Welcome to Our Community

Some features disabled for guests. Register Today.

KRE Mill V3 - Concept

Discussion in 'CNC Mills/Routers' started by Kevon Ritter, Sep 13, 2018.

  1. Kevon Ritter

    Kevon Ritter Veteran
    Builder

    Joined:
    Apr 30, 2015
    Messages:
    597
    Likes Received:
    293
    The spindle cooling seems to not be leaking, so that's good. For that, I'm just using distilled water with a little antifreeze.

    My relay board is working just fine on 240V. At the moment, only the spindle cooling pump is on it. I would like to trigger that when the spindle is on. I would gladly put the flood pump on it too, but I can't find any 240V/60Hz pumps. I had an 800lph (200gph) pump on the older setup. I'd like to step up to an 800-1000gph on this. In the picture, the battery is just a "signal". I have to make a case for it.

    With the signal voltage issue figured out (just a setting that was set to 0-5V instead of 0-10V), I ran a 12000rpm test for about an hour and a half with the pump running. The spindle was warmer than ambient. That's about it.

    Now there is some funny business going on.

    1) The more physical area of contact that I make with the spindle, the more the frequency (rpm) on the VFD drops. At 12000rpm, I was able to drop it by over 1000rpm. Two fingers drops it by 300rpm.

    2) The spindle itself is NOT grounded. The shielded cable is grounded, but the internal wire that is attached in the spindle is currently disconnected. I left it like this for the possibility of having a cordless probe setup. Considered what is going on with the rpm, I grounded it while it was running. Here is where it gets weird. The VFD is hooked up to my dryer breaker (240V). As soon as I grounded the spindle to the VFD earth ground, I lost power in my room. I lost power to my 120V room after grounding on a 240V circuit. The controller is powered by a PSU on the 120V circuit. Although each circuit is earth grounded independently, I know for a fact that there are no ground loops. The only connection between them are two signals (0-10V and spindle enable), each as a pair of wires.
     

    Attached Files:

  2. Kevon Ritter

    Kevon Ritter Veteran
    Builder

    Joined:
    Apr 30, 2015
    Messages:
    597
    Likes Received:
    293
    I did some more messing around and found some pretty shocking results.

    1) I decided to increase micro stepping from 1/4 back to 1/8. This is where the original 1300mm/min ceiling came from. With the 1/4 stepping, I was only able to get around 1200. I played with too many variables during my first run and apparently just confused myself. Steppers are supposed to lose torque at increased micro stepping, so I found this result to be counter intuitive.

    2) I know that steppers react well to voltage, but I was not expecting a 33% increase from running at 29V instead of 24V. The ceiling increased to 1600. I turned it back down to 24V out of fear of voltage ratings, but it was definitely worth noting. I've been looking into 36V or 48V just for the heck of it, but now I really do want to go that direction.

    3) Remember how the spindle RPM would drop if I touched it? That was discovered with only the spindle hooked up. I hooked up the steppers as well, and guess what. Touching the spindle during operation increases the rpm. The shielding on the stepper wires are not grounded yet, but the spindle shielding is. The spindle itself is not grounded. I have not retested the "ground power out" due to the fact that I would rather not shut my desktop down right now. One idea I had to transfer the the control box PSU over to the 240V plug.

    4) I have thought about grabbing a 30t pulley to replace the 20t for speed purposes, but we'll see what improvements can be made form the above developments.
     
  3. Peter Van Der Walt

    Peter Van Der Walt OpenBuilds Team
    Staff Member Moderator Builder Resident Builder

    Joined:
    Mar 1, 2017
    Messages:
    13,749
    Likes Received:
    4,070
    But higher microstepping provides smoother interpolation, which decreases the stall point. So lower torque, but higher RPM
     
    Kevon Ritter likes this.
  4. Kevon Ritter

    Kevon Ritter Veteran
    Builder

    Joined:
    Apr 30, 2015
    Messages:
    597
    Likes Received:
    293
    I'm officially upping the voltage with a Meanwell SE-450-48 (48V 9.4A). To support it, I could have continued using the MW LRS-350-24 (24V 14.7A), but I chose to grab a smaller generic 2A power supply strictly due to the size.

    The new Phoenix is also on the way.
     
  5. Kevon Ritter

    Kevon Ritter Veteran
    Builder

    Joined:
    Apr 30, 2015
    Messages:
    597
    Likes Received:
    293
    I haven't been updating here much, but the project is moving along. I'll post updated pics when I wake up. This post is more about numbers.

    I made a centralized power source with 240V and 120V on the same circuit. The rpm drop/rise that came with physically touching the spindle is gone.

    The 48V PSU definitely improved overall performance. I don't quite remember what X/Y maxed out at, but I currently have it set to 2000mm/min as an easy safe spot compared to the 1200mm/min on the 24V. The only issue now is the Z axis overheating. After a while, maybe 30 minutes, the Z starts to lose steps. The only thing I can think of is heat. X and Y are both mounted to aluminum plates which act as giant heatsinks. Z is thermally insulated due to it being temporarily mounted with an ABS printed part. The initial 1500mm/min has to drop down a bit after a certain duration of time. A temporary solution would be to add a fan (or two) directly to the stepper.

    Considerations to make are the 5mm pitch Z and 10mm pitch X/Y. Z is running at approximately double the steps/mm as X/Y.

    As for X/Y, I do wish I could get more from them, but they just don't have the torque at that speed (600rpm at 2000mm/min). We don't know if it will start skipping under heavy load. This won't be known until I take a massive cut with a massive end mill. Let's say that it does hold up at 2000mm/min. My goal was 2500mm/min. One solution is to swap the current steppers (269oz) for a closed loop setup. The other option would be to reduce the gearing which would easily accomplish this at the cost of losing torque......if the torque curve was constant.

    Take a look at the torque curve. Ultimately, the total torque available from the motor and gearing stays the same no matter what the setup is untill you exceed 300rpm (in this case). The only thing lost here would be resolution. The drivers are currently set to 1/8 micro stepping. That's 1600 pulses per revolution. After gearing, that's 480 pulses per mm. That's a theoretical resolution of 0.00208mm per pulse. That is not achievable on anything that doesn't have expensive servo motors on a granite and steel frame. Swapping from 20t pinions to 40t pinions drops the ratio from 3:1 to 1.5:1. That changes the theoretical resolution to only 0.00416mm per pulse which is probably still impossible to reach at this level. Since drastically decreasing torque should no longer be an issue, we should now be at 4000mm/min. Even if we want to factor in major drivetrain losses, we're still going to be well over 3000mm/min. Considering the extra torque available, I doubt that this will be an issue. At 1000mm/min, the X/Y steppers (after gearing at 24V) will be making 150Ncm compared to 135Ncm. At 2000mm/min, the X/Y steppers (after gearing at 24V) will be making 68Ncm compared to 55Ncm. Although I've been saying torque over and over, the issue isn't really torque. The issue is inductance. There's a point (rpm) where the energized state of the stepper windings can longer be changed fast enough. This is what creates the rpm ceiling.

    torquecurve269-2.8anema23.png


    The cast aluminum for the tooling plate will be here tomorrow (technically today after this long post) along with an assortment of larger end mills. I have Techniks Super Precision 3/8" and 3/16" collets on the way scheduled for Friday. I already have the included 6mm that came with the spindle and a 6.35mm, also from Techniks. They are about $15-16 each compared to the $25 for a full set of the ebay/amazon special, so I'll acquire them as needed.
     
    Andreas Bockert likes this.
  6. Kevon Ritter

    Kevon Ritter Veteran
    Builder

    Joined:
    Apr 30, 2015
    Messages:
    597
    Likes Received:
    293
    One issue was the alignment of the magnets to the hall effects. When I designed the mounts I assumed that they should have been centered. However, there is a required offset. I'll just have to do a bit of manual work.
     

    Attached Files:

  7. Kevon Ritter

    Kevon Ritter Veteran
    Builder

    Joined:
    Apr 30, 2015
    Messages:
    597
    Likes Received:
    293
    Every form factor has its strengths and weaknesses, but I didn't see this one coming. The plate is 0.5" cast and started out at 13.875" x 15.625". The gantry clearance is only 13.25". It's a simple ramping cut, part rotate, and repeat cut. This will give me a piece of approximately 12.25" square.
     

    Attached Files:

  8. Kevon Ritter

    Kevon Ritter Veteran
    Builder

    Joined:
    Apr 30, 2015
    Messages:
    597
    Likes Received:
    293
    The upper "U" will be lined with led strips.

    I'll pick up some triangle stock to act as an inner splash shield for the bottom edge.

    I can conveniently grab a 28" x 30" sheet of acrylic from Lowe's, which is exactly what's needed. I don't know if I'll make it double pane or not. For the bend, I'll stretch a scrap piece of MDF between the gap to prevent the middle from buckling. From there, a heat gun and another flat scrap should get the job done just fine.

    I'll probably just use gorilla tape for the back edge.
     

    Attached Files:

  9. Kevon Ritter

    Kevon Ritter Veteran
    Builder

    Joined:
    Apr 30, 2015
    Messages:
    597
    Likes Received:
    293
    The acrylic bent perfectly!
     

    Attached Files:

  10. Mark Carew

    Mark Carew OpenBuilds Team
    Staff Member Moderator Builder Resident Builder

    Joined:
    Jun 28, 2013
    Messages:
    2,742
    Likes Received:
    2,407
    Really well done @Kevon Ritter been following along, keep up the great work love it!
     
    Kevon Ritter likes this.
  11. Kevon Ritter

    Kevon Ritter Veteran
    Builder

    Joined:
    Apr 30, 2015
    Messages:
    597
    Likes Received:
    293
    Thank you! This was a heck of a design learning curve. From the other thread, I'm definitely considering stretching it now. Too many projects. It needs an ATC!
     
    Mark Carew likes this.
  12. Rob Taylor

    Rob Taylor Master
    Builder

    Joined:
    Dec 15, 2013
    Messages:
    1,470
    Likes Received:
    746
    Wish AMB would let people know if the 1050 ATC works on the newish 1400. Or just bring one out for it, either way. Continuous power between them is only 100W or so, should be fine.

    It's funny, looking back through this thread. Guess I know where I started thinking about M4 now, eh.
     
  13. Kevon Ritter

    Kevon Ritter Veteran
    Builder

    Joined:
    Apr 30, 2015
    Messages:
    597
    Likes Received:
    293
    I really like the FM30F from CNCDepot. It will even fit this machine nicely. Only problem is it's $4000.

    @Rob Taylor I haven't checked in on yours in a while. I know it was definitely a monster in many ways.
     
  14. Kevon Ritter

    Kevon Ritter Veteran
    Builder

    Joined:
    Apr 30, 2015
    Messages:
    597
    Likes Received:
    293
    Just for gits and shiggles, this is with the FM30. I raised the gantry 30mm. I would need a few extra extrusions, but for the most part, the entire machine is reused.
     

    Attached Files:

  15. that-jim

    that-jim Journeyman
    Builder

    Joined:
    Apr 19, 2020
    Messages:
    143
    Likes Received:
    148

    This is awesome. Now you have me thinking about what I would change before I even have my build done. Like maybe I could have a tool height probe off to the side of my table, for semi-automatic tool changes.
     
  16. Rob Taylor

    Rob Taylor Master
    Builder

    Joined:
    Dec 15, 2013
    Messages:
    1,470
    Likes Received:
    746
    Just noticed on these you also chose to do the "split 8080" orientation with the C-Beam. Hadn't even noticed that before. Surprisingly flexible at larger sizes, as I detail in my M4 thread, so careful if you're widening it.

    Yeah, I've looked at the S30 series of spindles a few times. They're beautiful, and not really that expensive relatively speaking, but beyond the budgets of most of these builds. Even the $2k ISO20 spindles at half the price are a bit of a stretch. The AMB spindles are half the price again and get probably 90% of the way there for most things, but the 2HP unit has no ATC. Yet- just did a bit of stalking of their youtube channel, and the quick-change 1400W model is already available and the ATC version is "coming soon." As of a week ago. So that's pretty exciting. I'll probably look back into it in a few months.
     
  17. Kevon Ritter

    Kevon Ritter Veteran
    Builder

    Joined:
    Apr 30, 2015
    Messages:
    597
    Likes Received:
    293
    Having tool height sensing is an invaluable feature. I used a thin strip of metal near the homing point with some custom code on my Sphinx and that saved so much time. Sadly, I lost that macro coding when GRBL Panel died. I wouldn't be surprised if that functionality is baked into OB software now.

    Another thing I can point out is to cover as much as you can. Check out the very first post in this thread. I didn't have a 3d printer at the time to get really fancy, but even using the slot covers helps immensely when it comes to keeping things clean.

    Flexibility was a huge concern. It's a U channel construction with 4080 beams on both top and bottom and a 60mm gap which is filled with a 4060 beam. The 4060 doesn't extend full length, but far enough to put in work. There are rear plates as well as internal corners. It's not a solid a single piece, but hands down much better than two floating beams. To be honest, I don't think the bridge is even necessary at the original 500mm rail length. I wouldn't skimp on it for the larger 1000mm ATC concept. Speaking of which, a little over 200mm worth of travel is off to the side with a 24" (600mm) wide bed.

    The other concern is just controlling it all. The two options I saw before were Planet CNC and Acorn.
     
  18. Rob Taylor

    Rob Taylor Master
    Builder

    Joined:
    Dec 15, 2013
    Messages:
    1,470
    Likes Received:
    746
    That's the same dimensions I came up with as well, so I'd have a usable internal channel for screws and maximum space between rails for rigidity. I'm just using 2060 spacers at the corners, but I'm gonna have full-area 140x1500mm or whatever plates covering the back sides which should help. On the (~1200mm) gantry I'm also putting full-width steel 1x3x1/8" rectangle tubing top and bottom. Not sure yet if I'm gonna fill anything or keep the open channels yet, we'll see how it goes. But that's what the FEA said I needed as a minimum to stop it twisting, so I just gotta figure out how to make it work.

    I'm sticking with LinuxCNC for M4 because it'll be my second LinuxCNC machine with Mesa hardware, keeps things consistent. If I were starting from scratch, I'd probably go either grblHAL+bCNC (great for macros) or maybe with the Centroid, which I've heard good things about from the few people who use it though I'm not really sure how it works.
     

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