would always flow out the group).

r

[matadero210]

I added mineral insulation to the robo-pavoni. It now sits with 0.1C stability, with a shot recovery time of 1-2 minutes. There's a quick snaphot at my grinder thread http://www.home-barista.com/forums/home-made-grinder-t3880.html. I hope to add a splash guard or boiler cover, but its working fine without.

raj

[espressme]

I added mineral insulation to the robo-pavoni... snip...but its working fine.
raj

Wonderful prototyping! Thanks!
richard

[arriflex]

Love it!! I've got a cnc mill that I'd really like to use to build a head since those don't exactly come up every day. Any ideas about what you'd want to see in such a project? Or for that matter have you seen anything like it done? I've been following the progress on the portafilter build http://www.home-barista.com/forums/olympia-cremina-67-portafilter-t4167.html as well. I'll keep all drawings and progress public if I actually get around to doing something other than 'thinking off.'
arri

[Chert]

Some people dream of good ideas and realize the simpler ones (blokes like me). It looks like matadero210 manages even the complicated ones. Now if only Expobar or the maker of the Vivaldi - not La Marzocco as it would be impossibly expensive - would buy Raj's idea from him and make the temperature-and-pressure-dialable-manual (Why not dual head, one manual and one spring operated) lever wondermachine! Great thread.

[mhoy]

Any updates on this amazing machine?

Mark

[matadero210]

Here is the complete kit, fully working and tested. From the right:

1. Robo-Pavoni. This is a pavoni head bolted to a home-made copper boiler and brazed-steel frame. The head has been drilled to show piston pressure (dial). The boiler is insulated (kinda ugly, but not to me) with fiberglass and has an anodized aluminum spray skirt (not cleaned in this photo) in case of shooters. The PID is on the back and is normally set to 90.5C. The system is plumbed directly to my cold water house supply with a 1.7bar regulator (with no back-pressure relief). When the piston is raised, the pressure increases to 3bar for the the initial pre-infusion as the gas bubble trapped in the top of the boiler compresses. Final pre-infusion happens at 1.7bar. Typically the boiler temperature fluctuates during the pull, but since the water is already in the piston this does not affect the pull temperature. By the time the shot is pulled, the boiler is back at 90.5. System takes < 4 minutes to be ready in the morning and can pump shots until the Hetch-Hetchy runs dry. Dry-pumps can circulate boiler water through the head before shooting. I also can keep the lever raised and flux liters of water through for daily cleaning.

2. Typical shot of beans on the 3rd day. On day 1 I get way to much crema.

3. Zassen-Robur. This is a single chunck of brass milled into a hand-crank grinder. The burrs are from the Mazzer Robur, the crank from an old Zass with broken burrs, and the rest from McMaster-Carr. The adjustor screw is 100tpi on a 3/1 lever, giving 80um/turn of adjustability. Its too heavy and hard to turn for my wife but it fits my hand just fine. About 30 turns for a 1.5x shot (what Pavoni calls a double, but I use a single pull with jiggles).

4. WDT tool and naked portafilter.

5. 1.6kg stainless tamper. One side is slightly larger, for the first tamp. Tamping is done by resting the tamper on the basket filled with grist, giving a half twist, and repeating with the narrower side. The pressure comes entirely from the 1.6kg--I don't push.

6. The newest member of the team, the Roaster. The guts are from a $10 popper from target, coupled with a multi-step, ramp-soak controller from auberins.com (same as the Robo-Pavoni's simple PID controller). One switch is used to toggle fan speed (low and high speed) since the beans are larger/lighter near the end of the roast. The T thermocouple is lightweight SS jacketed and at the center of the roast chamber. The screen in the background is the chaff collector, placed over the unit in operation. Typical roast is 80g: 4 minutes between 125C and 140C, then 8 minutes ramping from 180C to 230C, for Full City. Cooling is in-place and takes just over 1.5 minutes.

Can't think of any more coffee stuff to build. Maybe a reflux still .....

raj

[sweaner]

Yeh, I made all similar stuff....Ok, just the WDT tool.....OK, it is just a bent paperclip.

Seriously cool stuff. I wish I had the ability.

[r-gordon-7]

And I'd probably cut myself while trying to bend the paperclip! I can't begin to imagine how Raj could do all this - So very impressive!

r-gordon-7

[matadero210]

Thanks, guys. You'll notice that my "chaff collector" is just an asparagus steamer basket--I only build when I can't buy what I want.

[espressme]

Raj,
You seem to take a whole lot less parts than Rube Goldberg to perform a useful function!
Very, very, well done!
sincerely
Richard