18974RE: Re: [SherlineCNC] Stiffer x.axis (Stainless on Sherline Mill)
- Nov 6, 2013
Very cool! Sounds like an amazing piece of machinery built with the most precision possible. Thanks for the information, glad to know that they're are other lathe junkies in the world. Ha!
I will keep an eye out for the 3d renderings.
---In SherlineCNC@yahoogroups.com, <gcode.fi@...> wrote:Making an x-axis rigid on lathe - high end.
The job took about 200 hours and 1000€.
Took out acme screw.
Replaced with much thicker ballscrew.
The acme was (iirc) 14 mm thick.
The ballscrew is 18 mm (3/4 inches, actually).
This is about 3x more rigid.
Rigidity of screws is on all manufacturers tables - how much force required to move one Newton.
Also about the same ratio as max push force (rating) on the screw.
The normal screws are supported at front end only.
This means that the screw has a floppiness of length cubed.
If length is 2, thats 2x2x2 = 8.
If length is 1, that 1x1x1 = 1 or 8 times stiffer.
I support the screw at both ends.
This makes the max free length 1/2 of original design.
I built an outdrive station, with secondary sets of bearings, and bolts that attach to the saddle.
The bolts are 18 mm thick, two of, tool steel, and end in support plate of 125 mm diameter, 30 mm thick (it about 8 kg in mass).
Now, I put the screw in tension. At the back of the screw there are two sets of opposing threads, one is used to tension the screw, and one is used to lock the screw to the current tension.
This makes it about 2x as stiff as a standard screw (according to literature and specs from screw manufacturers).
I then built a cover for the whole thing, from an old PC box.
Finishing and painting took about half the time (100 hours) with 3 coats pant, sanding, 2 coats laquer (wont stain so fast).
All steel parts were blackened with a kit, it a controlled rusting process.
The end result it theoretically 3x8x2 = 48 times stiffer.
Due to inaccuracies, and lack of perfect industrial processes, I call it 14x.
The feel is about 20x more light. I can (but dont) use the gibs at VERY much stiffer setting with no effect whatsoever.
The ballscrew is about 90% efficient - the (very good, industrial) acme screw was 30%.
This also made the screw 3x more sensitive.
The actual screw has a real world resolution much better than 1 micron.
I am using 3Nm steppers at 1:3, for about 400 kg push on the screw, and 6000 (micro)steps / (5 mm, really 1/5 inch) turn for 0.8 micron step size.
(ONLY) When stiction is not a factor, AND the force is very great in relation to the stiction force, microsteps to 10 are accurate.
This is the case on my x-axis slide.
Using a 1 micron digital indicator, I get reliable 1 micron movements from the slide at all times.
Repetability is likely about 5 microns due to inaccuracies in manufacture.
The job was about 7.5/10 difficult.
The results exceeded my expectations.
The end result is 10/10 - I wanted a 1 micron resolution lathe and got one.
I needed to make the outside carriage plate, the 2 carriage bolts (250 mm long, 30 mm on thick end and 18 mm on long thinner end), 4 thrust bearings, and the ballscrew.
The ballscrew needed 4 threads, 2 left hand ones.
Turning a ballscrew is not hard with carbide inserts.
I did this about 8 years ago when I was still learning.
I am now redoing this with AC brushless servo motors, together with live tooling.
At the time I am waiting for my second turret components to arrive, and will fit the second turret on the lathe.
I am also changing the z axis to a grade 1 ground 32 mm ballscrew from an industrial turning center. Its good for a 40 hp machine, grin.
I am also changing the lathe motor a an indexed servo spindle - and will likely use a precision low backlash (2 minutes of arc) planetary gearbox for the spindle.
About 1:5 gearing, I think. This would allow turning at upto 600 rpm with the planetary, with no need for other transmission.
At 5000 counts on encoder this would give 25.000 counts, and at 10.000 counts 50.000 counts/turn (just right).
(I did this on a 12x industrial, a chester craftsman, but that wont matter).
Capabilities will fully match an industrial live-tooling lathe, soon.
I am also building 2 spindles for live tooling.
Threading is not yet working - I am waiting to connect a new hardware motion controller, with industrial capacities.
Costs are about 1/10 of industrial stuff, and I am building for production purposes/quality.
Ie you can look up a Haas ST10 with C axis for a cost of comparable one.
Ths big lathe is for using.
I will also make a small 7x as a demonstrator, and expect to take it to several shows next spring.
I also expect it will be on all sorts of web sites.
I will make barfeeder capacities part of it.
At the moment I have just received my 2 lathe collet lever chucks -
a 3AT for an old lathe, and a 5C, both working secondhand versions from the US.
When I see how they work on a dummy lathe, I will make an auto chuck for both lathes as well.
I am building for industrial purposes with industrial quality.
Thus I am willing to spend a reasonable amount of money as I know it will pay off.
For a hobby use, when starting, I used a DC treadmill motor I converted to servo, for example.
I have 3D renderings and pics of the x axis, and might put these up soon.
Some work-work is interfering for a few days, unfortunately.
"(with modified x-axis to be very rigid)"I am interested in how you have modified the X axis to make it more rigid. I have been thinking about and wanting to do that. Right now I use a M30.1 Flashcut or what I believe would be M47 in Mach3 on my finish passes until no more material is being removed to try and get as much accuracy as I can.
- << Previous post in topic Next post in topic >>