[el_fuego333]

Hello,
so after X years I finished my first milling machine.
The goal was to build a small milling machine about the size of a Tormach Pcnc440 or similar, on ball bearings and as accurate as possible. So that I can move all the parts myself.
I don't have any workshop equipment (except a little old horizontal router that I just drilled holes on).
The base the column with the coupling of the column and the base consists of thick-walled steel 200x200x12. (All units in metric millimeters) The base and column are filled with concrete.
I decided to use ready-made new linear actuators from THK - KR46 instead of the classic solution of rails and screws due to the simplicity of construction and assembly
in precision class P, which are preloaded in the line and the ballscrews have a maximum backlash of 0.003mm. THK declares high rigidity, but this has turned out to be a weak point, but too late.
Departures:
X - 380mm (only 300 are usable due to the planned fourth axis)
Y - 145mm
From - 310mm

Spindle: 1.5kW Chinese driver 12000 rpm 2.3Nm ER20
The headstock is made of one piece.

Engines and drivers:
Vexta PK566 5-phase stepping motors with gearboxes 4Nm, 6Nm and for Z 8Nm. - the engines had to be stronger due to the 20mm pitch of the ballscrews.
UDK5114N drivers

Table - steel 540x150x40mm, grounded

Total weight: a little over 300kg.

In total, I had only 7 parts made. All mounting surfaces are ground so that I don't have a problem with assembly with precision.
Nevertheless, after setting X and Y, I measured with a watch the parallelism of the table surface with Y displacement, and the table flew down the hill 0.05mm...
So I covered the couplings between X and Y with aluminum foil and it runs at a hundred of mm. Fortunately, the table was straight in the X direction, so the whole machine runs without drag up to 0.012mm.
Another problem was the imprecise manufacture of the headstock - the parallelism of the bearing surface to the Z axis and the axis of the opening for the spindle.
It also had to be lined with aluminum foil when adjusting the verticality of the spindle.

I tried to make the first shots in structural steel.
I thought that the weakest link would be the spindle, but that is not the case.
The limit for a milling cutter is a slot of 12 mm 2flute carbide milling cutter in a depth of 1 mm, with 9000 revolutions and a feed of 140 mm/min, when the surface is still nice and the cut is precise.
The milled surface with several rows of such cuts was very nice.
The 6mm carbide cutter goes in a depth of 2 mm at 9000 revolutions with 140 mm/min feed, and 12000 with 200 mm/min feed completely smoothly and beautifully.
The machine is intended for aluminum and non-ferrous metals, approx. hard wood. Steel only exceptionally, so I do not lose heart because of this limit.
With a milling cutter of 12 mm and a depth of cut of 2 mm, it vibrated and the cut was already imperfect. The weak point, in my opinion, is precisely the actuators, which do compensate
or the used 15mm pre-tensioned rails, but I think that THK is lying a bit about their stiffness and max. Retraction that is, they do not indicate the degree of their deformation for different types of retraction, only the maximum permissible values,
which the milling cutter does not even come close to reaching.
I don't have aluminum yet, but when I try, I'll let you know how it went.
I think that no one here has used THK KR actuators with this construction method. If someone decides, they are available on ebay, they are terribly expensive new, and only used ones in the highest
accuracy class P, the others are not good. The p-grade They are exactly suitable only for the construction of such small machines. If I had to decide again, I would go to the rails and screws.
And it has one more flaw: after switching off, the spindle slowly self-falls down. should have a counterweight.

[joeavaerage]

Hi,
I made a mini-mill about ten years ago and used it extensively up until about 18 months ago when I upgraded to a new own-design mill being mach bigger and more
powerful.

It had cast iron axis beds that I milled out of elevator axis weights I got from the scrap yard. It had 15mm BoschRexroth rails and cars, 20mm diameter 5mm pitch C5 Koruda
ballscrews and PK566 Vexta steppers matched with 10:1 low lash planetary gearboxes and RDK514 drivers. The travels were 180mm x 180mm x 180mm.

Like you I wanted mainly soft metal and plastics capability but be able to do steel. I had only an 800W 24000rpm spindle and so could not cut steel as well as you can
because the spindle would stall. What I did then was make a second spindle from a Allen Bradley 1.8kW 6.1Nm (cont) 35000rpm servo. Then I could cut steel up to the limit
of the rigidity of my machine. Despite being all steel and cast iron I ended up with similar performance to you, 12mm tool, a 2mm DOC and the cut was starting to degrade.

I did not blame the linear rails and cars, nor the steppers or ballscrews, but rather it was just the limit of rigidity of the machine.

I used this machine for quite some years, and it did very well, although it was limited in its steel capacity, but if used within its limits it was fine. I did decide that I wanted better
and so I designed and built a new machine.

It has three cast iron axis beds, I had them cast for me, 700mm long and 250mm wide, weighing about 115kg each. It has 32mm diameter 5mm pitch C5 double nut THK ballscrews, 20mm HD THK
linear rails and cars, a fabricated steel frame of 32mm med tensile steel, 750W Delta B2 servos. The travels are 350mm x 350mm x 350mm. I recently added a fourth axis and am in the process
of building a fifth axis trunnion table.

I also have another servo that I'm building into a spindle, its 3kW, 4000rpm, 12Nm (cont) 48Nm (overload). It'll will be an auto tool change when I build it.

And it has one more flaw: after switching off, the spindle slowly self-falls down. should have a counterweight.

My machine is not inclined to droop when unpowered, the Z axis is heavy but not really enough to back drive the 5mm ballscrews, but in any event I bought a 750W Delta B2 servo that has an
electromagnetic brake built in. It cost about $120 extra, but it will be handy when I put my much bigger tool change spindle on it.

This machine does pretty well, but if I really push it with the 1.8kW spindle I can cause the machine to flex with consequent degradation of the cut. So despite the much bigger and heavier
construction there is a limit, even if its much higher than my first mini-mill, it still has a limit that I have to work within. My experience is that just good linear rails and cars and good ballscrews
are not likely to be the culprits when determining the flexure of a machine, it is almost certainly the frame no matter how big a stiff you make it!

Craig

[el_fuego333]

Hello Craig,
Thanks for Your reply.
I would like to ask You about Your 4th axis. Is it custom made or did You buy it somewhere?
How satisfied are You with it? Magnetic brake for the Z axis is good idea.
Thank You.

Andrej

[joeavaerage]

Hi,
I made the fourth axis. It is based on an Atlanta Drives low lash (<2 arc min) servo reducer. It has a 19.5:1 reduction.

The deep reduction is good in the sense that you get very good torque multiplication from your servo or stepper which gives you very
good torque authority at the chuck to resist cutting forces. The downside is that the rotational speed it quite low. Many of my fourth axis toolpaths
are for small diameter parts and require a significant number of rotations to complete, which results in long cycle times. With larger diameter parts
that is not a problem. It is for this reason that I have fitted a 750W Delta servo, as it is rated to 3000rpm, and therefore gives me a max output
rotational speed of 150rpm. Initially I had used a stepper but it was just too slow, ie about 10rpm and so I uprated to the servo.

150rpm is quite useable but I don't want to go any slower.

You may have noticed in another of the pics I posted the parts I have for a fifth axis. There is another of the Atlanta drives servo reducers,
but a slightly different model, it has very low lash <1 arc min and a 6.75:1 reduction. I have another 750W Delta servo for it. The resultant max output speed will
be about 440rpm. This should help with those rotary toolpaths on small diameter parts.

The slightly larger drive, currently my fourth axis will have the chuck removed (it bolts on) and a trunnion table bolted in its place. The extra reduction and consequent
torque authority over the trunnion will be very suitable, and as the trunnion will rotate only +90 to -90 degrees , the modest speed will be more than acceptable.

The fifth axis Atlanta Drive servo reducer I bought on Ebay just six weeks ago for $180USD. The claim is that its new old stock. Now that I actually have it I would confirm that it does
not look to have been used or even fitted. In short I am very impressed with the quality and the price. By the time I paid GST (NZ based tax) and freight it worked out to $600NZD, or about
$400USD at my door, but still I think I got a bargain.

The fourth axis drive I bought seven or eight years ago off Ebay, new old stock, for $130USD, another bargain. It was my intention to use it on my mini-mill but when it turned up it
was just too outsized for my little machine and so its gathered dust ever since....until six months ago when I rolled up my sleeves and made the fourth axis. Its a perfect match for my
new machine.

All-in-all I have tried to make the best job I could afford with regards accuracy and rigidity and have a very useful fourth axis as a result. Naturally I'm hoping for similar success with
my fifth axis. My total expenditure on the two rotational axes to date is approx $5000NZD or about $3150USD. I still have some way to go with the fifth axis so I suspect the total
for the complete fourth axis trunnion and fifth axis will be $6000NZD or $3800USD. I have added capability but it does not come cheap......and the story gets worse.

Genuine four and five axis CAD/CAM software is expensive....very expensive no matter from what software manufacturer. I already use Fusion ($550USD/year) so I elected to go with
the Machining Extensions, and the introductory offer was $1200USD/year. The subscription is coming due in a few months and this time I'll have to pay full price which is $1600USD/year.
You might think that is expensive...and it is....but is very cost competitive with other software suppliers. But you absolutely need the CAM software. You can get away with
hand written code for simple four axis rotary and indexing paths....but anything more and you need genuine simultaneous four axis CAM. Five axis indexing might be
possible by hand but any simultaneous five axis REQUIRES genuine, read expensive, CAM.

The bottom line is that the expense (considerable on a hobby budget) for the trunnion and five axis is going to be overhauled by the cost of the CAM subscription over
three or so years.

I am overall delighted with how this project has come along....but if I'm being honest if I had accurately costed it out to start with I might not have proceeded.
Yet another learning experience!

Craig