DIY Mill Turn Center

[Khalid Khattak]

Great professional job...

[Generic Default]

Thanks for the support everyone!

This is the azteeg x3 pro wired with 8 steppers. They're all drawing 2.25 amps. I need to get a fan on this thing! And an enclosure.

On the first or second page of this thread I talked briefly about a repeatable limit switch for this machine. My current prototype version is shown below. I ran a few repeatability tests with a dial indicator to see how consistently it homes on the X axis (turret).
I did 20 homing cycles in a row and checked the position of the dial each time; the needle was consistently within 0.0002". Pretty good for a 2 dollar homemade limit switch made of stuff from the local hardware store and with one 3d printed piece! I can't reliably indicate any more accurate than 0.0002" so I'll have to get an expensive indicator if I want to determine the exact repeatability value. It's still much better than the clicky limit switches common on 3d printers.

And here is my test of a harmonic gear. Since I can't get the surface finish good enough for an elliptical shaped plug and ball race, I have to use bearings as rollers on the ID of the flex spline. The flex spline is actually a small assembly with 4 bearings in it to keep stuff stable. I'll work on a self contained unit capable of a lot of torque and zero backlash.

The hypocyloid gear type you suggest I make looks like it works the same in fit and function as a strain wave gear, but cheaper to machine on a mill or live tooled cnc lathe. A reduction ratio of 30 or more with zero backlash, high maximum torque, good efficiency, and no backdrive-ability is what I need on this machine. Plus it has to be compact and not too expensive.

So for rotary axes my choices are down to;
Dual Worm
Roller Strain Wave
Hypocyloid


Right now on the C axis (the one that the workpiece spins around) I just have a clutchable spur gear and I'm still getting low backlash and up to 3 rotations of the C axis per second, which is good enough for light threading operations. If I switch away from the spur gear system I'll lose the ability to thread at a reasonable speed but I'll gain the ability to take heavy milling cuts far from center. I need to get a sensored BLDC motor and put an encoder on it so I can thread with it, then use the low speed gear system for milling.



I'll get a video sooner or later.

[Captain Starfish]

Bump...

Any progress on this masterpiece?

[Generic Default]

What he said. More photos and videos of the beast in operation please, but what you've shown us so far is a thing of beauty.

How are the plastic guides holding out so far, considering smooth motion in the 'allowed' degree of freedom vs stiffness in the others?

Rather than the strain wave harmonic drive style, also worth looking at hypocycloidal eccentric gears. Nice smooth curves = relatively easy to machine the rotor and the ring can be done with some steel rod stubs. Considering putting one together to rebuild my el cheapo chinese 4th axis which is currently belt driven off a NEMA23 stepper and which I can backdrive by hand - even if it's only a bit of 8mm rod in the chuck.

Question: it looks like your X is driving the A spindle back and forth along the rails, how do you keep the toolstock synched if, for example, you're just using a dead centre in it? Slaved secondary lead screw on the tailstock, does it clamp to the same one, or what?

[youtube]http://www.youtube.com/watch?v=1GkhRZA5BUY[/youtube]
http://www.youtube.com/watch?v=1GkhRZA5BUY

Finally got around to making a public video. Sadly, I have had a video up since last July but I can't show it to the public for IP reasons. The above video doesn't show any of the extraordinary features of the machine so I don't have to worry about public disclosure with it.

To answer your questions:

1) "How are the plastic guides holding out so far, considering smooth motion in the 'allowed' degree of freedom vs stiffness in the others?"

The acetal pads aren't showing any signs of wear. The un-anodized aluminum boxways are slightly burnished where the acetal pads contact them. It's not a loss of dimensional tolerances really, more like the crests of the trochoidal tool marks are getting lapped into a mirror-like finish. The wear rate of the system is much lower than any metal on metal system. It's also quite rigid at the contact surfaces, the main cause of displacement is the thin (6mm) aluminum locking gibs on the bottom, and the small contact area of the set screws on the sides.
I improved the side-load rigidity by putting steel shims in the gap between the side plates and the acetal wear pads, so the steel shims distribute the set screw force over a larger area.

2) The C axis (main workholding spindle) has a clutching mechanism that switches between the BLDC outrunner for regular turning/drilling operations, and a 4:1 geared down stepper for tapping and threading and indexing for milling and stuff. It's great for small parts, but anything over 1" (25mm) is at risk of backdriving the stepper during heavy cuts. I have a completely different design for V2 of the machine that will be improved in every way.

3) The "tailstock" is identical to the main headstock, just on the other side of the machine. To synchronize the headstock and tailstock, you just command them to move the same amount in the same direction with gcode. I haven't been able to use this yet because I'm deprived of free time and my board doesn't allow me to use more than 8 steppers right now; once I upgrade to a board that allows more steppers, I'll be able to put the fully functional tailstock to use.


As-is, this version of the machine is much more useful than most of the hobbyist level machines around (anything under $2500). The problem areas of the machine are mostly in power transmission, which I didn't plan very well. I could have saved a lot of time by just using simple plum couplings from ebay on all of the motors. The second problem area of the machine is the base tube, which is much less rigid than I planned because the local metal place only sold it with 1/8 wall thickness. I can fix this by filling the thin tube with epoxy and sand, but it would take some disassembly so I'll just stick with the flexy tube for now. You can actually see this flex in the video during the grooving ops, at a few points the entire headstock assembly lifts up a tiny bit for a fraction of a second due to the cutting force exceeding the weight of the assembly. For reasonable cuts this doesn't happen.

Overall part tolerances are coming out around +- 0.0025" without putting much effort in, if I properly set the tools I could get this down to around a thousandth of an inch.


Version 2 of the machine is bigger and a hell of a lot more rigid, I just took every design flaw of the first machine and fixed it in multiple ways.

[Captain Starfish]

Ye gods

That is ... beautiful

<sobs with joy>

[joe]

Wow. Very nice. I was proud of the CNC router I had built from scratch but now not so much. Great job from what I can see. Can you show any other pics?

[Generic Default]

Bit of an update here....

I started out with the concept for this thing in Fall 2015, was building it by February 2016, and I had the first prototype working and making 5 axis parts by July 2016.

Now it's August 2017. I started building a version 2 prototype (V2) at the beginning of July 2017, based off of the design I had mostly completed by spring of this year. Now I'm approximately 50% of the way done with the version 2 prototype.

That's the back story. I also withheld some......strategic........information, specifically that I have been planning on doing a crowdfunding campaign since the very beginning. And also that this machine has additive capabilities. It killed me to have to withhold this from you guys, especially since I have had a video of it working with a hybrid additive/subtractive part since mid 2016. It's one of the reasons why none of my photos on this forum show the entire machine, even though it's right next to the camera!

V2 is similar in function and form to V1, but improved in every way. It's much larger and heavier. While V1 can cut plastics and metals with small tools (under 3/8"), it lacks rigidity and power. Plus I had to make it on a minimal budget, which forced me to go as cheap as possible on the machine and make design compromises. As a result it is difficult to align to the tolerances I want. V2 avoids this alignment entirely, and is significantly more rigid and powerful. I've attached a few photos, I'll be resuming this build log.

I realize that this forum thread may seem unrelated to 3d printing at this point. I still have my rostock max v1 behind me right now, and I'm using it to make certain non-rigid parts of this machine. The V2 machine will have 3d printing capabilities as well. I'm aiming for a Kickstarter late this year, my main problem right now is that I've spent $1,000 on this prototype so far and I need about $2,000 more to finish it, so I can't really make any progress until I get funding. I still have a bunch of pictures from the build so far, I'll be posting on this thread a bit over the next few weeks. Ask away!

[Captain Starfish]

Shiny. Real shiny!

I take it that the plastic gib on box ways concept is still going well for you?

Let's have some specs - work envelope, max tool sizes, spindle power, etc etc etc. Also if I bought/built something like this is there an option for a third or replacement X carriage to do more traditional plate style (as opposed to rod) 3 axis milling?

Now you've let the cat out of the bag, please keep us posted. My 3D printing has dropped off considerably over the last year or two since I got my CNC router, and I'll admit the only reason I come back here to visit now is when I get an update notify on this thread.

Keep us posted, and keep up the fantastic work!

[Generic Default]

Yup, the plastic slider boxways are great. If I were going to build a machine with very long travels, I would consider rolling elements for efficiency. But boxways are a lot more rigid, and are pretty much immune to contamination. Since most parts can be done with only a few inches of X and Z travel, it makes sense to stick with boxways.

Work envelopes:
6 inch chucks----> 7" dia x 10" length
ER40 Collets-----> 1.1875" dia x 12" length
16C Collets------> 1.625" dia x 12" length
5Ax Fixture------> 6" square x 10" length
3Ax Bed----------> 12" long x 6" deep x 6" tall


Each of of the things listed above is a different way of holding work. The machine is optimized for mill-turn work, but can do horizontal 5 axis work as well as basic 3 axis stuff as well. It won't have the same large Y overhang as most 3 axis mills, so plate/sheet work will be limited in size.

Turning spindles: ~10 ft*lbs torque, 4 horsepower, 2500 RPM for chucks, 4000 RPM for collets
Milling spindle: 5 ft*lbs torque, 4 horsepower, 10000 RPM, peak power at 5000rpm. Max tool size 3/4" shank (ER32)
Live tools: 3 ft*lbs torque, 3 horsepower, 6000 RPM. Max tool size 1/2" shank (ER20)

The turning spindles will definitely have an indexing mechanism, I have both a harmonic drive and a worm drive tested and working right now. All motors (except steppers) will have a 2048 count pulse encoder, which will allow closed loop control and rigid tapping on all spindles.

Machine axis movement speeds will be dependent on stepper or servo. Steppers won't slide an axis more than about 120 IPM. Servos might get it up to 600 or so, which is more than enough.

There are now 3 serious contenders for low-cost servos. VESC, O-Drive, and ClearPath. The first two will need firmware mods to work with any CNC controller. I'll be using the Smoothie v2 pro when it comes out, which allows an indefinite number of stepper/servo drivers to be connected. Plan is to offer the machine as a modular kit, so people can buy sub-assemblies and put it together in 30 minutes or less for the mechanical stuff.

Some close ups of the boxway system;

Note that the boxway locks are not tightened in the second photo, which is why they are at a slight angle.

[Captain Starfish]

Impressive. That's a lot beefier & bigger than the pics / drawings suggest. I didn't get a sense of scale and the green got me thinking "oh, similar to the PocketCNC or the taig micro stuff". Nope, sounds about the same ball park as my SIEG AL60 and X2 which I've found of great utility over the years.

It sounds like it has the guts and ratios for steel cutting, or is it not stiff enough for that kind of savagery? Thoughts?

Probably a little early still, but any thoughts on a ballpark price tag?

[Generic Default]

It will definitely be able to cut steel, otherwise I've been wasting my time. Cutting performance will be on par with bridgeport knee mills and hardinge HLV lathes. The bearings I'm using are the same size, the motors have the same power. My main concern is static rigidity of the B axis, which is the part that the milling head rotates on. I need to keep the overhang distances short, but even then, any torsional deflection in the B axis harmonic gear will translate to tool tip deflection. I won't be able to measure this for a while.

On the version 1 prototype, I have a 5 inch (125mm) 3 jaw chuck, and I can turn steel parts with it so long as I keep the depth of cut shallow. Version 2 is an order of magnitude more rigid since everything is solid metal. The base of this thing is a solid 4x4 aluminum bar, 4 feet long.

The boxway slides are 8x10 and 10x10, which is similar to knee mills and medium sized manual lathes.


I'll have to re-sum my costs to give a price estimate, but I can tell you that it will be less than you'd expect.

[Captain Starfish]

Damn, dude.

*hits 'subscribe' again so many times the button breaks*

[Generic Default]

No updates for the last month because I ran out of money; I just got funded by a very generous person a few days ago.

Back to making parts now, I just bought the metal for the vertical milling column (y axis) and the headstock block. So hopefully I'll have time to machine those over the next week or two.

Besides seeking funding, I redesigned quite a bit of the machine over the last month. The headstock(s) now have a pulley box on the back that is almost fully enclosed to keep chips and coolant out. It still has an opening for the spindle bore to allow long pieces of bar stock to pass through. I want to keep it easy to modify because I'm planning on adding a drawtube mechanism that will allow the machine to run automated, pulling the bar through after each part and clamping on the bar with the collet. Bore size is 1.89", which lets it use 16c collets for a through-bore capability of just over 1.5 inches. The 6 inch chuck can do 1.75" through bore. Spindle bearings have been sized up to 65mm to allow for a thicker spindle tube, this should increase rigidity as well.

The back side of the headstock will have a 1:1 chevron belt pulley; the pulley shaft that connects to the motor will have an absolute rotary encoder on it, looks like 4096 resolution and an index pulse. The pulley box also housed a worm gear system that can be engaged to give the machine a more rigid 4th axis. I still have to work out the pivot mechanism for the worm gear so it engages the big gear when an air cylinder is actuated. This is a very compact setup, it would be a lot easier for me if they made tapered roller bearings down to 1/2 inch bore.


Also I should mention the driving force for each linear axis. I'm still using 1/2 inch leadscrews with a fast travel pitch in acetal nuts. The nuts are 3 inches long, and they're held in place by a light press fit to take forward and reverse forces as well as 7 M6 screws. They're adjustable for backlash by turning a SHCS from the back, which compresses the nut around its flexible bottleneck joint a tiny bit. The acme threads are tapped with the same leadscrew that they run on, which means backlash is near-zero without compensation. The adjustment screw is there for long term use, I'm guessing the threads will wear a few thousandths after months or years of use. I'll be testing this quite a bit.....For now I only know that the backlash is well under 0.001".

[forrie]

An amazing bit of engineering you have going on here, very well thought out. I only found this thread recently but I have to say you have my attention. I know it's not a working prototype yet, but if it ends up having the work envelope, tolerances and capabilities you say it does and you can manufacture the thing in quantity....I think you are onto a real winner...I know I want to buy one!



.

[Captain Starfish]

Right? I'm camped on this thread for updates, hoping that my long dream of amalgamating my old lathe and manual mill with CNC and extra bench space.

[Generic Default]

Here's a cross link to that old post on another forum about the electric motors for this machine. Same thread from 2016:
https://endless-sphere.com/forums/viewt ... 30&t=76363
Looks like for now I'll be going with 130kv 63mm outrunners, since that's the only thing available. It doesn't make sense to spend twice as much on a larger motor when my wall outlet can only do ~1500 watts.



I made progress on the main headstock block today. I still have to rough out the center hole, line bore it, and make 100mm bearing pockets on both sides of the block. The block is 6x6x7", and I'm trying to keep the bearing pockets round, square, concentric within a few ten-thousandths of an inch over the six inch distance. I'll show the method as soon as I get it done. Japanese NSK bearings are on the way.

Here are some pictures I took while machining the block today;

[Generic Default]

Double post because nobody responded. This headstock is a pain to make without the right machines and tools. I had to do two interpolating operations to get a rough through hole. Then I had to move it to another machine to bore the through hole to act as a reference surface so that the back side bearing pocket can use the bored through hole as a reference surface.

Of course the probe stem was just a tiny bit too big; this means I'll have to wait a while for a special coaxial indicator so I can align the back bore. All of this effort to get some very accurately aligned bearing pockets.

I wish this could be built like 3d printers, where you don't have to worry about precision to get a working machine.

[Captain Starfish]

No lathe fun times for the bore on these? Or at least a boring head to keep things coaxial?

I guess if the plan is to flog kits ie make a few of them it's not just a case of knocking one out but of also working out how to most efficiently get good results.

I'm liking the photos. You look like you have a decent setup!

[Generic Default]

There was no boring head to use for a 3.94" bore. The tool used in the last post to plunge the through hole was just a fly cutter in a big ER40 tool holder. The purpose of probing with the VMC was to align the front bearing pocket to the through hole; the part will be flipped and aligned again for the back pocket. This is the only good way to do it without very specialized tools.

I got the main spindle bearings in yesterday; they're P5 grade angular contacts with a 15 degree contact angle. 65mm inner diameter is as big as I could get in this 6 inch headstock housing. The spindle will have a 6" chuck with a 1.77" through hole, so the machine can turn nice big bar stock without having to cut stuff off before chucking. Plus it can fit a 16c collet. Having the biggest bore possible makes the machine a lot better to use.


And today I got most of the milling column done. It rides on a 10x10 boxway, and the column itself is 8x8 for real rigidity. Most of the hobbyist machines use columns of 3x3 or 4x4 thin walled tube, some use smaller. Having a large column will contribute to the stiffness of this machine.

You can't really see in that picture, but the tall bars on the front are the boxways. The top will be braced, and everything will be bolted together and aligned.

I think I'll do this a little bit differently in production; I'll eliminate a few components and try to machine the boxway surfaces directly into the column supports. That way there will be no alignment needed, assembly will be easier, and the mechanical accuracy and rigidity of the system will be improved.

[Generic Default]

Almost finished the BY block; this is the part that slides up and down vertically on the Y axis while holding the B axis shaft and the milling head. See the render a few posts back if it's confusing. This part just needs a few M4 drilled and tapped holes to put a bearing cover on in the hexagon.

I assembled the milling column and put it on its boxway, then adjusted the preload gibs to check binding moments. I was able to slide the entire milling column on the boxway by pushing at any point on the vertical Y rail, which means the machine won't bind at a ratio of 2.5 or less with the cheap acetal pads I'm using right now. Plan is to make the production version with a better (and more expensive) type of sheet polymer, but since they will be laser cut from sheets, the overall cost will be justified by ease of manufacture. My cheap pads right now have a coefficient of friction of around 0.25 I think. The better ones will be around 0.1.

Good news is that my boxway system seems to be incredibly rigid, and there is no perceivable looseness or slop from any direction.

Anyone have questions?

[Captain Starfish]

No questions at the moment for me, you've explained over this thread pretty well how it will all hang together. Looking fantastic, can't wait to see more progress.

[forrie]

No questions, just interested in seeing this beast coming together and what it is capable of machining. Pictures are super cool!

[Generic Default]

I was hoping to have the turret block done today and in these pictures, but I broke a big carbide endmill while roughing out the pocket. I'll have it done by the end of the week I think.

I ordered the steel for the spindles and toolholders, so I should be turning those next week. Still waiting for the right type of spindle motors to get stocked. The linear axis motors are NEMA 24s, they're almost identical to NEMA 23s but with larger shafts. Of course I got them with dual shafts to attach handwheels.

I've been thinking about the kickstarter for a long time now, I'll need to have the machine working and making parts with all axes before I can film the pitch video. My main concern at this point is the control board. When I built the V1 prototype last year, I was planning on using the Smoothieboard V2. It never came out, so I had to use the azteeg pro board with 8 stepper motor controllers. The firmware didn't allow me to do 5 axis stuff like I wanted.

Smoothieboard V2 has been in development for a long time now, they say on their website that they're supposed to be released in late 2017. If that doesn't happen, I'll be in a bit of trouble. I need a board that can control 8+ motors that costs no more than a few hundred dollars.

[Captain Starfish]

Also worth considering how much the kickstarter is going to eat out of your potential funding, I get the impression from a few who've written about the experience that they ended up with a lot, lot less than they were expecting by the time the proverbial pound of flesh was extracted. I also think the kickstarter/indigogo et al crowdfunding boom has been and gone and punters tend to see the potential for doing their dough for nothing rather than the promise of getting stuff off the ground. Just a thought: if you can find a manufacturer, marketing and distribution partner you'd probably be a hell of a lot better off.

Looking at things like the SkyFire VMCs, you may find you don't need anything other than to advertise here and maybe on CNCZone. The SkyFire machines took crazy time, like a year or two, from order with payment to delivery. The guy took peoples' money, set up his supply and manufacturing chain, got things built and sent to a few early birds, then shipped to everyone when the kinks where ironed out. Same as Kickstarter just without the 30% kick in the wallet from them.

Pity about the controller. I was about to get excited and suggest something from CNCDrive (my UC-400eth has been lovely) together with PC but they only go to 6 axis and I seem to recall you wanted to avoid the whole computer connected thing. Might be worth talking to somewhere like Masso: they are doing monolithic controller boxes to replace the old Mach3 + BoB combo and are getting some decent reviews. They're fairly new too, so they may be hungry enough for sales to consider a variant with a couple more axes...

[forrie]

The Duet + expansion board would give you the axis count you need. No idea if the drivers are good enough to do what you want though, might need external drivers as well.