Welcome to Our Community

Some features disabled for guests. Register Today.

Robo Morten

Discussion in 'CNC Mills/Routers' started by JimB828, Jul 8, 2017.

  1. JimB828

    JimB828 Well-Known
    Builder

    Joined:
    Jul 23, 2015
    Messages:
    21
    Likes Received:
    15
    JimB828 published a new build:

    Read more about this build...

    [EDIT: I'm moving some of my update information from the build page to here, so that I can update the build page with more detail on the actual build, instead of my updates.]

    Update #1: 2017/07/15

    The concept works, and I'm getting somewhat accurate cuts in test pieces. However, the cuts aren't accurate enough for use. Part of the problem, because this machine does not have a lot of rigidity, is getting everything truly square. Also, because the router sits horizontally, weight is pulling the bit to a slight vertical angle. In order to combat this, I am in the process of trying to turn the vertical C-Beams around so that the moving gantry will be on the backend of the machine.

    This effort has uncovered another problem: the counterbores in the C-Beam end mounts are not only NOT flush with the end mount, each one has a slightly different depth. This is causing the two vertical C-Beams to not sit flat, and is making it very difficult for me to get each to be perfectly aligned. I've re-tapped the holes in the C-Beam--thinking that the taps were not deep enough, which was true in a couple of cases. However, holding the end caps in my hand and sitting the screws in the cap shows that it is the counterbores.

    View attachment 25562 View attachment 25563 View attachment 25564

    Update #2: 2017/07/16

    Because of the issue that I was having with the counterbore depth in the C-Beam end mounts, I had to find a way to raise the end mount off of the table surface. I had a piece of 20x40 which I laid flat on the table on the 40mm side. I also created a bottom frame using 3 more pieces of 20x40 (each approximately 500mm in length).

    View attachment 25615

    Once the frame was square, I used a 2 hole angle bracket to screw into the 20x40 and the front screws on the end mount (that are supposed to be for the C-Beam shield).

    View attachment 25610

    However, because there is no support on the back of the C-Beam, it was not staying true at a 90 degree angle from the surface of my table (the top of my table saw).

    View attachment 25612

    I knew I needed to recreate the "table" on the machine to hold material for both the mortise and tenon functions; in other words, to be able to clamp material flat and on its side to the machine. I added supports to the front of this table to push back against the vertical C-Beam to keep them at a true 90 degree angle.

    View attachment 25614 View attachment 25611

    The "table" has a front and a back, so that I can clamp pieces to the front for mortise work; using the back to clamp against. Of course, that makes it more difficult to clamp pieces for tenon work, so that's another problem I will have to solve.

    Checking everything again, the machine seems to be square, except the horizontal C-Beam supporting the router, which I need to fix; it is off just a hair from side-to-side.

    This is version 1.1 of the machine.

    View attachment 25613 View attachment 25609

    Version 1.1 Testing

    Testing the tenon job that I've created works generally well. However, it's oversized by .5mm (X) in length and .25 in width (Y). I can't tell if the gcode I generated out of VCarve pro is faulty, if the Tinyg is off, if Chilipeppr is causing me problems (it always does; I can't wait to find an alternative), or if my config is off somehow.

    Using the Acme 8mm lead screw, I've got travel per revolution setup at 8mm; which I don't think is correct. But, changing that setting results in crashing the axes. I've got more work to do to figure out what's going on.

    Testing Update - SUCCESS

    The table that I built into the machine isn't size correctly for the mortise work. I mis-measured the movement of the spindle, so I had to start taking the table apart to move the frame around. I will have to continue to work on the best setup.

    The 8mm per revolution seems to be correct. It's possible that my calipers are seriously inaccurate. I was able to successfully mill both a mortise and tenon; although, I need to work on getting the alignment correct.

    View attachment 25624 View attachment 25623
     
    #1 JimB828, Jul 8, 2017
    Last edited: Aug 16, 2017
  2. phil from seattle

    phil from seattle Journeyman
    Builder

    Joined:
    Mar 17, 2017
    Messages:
    47
    Likes Received:
    11
    Interesting. I've been thinking about a CNC based pantorouter for awhile now. I've also been thinking about how to CNC something along the lines of the Incra LS Positioner which I use to cut box and dovetail joints. The idea would be a machine that would cut a number of standard joints with a simple interface. No CAD needed - just input some basic dimensions and it makes the cuts.

    Please post about your progress
     
  3. JimB828

    JimB828 Well-Known
    Builder

    Joined:
    Jul 23, 2015
    Messages:
    21
    Likes Received:
    15
    Phil, right now I am using Chilipeppr on a Tinyg to control the machine. I really dislike Chilipepper; and, it's not necessary for most of what I want to do with this machine. Once I nail down the design, I am expecting that I will have to either switch to another controller, or write my own interface to do exactly what you describe--plug in some data and let the machine go.
     
  4. THE LINK

    THE LINK New
    Builder

    Joined:
    Apr 30, 2017
    Messages:
    20
    Likes Received:
    0
    Can you update the parts. Vlearly uses more extrusion.
     
  5. JimB828

    JimB828 Well-Known
    Builder

    Joined:
    Jul 23, 2015
    Messages:
    21
    Likes Received:
    15
    Yes. I rebuilt the table portion this weekend, so I have a few more tests to make sure that the current build works well for the scenarios I have in mind. I will update it this coming weekend when I am back from travel.

    I am also working on a cad model.
     
  6. JimB828

    JimB828 Well-Known
    Builder

    Joined:
    Jul 23, 2015
    Messages:
    21
    Likes Received:
    15
    I have updated the BOM for this as it stands today. There may still be changes.
     
  7. JimB828

    JimB828 Well-Known
    Builder

    Joined:
    Jul 23, 2015
    Messages:
    21
    Likes Received:
    15
    The machine is workly fairly well, but I'm still about 1.25 mm off on alignment between the mortise and tenon. Figuring out the math and making sure that I am setting 0 off of the correct corner for each piece is tedious.

    On top of that, Chilipeppr quirks are driving me absolutely crazy, and have crashed the machine a few times. I added homing switches to the machine, but even with those, I've had enough of Chilipeppr problems.

    I spent this weekend switching from Chilipeppr to CNCJS, which seems much nicer; if a bit simpler. But, in the process, I've gotten also quite frustrated with some quirks and possible bugs with the TinyG. I've just placed an order for the xPro controller to try out grbl to see if I can make this more reliable.
     
  8. JimB828

    JimB828 Well-Known
    Builder

    Joined:
    Jul 23, 2015
    Messages:
    21
    Likes Received:
    15
    The xPro controller seems to have been a no-go. I can't even get the motors to turn. I'm not sure if the board is bad or if I am missing something else. Instead, I spent this weekend installing the Planet-CNC Mk3/4 controller with (4) DQ542MA stepper drivers. It took me a while to figure out the software & configuration, but I have gotten the machine working again. After working with the Mk3/4 controller, I realize how limited the TinyG platform is given its price. I will never use that platform again for a new machine. I may use the board for an automated stop on my miter saw station.

    Now I have to figure out the differences in gcode support between the Mk3/4 and the TinyG so that I can get my probing routines figured out.
     
  9. phil from seattle

    phil from seattle Journeyman
    Builder

    Joined:
    Mar 17, 2017
    Messages:
    47
    Likes Received:
    11
    try setting $4 to 1.
     
  10. JimB828

    JimB828 Well-Known
    Builder

    Joined:
    Jul 23, 2015
    Messages:
    21
    Likes Received:
    15
    Thanks phil. I will try that when I am back home on Friday.
     
  11. JimB828

    JimB828 Well-Known
    Builder

    Joined:
    Jul 23, 2015
    Messages:
    21
    Likes Received:
    15
    I have not tried getting the xPro controller working again, but I have been testing out the Planet-CNC controller, with an upgrade to their TNG software. It's working well, and it seems like the controller will make it easier than either the tinyg or the xPro to create some macros for the future.

    I have also made some changes to the mounting of the gantry and the router, and added the C-beam cover to the Z-axis to decrease the amount of chips getting into the C-beam.
    _BI_7649.png _BI_7650.png
    I have had a little problem with the mini v-wheels loosening up inside the vertical C-beams. Unfortunately, because I originally used cast 90 degree brackets on the gantry plates, the only place to mount them was with the screws holding the wheels. I added spacers to each of the vertical C-beam gantry plates, and another gantry plate to mount the brackets.

    I'm going to add the C-beam shield to the other axes, but I need to change how the vertical axis is mounted to fit the shield there, and I need to change the way I am mounting the limit switch on the gantry axis for that shield to fit.
    _BI_7651.png _BI_7652.png
    In trying to test the tenon and mortise jobs, finding the edges, and accurately centering the cuts was a bit difficult using my triple edge finder. I purchased some aluminum duct tape, which makes it easy to find the edges on all four sides of the tenon piece.
    _BI_7653.png _BI_7654.png
    _BI_7656.png
    Finding each side of the part, makes setting the origin point to the middle of the piece easy. The tenon jobs cuts well.
    _BI_7648.png
    I'm also using the aluminum tape to find the edge of the mortise piece.
    _BI_7658.png
    I had either some backlash in the system, or I was missing steps, or both. The first mortise test I ran today started well, but quickly blew out.
    _BI_7660.png _BI_7663.png
    Somewhere, I must have put too much pressure on the Z-axis, because the gantry plate holding the router became loose. After fixing that (again), the mortise job ran successfully.
    _BI_7664.png _BI_7665.png
    So, I'm back to where I was a few weeks ago, but things seem to be more reliable now. Unfortunately, I'm still off 1.5 mm in one direction and .5 mm off in the other for joining up the mortise and tenon. I have no quite figured out where my calculations are going wrong. However, the fit of the tenon is almost perfect.
     

    Attached Files:

  12. phil from seattle

    phil from seattle Journeyman
    Builder

    Joined:
    Mar 17, 2017
    Messages:
    47
    Likes Received:
    11
    looks pretty good so far.

    I'm a little unclear on how you set up to cut a piece. It looks like you are finding the edges of the workpiece and setting their position in you CAM SW. Is that right?
     
  13. JimB828

    JimB828 Well-Known
    Builder

    Joined:
    Jul 23, 2015
    Messages:
    21
    Likes Received:
    15
    Yes, this is correct. I'm using the aluminum tape as a makeshift probe, since it wraps around the piece nicely.

    For the tenon, this is the process I am using so far:

    * Find the upper right X and Y, probing to each and using G92 to set them to zero. I do NOT correct for the diameter of the bit or the thickness of the tape.
    * Find the lower left X and Y. Since these are now offset from the zeros set in the previous step, for each, I divide by two and set G92 again (G92 X<posx/2> and G92 Y<posy/2>
    * Find the Z position and set it to the thickness of the tape (.005 for my tape) and set it using G92

    Now the zeros should be set directly in the middle of the piece.

    I am still trying to get an accurate mortise process. Since I am obviously miscalculating something, and ending up slight off on positioning. But this is the process I am using so far:

    * Find the upper right X and Y and set them to the opposite offsets from the tenon: X gets set to -<posy/2> and Y gets set to -<posx/2>
    * Find the Z position and set it to the thickness of the tape (.005)
     
  14. JimB828

    JimB828 Well-Known
    Builder

    Joined:
    Jul 23, 2015
    Messages:
    21
    Likes Received:
    15
    I have made a couple of stupid mistakes in working through test pieces, and I have crashed the router into the part numerous times. While I have been fortunate enough not to break the bit, it keeps causing the mini v-wheels and the Z-axis gantry plate to come loose.

    Because I have to take apart half of the Z axis to get the v-wheels tightened up again, I am really starting to not like using those wheels on the interior of the C-beam. I am considering switching to the X-large gantry plate so that I can use the larger v-wheels on the exterior of the C-beam. That way, when I make mistakes, it becomes a quick turn of a wrench to tighten things up.
     
  15. JimB828

    JimB828 Well-Known
    Builder

    Joined:
    Jul 23, 2015
    Messages:
    21
    Likes Received:
    15
    Since I had to take the Z axis apart to tighten up the eccentrics on the mini v-wheels, I went ahead and changed the X-large gantry plate, with the delrin v-wheels on the outside of the c-beam. I then ran another test cut, and it is almost dead on. I am very happy at the moment. It's still not quite as accurate as I would like, but it's usable now.

    With the X-large gantry plate, I had to move the Z axis a bit to prevent the bit from crashing into the table when moving down the Y axis, because where the Z axis was positioned, even with the router all the way to the top of the gantry, the bit still overhung the axis. If I stick with this setup, a 250mm Z axis isn't long enough. I need another 75mm or so to be safe. We'll see how this setup functions.
     
  16. phil from seattle

    phil from seattle Journeyman
    Builder

    Joined:
    Mar 17, 2017
    Messages:
    47
    Likes Received:
    11
    I know it's early but have you thought about how it will eventually be set up to run the canned M/T macros. I've been looking closely at the pantorouter and see it has one huge advantage - it's really easy to set up. It's biggest disadvantage (besides cost) is the need to have specific templates for each joint. I doubt the average joe could make a template so the power of the pantorouter is somewhat untapped. If rotomorten set up could be made as easy as the panto, it would be a huge winner, assuming there is an easy way to create new templates.
     
  17. JimB828

    JimB828 Well-Known
    Builder

    Joined:
    Jul 23, 2015
    Messages:
    21
    Likes Received:
    15
    Yes, I have been spending quite a bit of time, in between being able to actually test the physical machine, trying to figure out a usable way to make this consistent. On top of the somewhat inconsistent performance of tinyg with chilipeppr, the chilipeppr interface seems unnecessarily complex, and I don't like the way that it is architected. Trying to make a consistent macro interface seemed much more difficult than it should be. CNCJS seemed to be a bit easier to extend, but the problems I was having with tinyg meant I couldn't accurately test anything. Since I wasn't initially able to get the xPro (grbl) controller working, I moved on to checking out the interface for the Planet CNC Mk3/4 with their TNG software.

    While the TNG software is still in beta, and it is lacking any kind of documentation, I have already been able to figure out (in about 30 minutes last night) how to create a dialog box in TNG that will let me enter parameters for the mortise and tenon. It will require that I make my current gcode into a parameterized macro that can be run directly from the interface, but I know that it is now possible.

    Creating templates for the pantorouter will still probably be easier than creating new gcode macros from my machine. One advantage I see is that the CNC aspect of robomorten allows you to do 3D relief carving and similar functions, for which the pantorouter doesn't really seem suited. The pantorouter will still most likely be faster than robomorten, since I believe it uses the larger sized router supporting 1/2in bits.

    Eventually, I will have to figure out better ways to clamp and set up the physical pieces being routed on robomorten; because, right now, every piece has to be measured using an edge-finding process.
     
  18. phil from seattle

    phil from seattle Journeyman
    Builder

    Joined:
    Mar 17, 2017
    Messages:
    47
    Likes Received:
    11
    The g-code for most wood joinery should be pretty simple. What I've been thinking about is a sender/configure program running on something like a RasPi could be made to do it. I could be naive here but the sw shouldn't be that complex (especially for a guy with 30+ years as a sw engineer). The actual machine control would be done by a grbl based controller (or similar gcode interpreter) while the Pi (or what ever) would do the UI and generate the GCode programs to send to the machine. I've toyed with automating an Incra LS positioner via an iGaging scale hooked up to a teensy+LCD. No motion control involved but it used simple templates to guide the process.

    On set up, the panto just uses a fence and table approach. Measure your workpiece, clamp it down and position the template. For a woodworker, this is basically what you do with any tool. Getting the fit right on the panto seems a bit fiddly, by the way. A robomorten approach could just use fixed X and Z points (via limit switches like Z on a 3D printer perhaps) as a reference. Setup should then consist of selecting/mounting the desired bit, entering workpiece size, bit dimensions and joint location, placing/clamping the workpiece, homing the machine in X and Z, setting Y home and hitting go.
     
  19. JimB828

    JimB828 Well-Known
    Builder

    Joined:
    Jul 23, 2015
    Messages:
    21
    Likes Received:
    15
    I hadn't thought about using the limit switches as reference for positioning. I obviously use them as homing switches, but I haven't measured the repeatable accuracy of the mechanical switches. I also purchased some inductive proximity switches, but I have no experience with them, so I don't know if they are more or less accurate than the mechanical ones.

    Ideally, I want to get as close to the process that you've described as I can. Today was mostly spent on squaring up every single aspect of the machine. While it was performing well, there were just enough slight errors that things were off. I need to recut the tenon again to test edge alignment again--since the last one I cut was before I eliminated some of the errors. But, the fit was as close to perfect as I've ever accomplished--I had to tap slightly with a dead blow to get the tenon into the mortise, but I was able to pull it apart by hand with a slight bit of force.
     
  20. phil from seattle

    phil from seattle Journeyman
    Builder

    Joined:
    Mar 17, 2017
    Messages:
    47
    Likes Received:
    11
    That sounds like a perfect M/T - it should hold from friction but still be removable without undue force. When you glue it up, the moisture from the glue swells the tenon a little so you don't want it super tight when dry.

    In 3D printing, it's very common to just use a microswitch as the Z zero point. On my printer, It's generally within .1 mm of the actual bed and is quite consistent. This is despite the fact that the switch is zip tied to a mount. (that's getting changed soon). 1/10 of a mm is far more accurate than even the most exacting wood worker could expect. Probably the biggest challenge to accuracy would be keeping the switches clean.
     
  21. JimB828

    JimB828 Well-Known
    Builder

    Joined:
    Jul 23, 2015
    Messages:
    21
    Likes Received:
    15
    I am traveling for the rest of the week, so I will not be able to do any more testing until Friday evening. However, I did come up with an easier way to perform setup late last night. I'm a bit embarrassed that it took me so long to think of it.

    Instead of using the aluminum tape, and having to make sure it's flat when wrapped around the piece, I took a 3/8in zinc bolt with a washer and soldered a wire to it; and then, I threaded the bolt into a 1-2-3 block which is accurate to within .0005in on all sides (and I think most of that .0005 is actually the caliper, not the block).

    I created a macro in the Mk3/4 TNG software that requests the size of the workpiece for the tenon, homes the machine, and then runs through a process of using the 1-2-3 block as a probe to reference everything off of the table, and the left top edge of the workpiece. The whole macro and setup of the workpiece takes less than 2 minutes to run at the current speeds.

    The macro doesn't currently do any manipulation of the gcode to create the tenon, but it does make it very easy, and consistent to setup and clamp a workpiece on the table for a tenon job created in an external CAM package. While the tenon job has to be created externally (for now), the setup process includes a dialog that requests the tool diameter and the width and length of the workpiece, and all of the probing commands run off of those entries. I'll take some pictures this coming weekend to demonstrate the process.

    Of course, towards the end, when testing the macro, I made a mistake in the gcode and ran the end mill into the side of the workpiece pretty hard, which pulled the router gantry out of square. I will have to fix when I get home.
     
  22. JimB828

    JimB828 Well-Known
    Builder

    Joined:
    Jul 23, 2015
    Messages:
    21
    Likes Received:
    15
    Go ahead and rub it in. ;)

    Because I wasn't using the table as a reference, I needed a way to probe to the underside of the board. My edge finder wasn't working well, since I had to hold it in place, and it was easy to let slip. Now that I realize my mistake, the edge finder still isn't ideal, because it's not a consistent offset in both directions.

    The 1-2-3 block is incredibly accurate, and I can use it in any orientation. I already had 3 pairs, so using one for this purpose is easy, and I can remove the bolt to use the block elsewhere.
     
  23. JimB828

    JimB828 Well-Known
    Builder

    Joined:
    Jul 23, 2015
    Messages:
    21
    Likes Received:
    15
    I didn't make much progress this weekend. I've found another source of accuracy error, and that is that a couple of my v-slot parts are not completely square at the ends. I spent two days trying a couple of different ways to true up the ends with no success. I tried the router method used by DazTheGas, and all I found was the router tearing up the ends of the v-slot. I tried many different ways to trim up my miter saw to make sure it's making 90 degree cuts, and there is slop somewhere. I am incredibly frustrated at the moment.
     

Share This Page