Another off idea...

[mbele]

I'd like to start a discussion about my next diy project, please comment.

This would be a tool for a single/specific job, with following requirements (based on workpiece size and tools used):
travel length: 1100x100x50mm
precision (positioning accuracy,straightness): 0.05mm/300mm C7 ok
positioning repeatability: at least 0.05mm (minimum backlash)
working speed: 8mm/min
acceleration: 3m/s^2 or better
max cutting forces: 180N (8mm 2fl carbide TiCN DOC=14mm WOC=2mm HSM toolpath S19630 F4466 P=1.38kW CF=17.3kg says FSWizard)

Since one axis is extremely long compared to the other two, "traveling colum" seems to be best configuration.

I made an initial sketch to get some idea about size and mases:

Z axis~16kg
Z+Column~34kg
Z+Column+Y~50kg

Does someone have experience with chinese DFU ballnuts (eg from aliexpress/BST automation)? I was thinking 1605 for YZ and 2010 for X. Would they be stiff enough?

I'd like to use double ballnuts on ballscrews to eliminate as much backlash as possible (quality BK/BF supports too).
I have older chinese single nut rolled 2010 "C5" ballscrew with correct length but it has some backlash under load, so I thought that double nut variant would be better, sacificing somewhat the ease of motion.
Another option is to use used/nib/nnb brand name preloaded/double-nut ballscrews, but that would complicate things since I would have to work with what's available at the moment (sizes,lengths), and prices are higher even for the used stuff.

Thanks for looking.
Mihael

[mbele]

I've checked preloaded brand name ballscrews on ebay, and found few solutions for Y and Z. I have to select ballscrew for the min possible diameter (inertia) and required stiffness.
Considering 180N load and 0.05mm allowed error, can I assume that total stiffness of the machine at the tool tip should be no less than: 180N/0.05mm=3.6N/um? Would then a 12mm dia ballscew with rigidity(K)=120N/um be enough?

I've sketched a solution for X axis base. Tubular structure would be made from a 50mm thick granite slabs, glued and fastened together with screws. Steel rail supports would be fastened on top with screws and epoxy, and then shaped and ground to create plane for mounting rails. Would a rail span of ~200mm be enough (I see that span used on gantry machines)?

[driftspin]

I've checked preloaded brand name ballscrews on ebay, and found few solutions for Y and Z. I have to select ballscrew for the min possible diameter (inertia) and required stiffness.
Considering 180N load and 0.05mm allowed error, can I assume that total stiffness of the machine at the tool tip should be no less than: 180N/0.05mm=3.6N/um? Would then a 12mm dia ballscew with rigidity(K)=120N/um be enough?

I've sketched a solution for X axis base. Tubular structure would be made from a 50mm thick granite slabs, glued and fastened together with screws. Steel rail supports would be fastened on top with screws and epoxy, and then shaped and ground to create plane for mounting rails. Would a rail span of ~200mm be enough (I see that span used on gantry machines)?

Hi mbele,

This is gonna be an intresting build if your total allowed error is <0.05mm @ 180N

What servos can run at 180n at operating speed. Did you check the stepper calculator on this site?

Also please check what happens with 180n force in all directions at the tool tip on all (moving) parts combined.

One does realise quite quickly that everything looks like made of rubber.


Grtz Bert.

Verstuurd vanaf mijn SM-A320FL met Tapatalk

[mbele]

Bert, thank you for your comment.
180N/0,05mm may be wishfull thinking. I can probably reduce that requirement to half, with 8mm 1fl carbide endmill doc=20mm woc=0.5mm S24k F7238 P=0.82kW CF=8.5kg, but even if that proves to be unrealistic, it cannot get worse than my current setup which can complete the part.

I've made my own calculator (gulp) using example calculations I found in THK/NSK catalogues. It calculates inertias, drive and acceleration torque, required pulse frequency, resolution and max feed using following params:
- screw lead, diameter, length, efficiency
- sliding mass and friction
- cutting forces
- acceleration and direction (vertical or horizontal)
- gearing (number of teeth, gears inertia)
- dynamic friction torque for preloaded ballscrews and support bearings (basic torque based on BCD and preload for ballscrews)
- rotor inertia
It shows warning if max feed rate is compromised by ballscrew critical speed or there's an inertia missmatch greater than 1:3

I've just checked it against the calculator I found here:
http://www.mycncuk.com/threads/1524-...otor-do-I-need
basic calculations seem close (it has acceleration locked to max feed), and it seems to suggest motor with 3x the required torque.

With steppers I would need 10mm lead ballscrews rotating @800rpm. X axis max feed (rapid) would be compromised, and that could be handled with 20mm lead ballscrew and stronger motor.
With 3000rpm ac servos, Y and Z would need 2.5mm lead direct driven ballscrew or 5mm lead geared down up to 1:2. X would have to be 20mm dia 10mm lead ballscrew geared down to 1:2.5 to keep it spining under the critical speed.

I'd like to reduce the footprint of the machine as much as possible, so I would hide motors inside a column and use belt driven ballscrews, otherwise, direct driven ballscrews would be better solution in the long run since I wouldn't have to worry about belt wear (I'm using belts on my current setup).

If pulleys are used, I would like to use shaft-to-pulley clamping devices (sleeve coupler?), but they would increase the required number of teeth (diameter) of motor side pulley raising the motor side inertia significantly.

X axis 2010-1300mm ballscrew could be driven by 400W AC servo (1.3Nm, 360gcm^2, 3000rpm, 2x2500p/r) geared 24:60 providing 12m/min feed rate @ 250kHz, 0.8um per step resolution.

Inertia wise, for Y and Z, 16mm dia ballscrews would be ideal, but with gearing it seems that 20mm would work and it would be stiffer. I'm considering new ballscrews I found on ebay (could be cut and remachined, but calc for actual length)

Y 2005-284mm balscrew driven by 200W AC servo (0.67Nm, 134gcm^2, 3000rpm,16bit enc) geared down 20:36 (motor side sleeve coupler cannot be used since pulley dia is too small) providing 8m/min @ 330kHz, 0.042um/step resolution

Z axis ballscrew would not have BF support, reducing it's length to 160mm (w/ double nut). It could be direct driven with 200W AC servo, but if I want to hide the motor inside the column (and thus introduce pulleys), I have inertia problem. Again, if not using sleeve coupler for the motor side pulley, I could go up to 20:36, but with coupler there is no solution for 135gcm^2 rotor inertia (14mm motor shaft requires coupler with 26mm outer diameter, and I guess that minimum pulley dia is at least 10mm larger - 24T pulley, and that pulley introduces too much inertia on the motor side). So, solution is either with 20T motor pulley with set screws or direct drive and motor sticking out of column.

With 16mm dia (or less) ballscrews, total inertia drops, shaft end has smaller diameter so smaller sleeve coupler could be used (not very significant since it's on ballscrew side). It turns out that Y and Z axes could be driven with geared down 100W motors (smaller shaft dia - smaller sleeve coupler, less teeth on pulley), making 12mm ballscrews ideal in that respect. And there's my question from previous post :)

[mbele]

I've figured out motors and ballscrews...

Can someone comment on base structure and x/y configuration? I have option to make base out of 30mm or 50mm thick granite slabs. I used gantry stiffness calculator I found here to get rough numbers, and it seems that 30mm would work... I don't know how to upload modified excel file, so here are the params I used:
Length=1500, material 5.00E+04 / 2.7E-06 / 24000
Z load=500 simply supported
X load distance=150mm, 200N
Box section, width=250 height=230 thickness=30
(Young's mod for granite is 10-70GPa, I used 50)
It showed 4um deflection due to 50kg load, and 0.25um tool deflection in torsion. Actual base length will be shorter, depending on the x axis block span, so I expect better results for deflection.
Here are the sketches of granite slabs configuration:

I'm wondering if there's a gain in moving X ballscrew from midpoint between the rails towards the tool/load side, and if the overhang of Y axis bed would be a problem.
Thanks
Mihael

[mbele]

I think I figured YZ axis arrangement. Options:

and

Column would be made of aluminium, y bed 30mm thick, sides and Z bed 20mm, Z ballscrew support and supporting triangles 15mm. Rals for X would be Hiwin QHH20 or LGW20, for other two axis I can use SHS15R or LGW20 (that's what I have available).
Base structure looks flimsy compared to the rest of the machine, so I'm thinking of using 50mm thick granite slabs...

[mbele]

I've compared this idea with
https://www.youtube.com/watch?v=ewU8zWnuE3s
(I think Piotr's machine seems rigid enough to perform the job I had in mind)

The bigest revelation I got was that torison in the granite base should not be an issue in this build. In the moving gantry configuration, whole length of the gantry is subject to torsional forces since gantry is anchored to the base at it's ends with blocks on rails. In traveling column configuration, torsional forces act on shorter length - depending on how the bed (which stock is mounted to) is mounted to the base structure. So instead of building a ractangular closed section, ribbed construction would work better. Distance between ribs should correspond to X axis block span.
Deflection due to the weight of the base and the column can be somewhat controled by the distance between the mount points of the whole structure (eg. lengthwise 1/4 free-supported, 1/2 supported-supported, 1/4 supported-free, actually 310mm+780mm+310mm seems like the most balanced ratio)

In Dan Gelbart's lathe video
https://www.youtube.com/watch?v=sFrVdoOhu1Q
he connected two granite beams using precision spacers he made from steel tube. Beams are pressed against these spacers with bolts.

In his other video about large structures
https://www.youtube.com/watch?v=EeEhS3zmnDg
he uses tube and epoxy to connect two plates together.
I'd like to go this route using steel tubes as ribs, but I'm affraid that thermal expansion of steel would crack the granite.
This page
https://www.engineeringtoolbox.com/t...ge-d_1612.html
shows radial thermal expansion of 4"(~100mm) diameter thin walled stainless steel tube for 90degF (50degC) change to be 0.003" (0.0762mm). How would I check if granite would crack under this stress?

[AndyGuid]

Unfortunately I don't know enough to assist you Mihael, but it certainly looks like you're putting a great deal of thought into designing your machine!

Good Luck with it.

[mbele]

Thanks AndyGuid. I think I figured a solution with granite ribs. Currently I'm trying to figure out the best way for implementing chip guards...

[mbele]

Hah, back to the drawing board...

1. Z axis travel is too short, I use following calculation:
(max tool stickout - min tool stickout) + max stock height + clamping height + safety
minimum: (35-10) + 25 + 10 + 5 = 65mm

2. To implement apron type chip guard, I need to extend Z axis overhang, and since arrangement was optimised to make tooltip as close as possible to the linear components, arrangement does not make sense any more - Y overhang needs to be extended for 60mm and axis travel is only 100mm

3. rigidity of X axis ballscrew shaft is 7x less than Z axis ballscrew, so there's no point in beefing up Y and Z if the percentage of their impact on the whole system rigidity is small. X axis rigidity can be improved by increasing shaft dia, but that raises inertia, so 2.7Nm/840gcm^2 stepper would work better than 400W/360gcm^2 AC servo, but that puts me in the no-feedback scenario, so there's no point in using servos on other axes...