CNC Router Upgrades

I put my previous build on hold, mostly due to space constraints in my shop but decided to upgrade my present MillRight MegaV router. I made a kit to hopefully perform the upgrades in a day.

Increased Z travel to 6"
Increased height under gantry 6.25"
Conversion from V-wheels to linear bearings on X & Y axis
Conversion from Acme to ball screw on Z-Axis
Improve rigidity and overall accuracy of base/platen

This makes the machine very similar what MillRight calls their MegaVee Pro except that has a Masso controller. They want $8k+ for that. I have <$2k into the original purchase, $270 for the ball screw and all linear rails and bearings, and $200 for the aluminum tooling plate. -I like that price much better!

I will retain the rack & pinion drive on the X & Y axis, Nema 23 steppers, limit switches, controller, and beams. I debated replacing the beams too but at that point I may as well just sell the MegaVee and start a from scratch build of the other machine. Instead, I think I will make that other machine a smaller envelope dedicated metal/hard material machine.

Anyway, due to the previous effort, I had already done a lot of CAD modeling of the components which was very helpful for the design work.









Here's the kit:

Structural steel tube weldment skinned with .500" aluminum tooling plate:



The rest:









-More to come.

Best,
Kelly

 

I got it swapped over. Took the better part of a day then another half day to make all the little brackets for the drag chains, limit switches, and little details. The motion felt really nice after assembly. Need to do some leveling, tramming, and general tuning. Been busy but should have some time this weekend to get back down to business.



Best,
Kelly

 

Thanks Mark. The CAD work occurred over a long period an hour or so at time in my easy-chair so no idea what I have into that......it was more of entertainment.

I've been using it and continue to tune and tweak. The biggest chore will be leveling the bed plate. Though the aluminum tooling plate is very uniform thickness and flat, the steel weldment it mounts to is not as precise. Right now it varies about .010" across the entire surface, mostly low in the center, which isn't too bad considering it's just a structural steel weldment, but not good enough for me. For now, I can just mount a particle board platten and plane it with the router to correct it. In the long run, I may shim the underside of the tooling plate at the low spots. It's good enough for foam pattern work now and definitely an improvement over the original machine. -I'll keep plugging at it and make refinements over time.

I also don't know for sure how straight the aluminum beams are. I have steel beams that are precision ground but didn't use them. I may make a more rigid, precise, dedicated machine for aluminum work with a smaller work small envelope.

Best,
Kelly

 

Yes, but only for the initial rough adjustments. It really depends upon what metrology equipment you have to work with. For me that's pretty basic and not much. I actually started with a precision level and adjusting leveling feet on the table legs, but again that is just a gross adjustment. It quickly becomes a tail chasing exercise if the table isn't flat.

So I move onto getting the variation in distance from quill to table height as good as possible with a dial in the collet. This is a good next step, but again, it doesn't mean the table is flat, just that it will cut at consistent depth of cut. It can still have warp in any given direction. I have a precision straight edge that I borrow from a friend to check this, but there isn't much sense in doing it until you get things pretty close. But if you do all that, then precision level the table again, then check with trammimg head/dials in the collet to get the carriage and gantry plumb, you can usually get pretty close. It's a lot of doinking around, but I only need to do the whole dance once, and that is usually in multiple sessions of improvement as jobs require.

On my previous particle board bed plate, I had a set screw next to each mounting screw, so I just jogged the head/dial to each mounting screw, set the height with the set screw, tightened the mounting screw, and depth of cut was set. I didn't want to drill all those extra holes in this tooling plate. I'll need to shim the plate, which will be tedious.

I had seen the method in Mark's post before, but I don't have the laser and getting one and set up is as big of a task as just getting on with the iterative process which though crude, is probably good enough bearing in mind these are just extruded aluminum beams and although they are very good for what they are, they won't have the straightness of a precision ground and scraped beam. If I make a dedicated metal machine someday, I'll have to get more serious about the degree of precision, but for now, for the work I do, doing what I describe, then planing the surface of a sacrificial bed plate, or mounting, dialing and tramming a secondary platen will be close enough.

For doing foam patterns, if I'm within .005" variation depth of cut across the entire workspace envelop, it usually means a small fraction of that depending upon the size of my part. If I mount a piece of MDF and plane it, it can reduce variation to about what I can measure which obviously is well beyond that needed for foam patterns. The rack and pinion drive seem to repeat to <.003 in 24" of travel in X&Y. The Z ball screw seems to be repeating to within what I can read on a .001" graduation dial.

It's still just a hobby grade cnc router........need to be realistic about what it can do, whether the effort is worth it, and what's good enough.

Best,
Kelly

 

That tension stretching method has actually been quite common (and effective) in precision extrusion for many years now. I use to specify them and it was actually quite remarkable the profile tolerances they could hold. You're right, because if they are not handled carefully, it's all for not.

Best,
Kelly

 

Did some test and tune cuts and then took it on its maiden voyage.........machined the valley pan plenum shelf on one of my intake manifold castings. Then I machined the cover for the plenum from .25" MIC6 tooling plate. It has a number of O-ring glands for sealing the plenum. Came out Nice. The surface finishes were better than expected.

I made that MDF platform from my old platen bed board. It gets the stock up close to the gantry so I can use short stiff bits on thin stock. The aluminum bed plate still isn't level across the entire surface so I just shimmed this MDF platform until its surface was within a couple thousandths and went to it. I'll have a video of the build in the not-too-distant future.





Best,
Kelly

 

The video of the deed.



Best,
Kelly

 

Even though it's the same controller and motors, makes me pucker a bit since I have so little time on the machine since the upgrades. Lose a few steps and.......disaster. Scrapping a hunk of foam is one thing, but a large, finish machined, casting quite another. I wouldn't have chosen the castings for the maiden voyage, but I had to get a couple of them done, and thankfully things went smoothly. Large foam patterns are better candidates because cost of stock is cheap and it exercises the range of motion at high speeds. I'm looking forward to getting some more hours on it.

Best,
Kelly

 

Well, I’m at it again. I decided to convert my CNC Router from rack and pinion to ball screw drive on all axis. I guess I should have just done it in the previous linear rail upgrade. The R&P served me well for years, and though it does operate the steppers in the sweet spot of their torque curve, I’ve grown tired of fiddling with them. When I opened the clearance on the R&P such that it’s be reliably bind-free, there is lash/hysteresis that causes suboptimal positional accuracy. When I tightened it up to get the desired accuracy, small debris would occasionally finds its way onto the rack which inevitably causes binding and lost steps.

I looked at various methods and mechanisms for eliminating R&P lash (like Avid CNC uses for example), but for all the effort, compared to $275 worth of imported ball screws, I decided it just wasn’t worth the hassle, especially after observing the accuracy and repeatability of my previously converted Z-axis driven by the same ball screw, which performs excellently.

So, similar to my previous linear rail conversion, I designed a ball screw conversion kit that could be installed on my existing machine with minimal down time.



The lead screws are 16mm x 10mm I ordered online, three of them, all 1100mm long. The more common pitch for the Chinese ball screws is 5mm but that requires pretty high speeds from the steppers. 10mm pitch is a little less common, but it is just a double lead version of the 5mm, so in addition to only needing half the stepper speed, with twice the contact area, they should have double the load capacity, so that’s why I chose them. I already had a complete 2D CAD model from the previous iteration, so all of the dimensional interface was readily available.

Of course, I did have to design a couple castings into the conversion kit.



I fabricated the X-Axis steeper mount from a couple small chunks of ¼” steel plate. I wanted this to be as stiff as possible and since the modulus of steel is about 3x of aluminum, this should be about as stiff as a .75” thick aluminum version and take up less space. I added an outboard support bracket on the thrust bearing mount for good measure. I just designed the mounting brackets to use the FK12 ball screw thrust block, FF12 end support, and ball nut mount supplied with the SFU1610 ball screws for this axis. I did utilize a small casting for the end support since it just stabilizes the end of the BS and doesn’t really experience any significant loads.



I re-made the plate to mount the thrust block because two of the thrust block mounting holes landed in the previous stepper opening.



There’s a LH & RH ball nut mount for the Y-Axis. I vacillated a bit about their placement under the beam but the ball screw/thrust centerline is close to the contact point of the cutter and thus opposes the cutting force moment about the beam, and it made for easy packaging, and somewhat more protection under the beam rather than on top.



To minimize the level of disassembly, I removed, modified, and replaced the corner posts one at a time the day of conversion. Similar to the X-Axis, as opposed to machining the length of the ball screws, I just made the hardware to use the ball screws at the purchased length. I had to turn a dozen .5” diameter aluminum standoffs to mount all the steppers in the correct position, make a few simple brackets, and machined parts.



I put the steppers on the backside of the table. I could have put them on the front to make them more accessible, but I figured they’d just be in the way and how often was I really going to touch them once installed and up and running? I removed the R&P hardware but everything is reversible if I ever wanted to return to R&P drive, but seems quite doubtful I’d ever elect to do that. Here’s the completed conversion.



Besides accuracy one other small advantage of the Y-Axis/Gantry ball screw drive is two of the steppers that once traveled with the gantry are now stationary. This eliminated the need for one cable carrier and removed one cable from inside of one of the remaining two.

I relocated the Y limit switch, changed the stepper resolution settings to match the ball screw lead, fired it up and started jogging around the table. Initially I had a problem. When I called for high speeds, I was triggering alarms and causing quasi-crash like behavior. The microstep settings on X & Y stepper drivers were set on 3200 steps/rev which was ok for the high displacement per step of the R&P, but since the ball screws require higher stepper speeds for the same displacement, this was exceeding the stable output frequency of the Arduino Mega 2560 controller and high travel speeds. Reducing the driver microstep settings to 1600 steps/revolution solved this problem and still allowed speeds in excess of 230in/min. Some of my friends on the CamBam forum set me straight on this.

So I ran a couple sample programs (air cutting) and then cut a couple sample parts. I had previously damaged a couple of my dust shoe baffles so I decided they were a good candidate. They register in a step in the shoe casting and are retained with magnets.



My next upgrade may be controller related. Though the Y Axis has two steppers and ball screws, it uses a single driver to run both of them. This of course saves the cost of a driver but means you cannot separately home both sides of the Y drive making it self-squaring. My controller has a 4th axis. I may conscript that driver into service on the Y-Axis.

Funny thing with my various machine mods over time, the only thing left of the original machine is the beams, steppers, and controller.

Best,
Kelly