Pre-Build Log, Looking for Lessons Learned

[Thomas Utley]

All,

I've embarked on a mission to convert a discarded piece of industrial machinery into a three-axis CNC mill and would appreciate your sage advice prior to finalizing my design.

The salvaged equipment was some sort of pattern duplicator that allowed the user to follow a template with a stylus and replicate the design in what I'll call 2.5 axes (see photo below--the Z axis is nothing more than a compound slide from a lathe, with the toolpost replaced by a spindle mount for a small router).

What caught my eye were the good quality (and hardly used) Thomson linear bearings. The longer rods are 1.5" OD. The smaller ones are 1" OD. I've since purchased a third set of bearings for the shorter 1" OD rods shown in the photo to fabricate a real Z axis.

I'm looking for lessons learned as I finalize my design. My goals are:

1. Accuracy over speed
2. Easily handle anything from foam to hard woods and plastics
3. With care and feeding, handle softer metals including aluminum and brass, with no aspirations of milling steels

The salvaged hardware dictates my working envelope, currently X = 28", Y = 22" and Z = 6". My first spindle will be a Porter-Cable 7518 variable speed woodworking router because that's what I have on hand. I think I understand its limitations, but would appreciate any thoughts on making it more capable of milling aluminum as I start looking ahead to using the machine to rebuild itself with more durable components. Eventually I want a high speed VF spindle that runs quietly, but that's most likely an option for my next machine.

My bed is a torsion box, 5.5" thick. It's mostly air but with stiffening goodies in all the right places inside. It will be overlayed with a sacrificial bed made from MDF or melamine slotwall after I get everything dialed in. I'm happy with its rigidity and flatness given the conditions under which it was fabricated--certainly nothing the mill itself can't flatten once it's up and running.

The gantry is made from shallow torsion boxes made from hardwood plywood sandwiched between Masonite skins. The hardboard was chosen for its ability to withstand compressive loads from all the hardware being bolted to it/through it. That, and it was what I had in my scrap rack at the time.

You'll note from the CAD screenshots that I opted to hang the X-axis pillow blocks from the sides of the bed rather than set them on top. This was to open up the working envelope as much as possible. I recognize there is no increase in usable travel from the Y-axis rods, but I can still reach the center of a much larger workpiece this way.

My steppers are 305 oz-in run by a home-built controller kit. They seem to run fine on the bench.

The leadscrews are all 1/2"-10 single start ACMEs to give the little steppers a fighting chance at starting (and stopping) the mass of the various stages. Each axis will have thrust bearings riding short pieces of drill rod nearest the stepper motor with the far end of the leadscrew free to float axially.

The A/B nuts are Dumpster CNC's acetal version on all three axes. Motor couplers are helical aluminum units. I actually bought Lovejoys, but they arrived with about 2° of axial slop from the supplier and I quickly decided against using them.

Finally, nearly all of the fabricated pieces (most of the pillow blocks, stepper mounts, thrust plates, etc.) are made from what used to be the base of the salvaged machine shown in the photo. It's glass-reinforced phenolic sheet, 3/4" thick. If you're familiar with the material, it's about 2/3 the density of aluminum but roughly 90% the tensile strenth and 60% of the shear strength of 6061 aluminum. Not quite as tough, but I can cut it with my table saw and, much like the Masonite above, was free from my scrap pile after removing all the old hardware. Free is good, especially with the wife scrutinizing this project as my "most" insane so far.

What am I missing before I start cutting out the fabbed parts this weekend?

Thanks,

Tom

[acondit]

Hi Tom,

Welcome to the Zone.

The thing that I see immediately is the X rails appear to be unsupported along their run (supported only at the ends). They are obviously heavy (I think you said somewhere that they are 1 1/2" rods). Have you calculated the sag at the center of travel with the weight of the gantry and Z?

Alan

[Thomas Utley]

Alan,

Great question, and yes I have calculated it. Had to go back and dust off my mechanical engineering beam deflection formulas for a prismatic beam supported at both ends with a point load in the middle. The reality is it's not a point load since it's spread over two bearings which span 25% of the beam. Still, taking it as a point load is conservative, and rounding up to the max gantry weight I anticipate seeing:

max deflection = (Weight * unsupported length ^ 3)/(48 * modulus of elasticity * moment of inertia for a round beam)

= (30 lbf * 37 in^3)/(48 * 30,000,000 psi * 0.2485 in^4)

= 0.0042" max deflection

Taking into account the real loading conditions which tend to disribute the load away from center, I'm anticipating about 0.0035" deflection in Z when working toward the center of the envelope.

Note that as the gantry is moving away from or toward the center (the low point), it's also tipping slightly. This causes X-axis position error depending on high high above the X-rods the business end of the spindle is located.

Could I improve on that by running on supported bearing rods or along the edge of the torsion box? Absolutely yes. For the price, however, I can live with that much deflection, especially since it's "mostly" in Z.

[Thomas Utley]

Followup note...

What I didn't do, before now, was run the same deflection calcs for the Y-axis. It's obviously less mass and a shorter span, but it's riding on 1" OD rods vs. the 1.5" OD X-axis rods. What a difference a half inch makes...

= (20 lbf * 29 in^3)/(48 * 30,000,000 psi * 0.04909 in^4)

= 0.0069" max deflection

Ouch.

I suppose by the time I get the machine squared up and the sacrificial bed surfaced by the machine, it will have a nice concave working surface before I clamp any stock to it...

Any experience with harmonic vibration on a system like this? My worst fear is that the first time I run the tool into something moderately hard, that it starts singing like a canary as the tool hits some harmonic of the gantry and the whole thing tries to self-destruct and kill me with shrapnel. Ha.

[acondit]

I got the final pass set a little deep once and I had steel screws just flush with the top of the spoil board. I was cutting 75ipm with a 1/4" carbide bit in my Porter Cable router. It changed pitch for a second but never slowed down and milled the head of the screw just as nice as you please.

Alan

[JD_Mortal]

Please forgive my limited knowledge and observational skill, as I have not built a CNC, however, I do have a lot of mechanical knowledge through use.

On your X-axis, (I call it the suspension bridge), you have a single-drive screw in the center with the guide rails being real far from the drive. Even with the best linear guides on the rails, you will get a lot of left-right angular play and jitter as the drive pulls from the center.

Eg, the left will "rest" but the right side will "slide", then the left will slide as the right side rests. This will be more prominent and damaging as the tool above mills and hovers over the left or right sides. (That side will become a pivot and the other side will be the only side moving, causing motor and drive-screw stress and damage.)

You can alleviate this with two things...

You need a longer mounting-length for the bottom of your bridge, x-axis, so the four points form more of a "square" and less of a "rectangle". The rectangle is shorter on the drive-length, which is the reason you will have no true linear length-pulling power. The long side has to be on the length, to get equal left-right pulling forces on the side mounts.

However, you can regain your linear control by adding a slide-rail under the drive-screw, (The slide-rail mounts would still need to be as horizontal in-line with the drive-screw as possible.) This will give you back your linear-accuracy with the left-right sides of the x-axis, suspension bridge base, and also reduce stress and wear on the drive motor.

EG... you have this now... in your design...
(| = x rails, - = x base, # = drive screw)

|------#------|
|------#------|

(] & [ = linear guide) Note, the linear guide would be along the full lenght of the mill-table, this only shows the closeness of the mounts on the x-axis base.

|-----]#[-----|
|-----]#[-----|

Alternative, not unless you have lots of room, or want to loose milling length... (Square pulling forms a triangle)

|------#------|
|------#------|
|------#------|
|------#------|
|------#------|
|------#------|

The triangle in there... (x)

x------#------x
|------#------|
|------#------|
|------#------|
|------#------|
|------X------|

The triangle in the linear guide... (x)

|-----x#x-----|
|-----]X[-----|

The triangle in your design... (x)

x------#------x
|------X------|

Sorry, there was a third option...

You could add another drive-line (screw). Place each one of the screws and mounts, as close to the linear slides as possible one left and one right. Use two motors or one motor and link them with a chain/belt drive.

|#------------#|
|#------------#|

The triangles... (x)

x#------------#x
|X------------X|
.
.
.
.
..........X

ignore the ..., the X is the virtual intersect of the triangles. (There are actually two overlapping elongated triangles, this is the virtual average.)

Also note, I protest against multiple motor drives... due to slippage, they can burn each other out and cause even greater wear and hardware damage, without "smart" encoders and self-adjusting correction computer hardware.

[Thomas Utley]

JD,

I must compliment you on two things. First, for a guy who's never built a CNC, you have a very good grasp of the challenges of laying one out, including the balance between ensuring smooth linear motion and maximizing travel. Second, yours has to be the best "ASCII CAD" I've ever seen--that was a brilliant way of conveying a fairly complicated mechanical concept without geometry on the screen!

I do appreciate the warning, and I sincerely hope it doesn't come true once it's all bolted up and running. My only insurance against racking (the slip/stick condition you're referring to) is the oversize X-axis bearings I'm using. They're 9" long, which is roughly 25% of the total length of the "suspension bridge." There's always a trade between length of your rails and max travel--each inch you steal to separate your bearings from one another for motion stability is an inch you give up in travel. I'm stuck designing around the salvaged hardware I started with and until I get this one working (and hopefully making a little bit of cash for the next one) I'm bound by its limitations.

If I do wind up with motion problems when cutting away from the X-axis leadscrew centerline, my only recourse will be to center up my workpiece on the bed and work as close as possible to the middle. If that doesn't work, then it's back to the drawing board.

I wonder else what I could make out of a bunch of round rod and bearings...? Ha.

Stay tuned for progress posts, and thanks for the insights!

--Tom

[JD_Mortal]

Thank-you for your compliments...

Depending on the forces the machine will be generating, a simple glide-block using "L" framing from a bed-frame, and four bearings aligned on the horizontal, will offer superb linear guidance. (It will also reduce racking-stress, keeping your expensive slide-bearings lasting longer. The bearings are a sacrificial element, as the slides are, compared to the steel-guides.)

If you setup the brackets so they are about 1" away from the screw drive, running the length of the screw-drive, to the ends, and the rollers mounted on the four corners, as close to the forwards/rear edges as possible... Your minimum play should be less than 1/1000th of an inch, at the slide-bearings on the outer guides. Unless the force is too great, then it will bend or warp the brackets. The "L" brackets forming an upside-down "U" around the screw-drive and drive-mounting block.

I just don't trust single-pole glides, after seeing them bind so easy on injection molding presses. (Granted, the injection molding presses I operated were as small as 5-ton and as large as 100-ton devices, and they all had multiple guides, so each helped keep the others aligned under the most stressful extensions. They only remained close when they free-glided in an open position, chattering and digging in to the 20" steel poles, even with all the grease and oils used to reduce friction.)

Ok, done blurbing... lol...

Yes, I am now learning the lingo, so I feel less alien, before I begin my first cut into the design I am working on. Trust-me, I am researching, crash course style, every available option on the market. (I am leaning towards something that doesn't involve the use of 100-lb $100.00 poles as guides.) I am actually falling in love with all the MFD designs, and the "V" guides... This is the next best thing to nothing, and part of an "every-day man's" budget and skill. But around here, hollow plumber pipe and fence-pipe is cheaper than any extrude aluminum sold at a hardware store. They are so greedy on that stuff! They have a 800% mark-up on the lowest quality junk.

[Thomas Utley]

JD,

I like the concept, and I'm sure it would work if I end up needing to add it.

As for different layouts you might consider (since you're starting from a clean sheet of paper), it depends a lot on how big of a working envelope you need. If you want to cut parts from full sheets of 4' X 8' material (and don't forget MDF is actually oversize at 49" X 97"!), then it's tough to design around leadscrews. Most wind up going with rack and pinion layouts, at least along the long axis.

Why not start your own build log here so we can see what you're thinking? Hand sketches can be posted just as easily as CAD screenshots, so don't be shy about adding pictures.

--Tom

[Thomas Utley]

So this weekend I drilled a LOT of holes. Small holes. Big holes. Holes perpendicular to other holes. 190 holes total just to fab the parts for the Y and Z axes. There's another 40 or so left to finish up the X axis parts next weekend. Can I just say I really wish the CNC was up and running to do the positioning for me? Oh, wait, nevermind...

Ironically I do have a really nice X-Y table on my old Wilton drill press. Unfortunately it only has about 5 1/2" inches of Y travel so it wouldn't reach all the holes in the larger plates without repositioning. And if I had to reposition, then what's the point of using it, right? So I reverted back to paper...CAD drawings printed 1:1 scale and taped over each blank. The axis lines served as centerpunch crosshairs, followed by a centerdrill, followed by the real drill. Plus more than a few counterbores where I need clearance for the socket head cap screws.

Patience...

[acondit]

Tom,

One way to deal with repositioning accurately, is to have a hole at a fixed distance that is a snug fit for a dowel (in the fixture). Then before moving the part, drill a new hole in the part that is a fixed distance from the dowel hole (it can be a hole that is required hole for the part if you lay out your fixture to allow for it). Then when you lift up the part, insert the dowel in the fixture and position the corresponding hole in the part to fit over the dowel. (Clear as mud?)

Alan

[Thomas Utley]

Thanks, Alan. That's a great suggestion!