1st group body prototype

The prototype of the group body is done. Although there are a few minor surface blemishes, the overall finish looks to be on par or better than the two vintage groups that I have. Certainly, the lines are a little crisper. A closer inspection of the photos indicates that there may be an issue or two to address with the secondary machining processes. I've requested a few more photos and some QC measurements. The part may have to go back for a little more machining before it is shipped.

Coffee project goals

The goal of this project is to build an 'update' to the aurora, which is my favourite vintage single group machine.

There are several variants of the aurora, at least four that I know of, which were imported to in North America by various resellers. I am lucky enough to have two of them. The one in the picture in my first post was made in Milan and imported to Canada. The one below, with a different boiler and HX, was restored by Orphan Espresso. I believe that it was assembled and sold in the US by Termozona, a company from Connecticut, who added the "Europa" branding. I'm not sure how much of the case of the Europa is original, but the design and build quality on the European import are better.




Dividing the machine into its component assemblies, the plan at the moment is as follows:

Group - everything that I have read about this group is positive. The main casting contains over 7lbs of brass and consequently has plenty of thermal mass to maintain temperature stability. It makes a killer shot and I prefer its aesthetic qualities over other lever designs I have seen. Consequently, this, apart from a change in a few of the materials for reasons that I will go into later, will remain faithful to the original design. As this part of the project is the most difficult, this is where I am starting.

Boiler - after the group, the boiler is, imho, the most important part of the design as it determines how well the whole machine performs. Here again, I want to stay true to the original design while benefitting as much as possible from the performance improvements that are offered by contemporary control electronics. So, much as I like the project that EspressoForge is leading that pairs an Ascasa Thermoblock with a lever group, I am going with an old school boiler and heating element, with an HX.

Control - this is the area where there is the most room for improvement over the original design which, depending on whether it is an autofill variant or not, uses a Gicar and Sirai pressurestat. This last is a rather blunt instrument with a deadband, when it is new, of around 0.2bar (considerably wider when the diaphragm of the stat gets old). Although I may start with a pressurestat as it will work "off-the-shelf", ultimately the control will be handled with some flavour of micro-processor which will drive the element using an SSR. A starting point for this may well be EspressoForge's combination of the arduino + TC4.

Autofill - yes. Fairly straightforward with a solenoid valve.

Sight gauge - not currently planned. The sight gauge adds a lot of complexity to the plumbing and the case design which would be eliminated by providing feedback from the water-level sensor with an LED.

Case and chassis - with the goal of making a commercial quality machine that can live on the counter top in the kitchen, I am going to try to slim down the profile and footprint a little while remaining faithful to the styling. The drip tray and sump designs also have room for improvement.

Cold Aurora in February

One particularly cold February morning, I got up to discover that the auto-fill circuit on my Brugnetti Aurora had failed. As I opened up the Gicar controller a few days-without-proper-morning-coffee later to figure out what was wrong (an old and faded capacitor), I realized that I had now either fixed or maintained every system on the machine except the lever group. So I took that apart too and replaced the spring and the gaskets. As I was putting it back together I started wondering about what it would take, given that the original Brugnetti company no longer exists, to make more of these fantastic machines.

A few days ago, after nearly three months of research, drawing, discussion and negotiation, I received these photos:



... a wax positive of the group body which, over the next few days, is going to be encased in plaster and replaced with molten brass.

The photos are also a sign, at least to me, that this project is now well and truly underway and that it is time to share it with the generous people on this forum, whose advice I have greatly benefited from since my obsession with coffee began. I am a long, long way from a working machine, but you have to start somewhere.

CNC lathe conversion part 2 - X axis

So I decided to undertake a non-destructive conversion of my King 10x36 Gear head. Not replacing the acme screws with lead-screws means that the mechanical conversion is pretty easy. Backlash and precision are not going to be great, but they may be good enough for the application. Time will tell.



Here is the saddle with the compound and the cross slide removed.
The lead screw is 5/8 x 10tpi acme. This has always been the weak point of this machine. It seems impossible to torque the acme nut to the compound and have the acme nut correctly adjusted to eliminate unreasonable backlash. Fixing that would likely require replacing the bearing, mount and probably also the nut and screw. More trouble that I think it is worth going to at this stage.





A shot of the existing screw and fixed bearing assembly. The double nuts at the end of the journal bearing are a very poor solution as counter-tightening the second nut tends to dramatically alter the load on the bearing.





I'll be using 5mm HTD timing belts and pulleys from SDP/SI (who have a very useful pulley designer on their site) rather than direct drive for this build. The X axis is the tricky one as the pulley plus the belt and the cover have to fit under the cross-slide. I was trying for around 2:1 advantage, but couldn't quite get there on the X. The configuration for the X is 13 double flange : 22 no flange with a 60 tooth belt. The Z is 19 df : 38 nf with a 77 tooth belt. Three of the four hubs all had to be re-bored to the correct shaft size. The 13 tooth was a total pita. 

A 6A25M022NF1508
A 6R25M060150
A 6A25M013DF1506
A 6A25M038NF1510
A 6R25M077150

A 6A25M019DF1508






The disassembled fixed bearing and, at the bottom of the shot, the journal extension with a 0.3745" stub, a section of M10 x 1.5 thread and a 0.5000" shaft to receive the drive pulley.





Here, the existing threads have been cut off the end of the lead screw journal and I'm just about to centre drill, drill out and ream a 0.3750" hole that will receive stub of the new journal extension.





A little loctite and a coffee break later and the new journal is done. 




All finished except for a cover (which turned out to be a little hard than I thought and after a couple of false starts isn't done yet). Note the brass nut replacing the double M10 nuts. This can be tightened to put just the right load on the bearing and then locked in place with a set screw. I dropped a little brass slug into the set screw hole before torquing it down to avoid damaging the screws on the journal.






The stepper (an Oriental Motor PK296DAA I think) and the mounts just fit inside the splash guard. A couple of final thoughts... There is an alternative to this mounting position as the King has an excellent gear box on the saddle to power either the Z or the X axis. Mounting the motor on the drive shaft of the gear box instead of directly to the end of the X lead screw would be very easy. While taking the X axis fixed bearing apart and reassembling it a few times I realized that, in addition to it being the least well-built part of this machine, it also isn't really intended to handle radial loads. While the motor and timing belt aren't particularly large loads, they still shouldn't really be on this bearing. If this proves problematic, I can either attempt to replace the whole bearing assembly with something better off-the-shelf, make a housing for the existing bearing that will handle the load or move the whole shebang to the gear box drive shaft. I decided not to go this route at the start because the gear box will induce considerably more backlash.

CNC lathe conversion part 1 - research

I was not really planning on going here at the beginning of the coffee project, but the prices I'm getting back for some of the parts are making me rethink at least the initial production run. There are some things that I cannot do in the shop: castings, complex cnc parts and plating etc. Cutting tapered parts is also not possible (or at least not practical) and the time I would spend on the learning curve for manual single point metric threading on an inch lathe seems like it might be better spent on tackling the CNC. However, my lathe is manual and possibly not the best starting point for a cnc conversion. I'm also not terribly keen on no longer having any manual control. So, some things to consider. 

(Note - here is a technique for manual metric threading using the half nut to disengage the lead screw.)   https://m.youtube.com/watch?v=HXt4TWa382Q  


Useful lathe CNC conversion posts:

8x12 Harbor Freight conversion
http://plsntcov.8m.com/CNClathe/CNClathe1.html

Excellent conversion of a Jet 13x40 http://www.hobby-machinist.com/threads/converting-a-13x40-manual-lathe-to-cnc-with-servos-and-mach3.33405/

Another excellent conversion using a kit from Billy Tools
longezproject.blogspot.ca/2015/02/cnc-lathe-conversion-part-1.html Possible converts:

Busy Bee 7x12

At 75 lbs there isn't really enough meat on the bones and bore is only 20mm.

Old and tired gearhead on kijiji

More old and tired 36" Logan on lespac

Converting a manual lathe to CNC is quite a bit more invasive than doing a manual mill. All of the elaborate gearing for threading and power feed are dumped along with the compound and the worm screws. All that one really needs is a solid frame, a good apron and the head and tail stock. Finding one with a 3 phase 220V motor would be a bonus, but unlikely. 

...so its back to ballscrews...

For my mill conversion I bought used ground ball screws for the X & Y and a new surplus for the Z from ebay. The precision of all three was C5 if not better. However, the X axis (which is the longest, most expensive and was the hardest to find) has some issues, possible because of a crash courtesy of the previous owner. So I am somewhat ambivalent about going the used route again. However, finding surplus screws of the right dimensions with support blocks is really tough, so the only affordable and convenient options for buying new are Chinese rolled C7 precision grade that come with journals and matching supports. C7 is 0.002" in 12" (in the worst case over the length of the screw) which, given what we are proposing to manufacture here, isn't that bad.

X axis

used THK - 14x4mm 358mm with 235mm travel - with supports and stepper frame - $189

used NSK - 16x5mm 407mm with 282mm travel - with supports and stepper frame - $189

used NSK 20x5 460mm overall with supports - $250

new NSK 25x5 ~440mm overall no supports - $270

surplus THK 14x2mm 416mm overall no supports $200

new Kuroda 12x2?mm ~15.5" overall no supports $90 plus shipping from US



Z Axis - the hard one...

used THK - 20x5mm x 1385mm ROLLED with supports - $239

new C7 Chinese - 20x05 x 1352mm with supports- linearmotionbearings2008 - $105 plus shipping



Update 2016 4 27

A couple of simple conversions using the existing lead screws. The second one is particularly interesting because it doesn't require giving up manual control! 

flashcut cnc
https://www.youtube.com/watch?v=_Polq5piWhQ

optimum
https://www.youtube.com/watch?v=FWg8NzfP108

A conversion of a small grizzly using a kit from BD Tools. Excellent blog about building an airplane too.

http://longezproject.blogspot.ca/2015/02/cnc-lathe-conversion-part-1.html

Joining dissimilar metals - Soldering on with the boiler (wah wah wah)

It would seem that the boiler will be composed of a few metals.

There are a number of possibilities. The two Auroras that I have taken apart have two different copper boilers made in different ways. The U.S. model has a hemmed and (presumably) soldered joint along the cylinder - the two ends of a sheet of copper are folded over and hooked into each other before being soldered together and flattened slightly with a hammer. This one has one semi-spherical end and flange at the other (more about this later) which are both (again presumably) soldered in place afterwards. The European model has a much neater looking seam along the length of the cylinder which looks suspiciously like a butt joint. The two ends are bossed and fit around the exterior diameter of the cylinder. All the joints look, to my inexpert eye, to be too perfect to have been made by hand. The neatness and the likely presence of a butt joint both suggest that this exotic beast may be machine brazed. Funnily enough, I don't have a custom-made brazing machine, so replicating this is looking unlikely.

Aurora 'euro' model boiler - note the perfect seams.

I have fooled around a little with some scrap copper, experimenting with the techniques to see just how hard making the boiler out of sheet copper stock would be. 

      

   


Top left to bottom right: Silvabrite silver solder from the plumbing supply store, slit and annealed copper pipe, flattened and hemmed sheets (MAPP gas in the background), joint before soldering, soldered joint before clean up, detail of the joint cut through on the band-saw (note the small void, which may or may not be a problem). 

So back to that flange on the U.S. model. The tricky part about this is that it is made of brass and a stainless steel ring ...and the flange bolts are steel. 

I have no solid theory as to why the stainless ring is there. The only plausible explanation I can come up with (other than strength) is thermal expansion: perhaps the stainless band acts to keep the expansion of the brass flange in check. Although copper and brass are soldered all the time, perhaps the size of the parts concerned makes the joint prone to failure from the thermal cycle. Certainly, the numbers for the materials lend this hypothesis some credibility.

Metal                          Thermal expansion (micro inch/(in deg F))

Copper                                  9.8
Brass C23000                         10.4
Brass C28000                         11.6
Stainless 303, 304                   9.6

Stainless 316                          8.8

But, taking it as a given that the engineers who designed the boiler knew what they were doing, ours not to reason why...

Silver brazing these parts together is possible, but exposing the stainless to high heat is potentially a problem:


While brazing can provide a stronger joint, the high brazing temperatures can be bad for stainless steel. At those temperatures, carbon in the stainless steel can form chromium carbides which takes the chromium out of solution, making the steel non-stainless near the joint. This area is prone to rust and cracking after it is in service. The problem cannot be fixed by re-passivation so it is best to avoid excessively heating the parts during the braze and keep the total time at temperature to four minutes or less.

So that means brazing the steel bolts to the brass flange, and then either shrink-fitting the stainless ring to the brass or soldering the flange to the ring and the copper boiler skin in one shot - without destroying the hemmed copper seam. Soldering the stainless to the copper (and presumably also to the brass) can be done using the silver solder and a special flux: Harris Stay-Clean® Liquid Flux - which can apparently be bought at Linde and other Canadian welding supply places, otherwise on ebay in the States. I have come across some mentions of using muriatic (i.e. hydrochloric) acid or plain old plumbing flux, but the Harris stuff is supposed to make it easier.

All this makes me wonder whether an all stainless construction (with the exception of the boiler to grouphead flange which will necessarily be brass) might not be simpler. Replacing the flange and the end cap with laser or water-jet cut stainless parts and swapping in welded stainless schedule 10 tubing for the copper might be more expensive in materials but considerably less labor. Additionally, this would eliminate the cost of the molds for the brass parts.

The only complication (other than potentially having to drill and tap stainless) is the connection between the boiler and the boiler to grouphead flange. This last part serves as a reservoir for the grouphead and is heated by conduction through the wall of the boiler. Copper has about twenty times the thermal conductivity of stainless so a 1/8" thick stainless instead of a 1/16" copper boiler wall is going to have a significantly effect on the heat transfer to the reservoir. 

For the shrink that fits - Alice's adventures in tolerance land

More thoughts about the group body casting.

Having stared at the casting after looking directly at sun for a few minutes I am now pretty sure that it is an assembly of two parts: the bulk plus a small internal sleeve that holds the dispersion screen. Putting these two parts together could be done in a number of ways, but to avoid the possible consequences of heating the casting to soldering or brazing temperatures, I think that a shrink fit is probably the best bet.

From the engineeringtoolbox:

Shrink-fits are assembled by heating them to temperatures where the expansion exceeds the interference. Required temperature heating can be calculated as

dt = δ / α d_i         (1)

where

dt = temperature heating (^oC, ^oF)

δ = diametric interference (mm, in)

α = coefficient of linear expansion (m/m^oK, in/in^oF)

d_i = initial diameter of hole before expansion (mm, in)

Diametric interference can be calculated as

δ = dt α d_i         (2)

So assuming the sleeve is at room temperature and the body is heated in boiling water:

Nominal hole diameter d_{i =} 58mm
Temperature heating dt = 80^oC
Coefficient of linear expansion α for C23000 brass = 18.7x10^-6 (°C)^-1
Coefficient of linear expansion α for C87850 brass = 18.5x10-6 PER °C (20-100 C)

Therefore:

for C23000
δ (80^oC) = 80^oC x 0.0000187 (°C)^-1 x 58mm = 0.086768mm

for C87850
δ = 0.08584mm

...or more or less the same.


So over-sizing the sleeve diameter by 0.08mm (which is pretty close to 0.003") will allow it to be shrink fit into place with minimal force.

And after drinking the potion mark "drink me" Alice can now go down the rabbit hole to tolerance land. 

Interference H7/s6 S7/h6  - Medium drive fit for ordinary steel parts or shrink fits on light sections, the tightest fit usable with cast iron. 

The tolerance for the sleeve and the hole diameters is important as the two add together. So a symmetrical tolerance of +/- t will possibly result in a total tolerance of 2t.

So allowing for a minimum of 0.05 interference which should still provide a 'medium drive' shrink fit, the diameter of the hole should be 58 +0.015/-0 and the sleeve diameter should be 58.08 +0/-0.015

So in the worst case scenario, the hole is 58.015 and the sleeve is 58.065 which still results in 0.05 interference.