Welcome to Our Community

Some features disabled for guests. Register Today.

5 AXIS offset compensation

Discussion in 'CNC Mills/Routers' started by HeartOfGermany, Nov 20, 2018.

  1. HeartOfGermany

    Builder

    Joined:
    Oct 29, 2018
    Messages:
    5
    Likes Received:
    0
    I know how to measure the XYZ offset of the workpiece with the arduino with grbl on it. If I convert it to 5 axis there is a problem if I understand correctly.

    The a or b or c axis, yeah the whole toolpath should be wrong, if the workpiece is offset. If it rotates, the workpiece is shiftet, at a 90° however it might match the X axis, but the offset is on the Z axis. If it is 45° the offset is half, both at X and Z axis. I hope you know what I mean with that. So the only thing that comes to my mind in fixing that, since the grbl should not be able to compensate the gcode than, is to precicely place the part, am I right?

    If for whatever reason the workpiece is not able to be placed correct, I would need to input the offset manually in Fusion 360, right? (There is no direct communication possible between Fusion and GRBL as far I am aware of).

    My planned solution:
    - Start program (measuring probe in mill)
    - Get XYZ absolute 0
    - Drive to XY .......... (Position of the rotation axis with a spline inside and measure where this exactly is) Zero X and Y around there (with offset of the workpiece size)
    - drive to rel. X 0 and halt
    - Insert Workpiece to the Vise and push it precisely against the measuring probe till it triggers
    - check rel. X 0 again to make sure it is perfect
    - same procedure for y and z

    Yeah, I need a tought vise for that, that is completely free to move a bit in every direction (got a nice plan for that thanks god).


    So to speak: my plan would be, to precisely model in the CNC itself (also as a design for any upcoming 5 axis project). Than I add the model and workpiece to that.

    Is that route ok or do you have a better solution?
     
  2. Rob Taylor

    Rob Taylor Master
    Builder

    Joined:
    Dec 15, 2013
    Messages:
    1,470
    Likes Received:
    746
    The exact nature of the offset is gonna be different depending on if you have table/table, head/table, or head/head 5-axis, but in general, you do usually model the workholding, spindle and rotary axes in CAD/CAM in order to effectively produce useable G-code. The part is not normally assumed to be perfectly centered on all axes, and in fact this greater flexibility of workholding is one of the major advantages of 5 axis. It's also important to model the machine in 5-axis because the risk of crashing is much higher- when parts of the machine are in different places at different times, things get unpredictable. Simulation becomes a critical part of the CAM workflow.

    As long as you're probing the part in (I know bCNC can do probing, but I don't know if you can get away with just doing it in 3 axes... I'm inclined to say yes, though, because you should be able to extrapolate based on known rotary positioning), then where it is is irrelevant- it becomes part of your g54 work offset settings and you don't have to worry about it.

    If you're not probing and relying on precise positioning, it could be trickier, but with proper workholding- vice stops, soft jaws, you name it- you could make it work. I suspect your rejection rate would be higher this way, though, depending on the level of precision you're working toward.
     
    HeartOfGermany likes this.
  3. HeartOfGermany

    Builder

    Joined:
    Oct 29, 2018
    Messages:
    5
    Likes Received:
    0
    Ok that is a very good answer. I am not sure...

    Should I use BC or AC? I somehow find it better to be able to rotate the milling spindle. However my friend over in a professional environment says you are better of rotating only the table. I am not sure yet what is better.

    With AC (2x table) I would make the Table for Rotation on the right side. If I would do BC, B would rotate the milling spindle and C the table. My opinion is, that it is much easier to rotate only on axis on the table. 2 Rotations of one segment are much more complicated to archive. With BC the Table would be precisely flat over the whole area (which is huge). With AC on the other hand I would need either a huge cutout on the table or need to rise the rotating table over the fixed table to get to 90°.

    Since your answer seems very very professional (realy nice to read such a good answer after a hard late-shift), I guess you can make a suggestion. The physical build of the machine starts on January (Just if someone wants to know. :)

    The Motors by the way will be closed loop to get max precision. X, Y and Z will be 1:3. The rotation axis will be spontanously designed to the rest of the setup. Will definitely weight over a ton.
     
  4. David the swarfer

    David the swarfer OpenBuilds Team
    Staff Member Moderator Builder Resident Builder

    Joined:
    Aug 6, 2013
    Messages:
    3,239
    Likes Received:
    1,815
    assuming 3+2 operation:
    I would just set G54 for the top, G55 for left side, G56 for right, etc.
    Then Fusion just needs to be told which system to use for each operation.
    That is how CAM 'communicates' with the controller (-:
     
  5. Rob Taylor

    Rob Taylor Master
    Builder

    Joined:
    Dec 15, 2013
    Messages:
    1,470
    Likes Received:
    746
    Having now looked at the grbl-Mega-5X/cn5X system I think he's looking to use, I don't think simultaneous 5 is off the table (heh). But yeah, for indexing, just offsetting per side would be perfect.

    Honestly which you go with depends on your anticipated end usage. Table-table (ie. AC trunnion) is primarily for smaller parts at higher speed. You have to be able to fit the parts inside the trunnion bracketry. As you say, you also lose a fair bit of Z-height to the A axis mounts and the size of your workholding, plus the diameter of your part when it's vertical (A = +/-90 degrees). That's unavoidable with this setup and generally taken into account when sizing the machine and determining the z-axis table offset (that is, how far above the table the spindle nose bottoms out in its Z travel). You're generally going to get lower rigidity, unless your trunnion is heavily over-built. If you're anticipating mostly using small diameter high speed end mills, that's not an issue.

    Head-table is much better for large parts. Rotary C table is good for bulky cylindrical parts or mounting tombstones, rotary A axis allows for longer, more "human scale" parts like carved columns and legs, rifle stocks, photographic equipment, etc. depending on the exact size and nature of the machine. A rotary table that actually becomes a B axis on a horizontal machine is heavily used in high-volume, automated manufacturing because it becomes the base of a tombstone/pallet system.

    That brings me to another consideration. Precision is a stack. In any odd number of total axes, one of your motion chains is going to have one more link in it, and therefore less rigidity and lower precision- figuring out which it should be based upon the geometry of the machine is where the difficulty lies. In a vertical machine, due to the lower rigidity of the column/gantry, should you only have two axes on the head? If so, should that be X/Z or Z/B? Should you overbuild the gantry and add another axis, so it becomes X/Z/B, or Z/B/C (in the case of a head-head machine)? What's the available spacing on your table rails to maximize rigidity there? Should you do a single axis table and a trunnion? How about a single axis table and a rotary table with three axes on the head? If you make the head a horizontal, which axes should you put on the head then? Could the head basically be on an X/Y table, or should you split the linear axes? But if the table has the Z axis, you might get more rigid drilling? What's the precision and backlash of different types and sizes of rotary drive? The larger the sweep, the lower the precision toward the outside of the radius. What about backlash free geartrains like harmonic drives? Is a belt drive sufficient to hold a table without flexing under load?

    You need answers to all of these types of questions- and then the sub-questions they bring up themselves- before you can fully answer the question "what setup should my machine be?".

    If it's going to weigh over a ton, I'd go for a horizontal. X/Z head (X being vertical up in the air, in this case), Y table (toward and away from the operator) with a half-trunnion. This is a machine design that's been floating around in my head for a while now, and seems to be the best bang for buck. But again, all of this depends on what you actually want to make. There are so many machine manufacturers and product ranges for a reason.
     
    HeartOfGermany likes this.
  6. HeartOfGermany

    Builder

    Joined:
    Oct 29, 2018
    Messages:
    5
    Likes Received:
    0
    Than you will definitely call me insane. I want a stationary Table. Unfortunately there is no space to get the Table right to left or "behind" the machine, since it needs to be quite close to the wall. Z axis will be a single beam/rail as you called it. X axis will be also a single beam/rail. Y will be mounted on 2 beams/rails. To get rigidity and not need to move anything to the back of the machine (since space is limited), the rail will stick out slightly to have a higher lever length to fix the problem of Torsion (which I think would be insane). The carriage will be about 30cm long and consist off bearings rolling over an "oiled", precision planed surface/beam. Why? Well, bearings are much cheaper than buying professional carriage things. A single Rail (that will never even just hold up the forces propably and tear apart) costs as much as the spindles. :facepalm:

    The first axis that will get constructed is the X axis (longest). This alone will be used to plane the other beams. Will take ours to mill with this setup, but will at least work and give enought precision for basically no cost. The total Cost of everything (except milling tools) are planned to come below 800€ for X and Y 1400mm and Z 900mm travel distance.

    If you ask, what I mean with X is longest, whilst same travel distance as Y -> Only the beam will be longer, since I initially need a longer travel distance to build the Y axis, since the carriage also has some size and can't hover past the end of course. The drive spindles are 1500mm with a usable length of 1400mm. So to initially mill the other axis with the only precision axis I have, i will simply use a threaded rod which is almost free and will survive a single operation.

    Why all that effort? Well, we simply cannot find anyone who planes us the beams for free, so to avoid labor cost we keep it as low as humanly possible. The beams/rails are planned to be of 2 s-sections welded together to basically get a tube for much better torsion resistance. We will measure (before surface finishing of course) if this is even strong enough. I hope so, because these are freaking heavy. It is impossible to move one as 1 person. If not we need to improof on rigidity even more, but that would mean thicker spindles for much stronger, rigid drive. The Z axis basically floats in the air, so it will propably be 3 s-sections welded together. We still need to test the rigidity of the materials, so not going definite on that.

    About the rigidity with a belt drive: Well, driving the spindle it will be enought, but using a belt itself for the whole length of the axis will propably result in an insane stretch. 3D printer yes, but driving several 100kg will definitely be impossible with any precision.


    And I know, the machine will definitely flex if loaded with enough feeding pressure. So I will definitely avoid high feeds under any circumstances if precision is of importance. I guess at least 2/100mm should be possible.


    So my desicion so far is:
    - XYZ (all moving the milling spindle)
    - solid table with round cutout at max possible size for the C axis
    - B axis for the spindle. The rotation point will be placed as low as possible to increase clearance as much as possible
    - the C axis will be slightly offset of the center (in the X direction) going outside of the milling region (no problen, since it can rotate any part of the workpiece in) to allow for enough spindle clearance on huge parts. The good thing with a basically solid table is, it allows for very heavy parts without loosing much precision because of elastic deformation of the whole structure. The friend giving me the place in his shop for that project is gonna kill me - definitely. :D


    Wow, this project will kill sooooo much time, yet I can't wait to start it. It will be the most insane project of my life. (or is it?) ;)


    Thanks a lot for Answering my questions. It realy did help a lot!


    Only one last question:

    The C axis is simple - the gear will be huge and it will be zero backlash (2 offset Gears with springloading)
    The B axis is much more difficult. I need quite some reduction, that is no Problem. But I also want zero backlash. What kind of steel is a good choice to hold up to such huge loads? The spring load needs to be much higher, since a smaller diameter gear will have less leverage, so I am affraid, the teeth will eat up themself after a short time. I am not a professional. I would like to avoid hardening and tempering the gear myself. What is a "low" cost material that can do the trick and needs no hardening afterwards?

    Was a long day. Hope this long text makes any sense.

    Thanks.
     
  7. Rob Taylor

    Rob Taylor Master
    Builder

    Joined:
    Dec 15, 2013
    Messages:
    1,470
    Likes Received:
    746
    I get the idea, but I suspect once you start seriously looking at bearings with the precision and load capacity you need for machine rails in and reasonable quantity, not to mention the price of precision ground barstock(!), you might end up back at square one. Remember that THK-type rails can typically be undersized for the application; their static and dynamic load capacities are monstrous for a given dimension. Most people here could get away with 9,12, or 15mm rails on everything and see no issues. Even 20 and 25mm rails can be had for moderately reasonable prices.

    I also don't understand the idea of using a single rail. For a simple conveyor system, sure. For a machine tool, it's a non-starter.

    I'd be fascinated to see that work in practice. That's an extremely small budget for a heavy-duty machine, I wouldn't build anything like it with less than a $6000 budget. I'd put $2500 into the linear axes (rails, screws) alone. But with the will to see it through, maybe it's possible!

    How are you supposed to machine the first axis, though? If you need the other axes to be +/-0.02mm, the first axis needs to be close to +/-0.002mm!

    This level of bootstrapping would be impressive! Might be doable.

    Careful if you're welding. Tacking together with TIG might be ok, but full weldments will need a post heat treatment (more or less along the lines of holding at around 950C for several hours and then cooling very slowly) in order to remain dimensionally stable. Welding is a very stress-inducing process, which is why almost all machines are cast and/or screwed together.

    Mass is good for damping, but it's not an end in itself in terms of rigidity. You may be better off with some kind of spaceframe design which you then add mass to after the fact.

    Yeah, a machine like this needs ballscrews. They're quite inexpensive these days, but will certainly add up significantly at the lengths you're anticipating using.

    High feeds are fine if you use shallow widths of cut. Look up high speed machining (HSM). 20um (0.0008", or 8 tenths) isn't gonna happen with the rails like they are. You need high precision rails, built by a machine with an order of magnitude more precision than the one you're trying to produce.

    I think the plan, as it is right now, will result in a relatively low performance machine despite the effort and expense. In fact, I think a V-Slot machine reinforced with some steel plates and partially filled with concrete would actually perform better with higher accuracy.

    - One of your axis chains is longer than necessary
    - You have single rails, so all of your moving parts are going to have no torsional rigidity
    - Moving gantry, unless it's seriously overbuilt, is generally bad for accuracy under load. Overbuilt moving gantry means big motor and bigger driver!
    - Enormous C axis means large angle error at larger radii without massive down-gearing that results in a very slow axis, and therefore very low feedrates needing significant depths of cut to avoid tool rubbing... It's not good.
    - Do you know what you're actually trying to machine? What are the parts and materials? You have to work backwards from there. This machine, as it is, is basically only good for milling plastic foams and light clays.
    - Can't you move the table side-to-side instead of front-to-back?
    - Does the machine actually need to be this huge? Big machines are low rigidity.

    You basically want a high-performance alloy steel for the price of aluminum and the ease of use of mild steel. Not gonna happen. High-hardness, high-resistance steels (D3, H13, etc, anything they dump a bunch of chromium, vanadium and molybdenum into) require specific hardening and tempering cycles- that's how they gain their properties. Without a proper electronic heat treat oven, it'll be difficult. D3, being an air-hardening steel with good dimensional stability, will be the closest possible option for you since it shouldn't have special quench requirements. The rest of the cycle is probably hard to do though.

    Check eBay for harmonic drives. Compact, zero-backlash, high reduction. Might find a used one for "cheap" somewhere.

    It's funny, I was in this exact same place 5 years ago, where I thought I could build a 1um, 1000x750x500mm machine from scratch for under $1000. The last 4 years have been me learning about why that's not remotely physically possible. Glad I've gained all this knowledge, but it's a lot harder to design a machine now!
     
    HC.Carbo likes this.
  8. HeartOfGermany

    Builder

    Joined:
    Oct 29, 2018
    Messages:
    5
    Likes Received:
    0
    Side by side will take to much space for us unfortunately. However if we find out we have not enought stability, we will change the aproach on the fly.

    Well the C axis takes 3.75 sec for a full rotation. By watching some 5 axis machines working I think this should be ok. The force should be 120Nm. Should be strong enought. However about the exact gearing I will talk to my college who comes from a CNC job, if it even needs that force.

    Single beam sounds weak, I understand. But the Beam will have 20cmx20cm. Isn't that huge enought to resist quite a punch?

    Before buying anything we will definitely do severe testing. (Using a micrometer) we will measure, how much the beams deflect over a given length, also how much by its own weight.

    To get the precision beam, we need to pay a professional to get it done. Actually I want the beam to have at least 1/100mm. If not much more expense, I will even go to 1/1000. We still need to find out, how much we need to pay. If it is cheap, we might aswell just mill the rolling surfaces on every beam.

    Every beam will have a carriage consisting of 8 bearings, individually preloaded with a screw and than fixed with an additional nut.

    The only welding planned is with the beams. The sections have no enclosed volume, so are not torsion resistant. We need to weld 2 together side by side, to get an enclosed area like a tube has. I know it is not ideal without heat treating unfortunately. Just pray at that point. Everything else will be heavyly bolted with oversize holes, to have the ability to fine adjust everything on the fly.


    Don't be affraid, we will do severe testing before wasting any money. (Beams and electrodes are free and this is the starting point where it either works or the project fails at this point).

    Thanks.
     

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice