[Hamzter]

Hi All,

I’ve had my SX3 for about 6 weeks now so I thought it was about time to start converting it to CNC, and decided I should document it while I’m at it. I’m doing this one slightly different to most other builds, in that I’m planning on using belt drives on all 3 axes. I’m doing it that way so I can easily change the gearing to get the optimum speed/torque, and also to minimise the dimensions of the machine (therefore decreasing the size and cost of the enclosure). The design’s still at a very early stage, but I thought I’d post here to get some opinions before getting too deep.

Attachment 216114

I’ll be using NEMA23 stepper motors on the X and Y axes, and either a NEMA23 or 34 stepper on the Z, I’m thinking of upgrading the Z to a servo drive in the future too to improve z rapid speeds, but I’ve not given it any serious thought yet. I’ve managed to arrange the motors in a way that don’t crash into anything at maximum travel, but are still below the table surface in case I want to clamp down something particularly large.

Attachment 216116
Attachment 216118

For the X axis I’ve designed some mounting brackets with the thrust bearings on the LH, and a radial bearing on the RH (to take the radial loads from the belt). Travel looks like it will be about +179mm/-245mm, though anything above 165mm starts uncovering the ways so that limits the usefulness past there.

Attachment 216120
Attachment 216122

For the Y axis, I’ve got the thrust bearings and radial bearing at the same end, the thrust bearing spacer block’s been designed to maximise travel by ensuring the saddle touches the column before the ballscrew nut touches the bearing. Currently the travel looks like it will be about +95mm/-82mm (total 177mm/7.0”). If I flipped the ball screw nut around, I could actually get +95mm/-112mm, but this would leave less than half the ways in contact, plus complicates the bearing block. Does anyone know if any useful work can be done with this ammount of way exposed?

I’ve added a bolt and washer to the end of the screw to prevent losing my balls in the event of an overtravel (at the cost of about 5mm travel), I think it will need a LH thread to prevent it unwinding when it hits the nut though.

Attachment 216124

The Z axis will be a fairly typical mounting affair, using the column bolts. Rather than machining out of a massive block of aluminium like I’ve seen on a few machines, I’ll be bolting together plates.

I’m still considering how best to mount the pulleys, currently I’ve just got a set screw against a flat on the ball screw/motor shaft, I’m concerned there’ll be slop in this though so I’m thinking about drilling and threading straight through the shaft (might be tricky on a Ø8 motor shaft). Anyone got any advice on the best clamping method?

For tensioning the belts I’ve currently just slotted the mounts (10mm slot to allow for different pulley/belt combos), hopefully this can hold the tension when running, if not I might be coming up with a more elaborate mechanism.

I’m looking at getting ballscrews with preloaded nuts from CNCFusion, I’ll be needing slightly different lengths to their standard kits though so I’m hoping they can accommodate. While I’m waiting for their response, I’ll be designing the belt covers, limit switch mounts and starting on the electrical design, expect updates soon Any comments welcome!

Graham

[neilw20]

No point in Servo on the Z. A stepper can crash tools at more than acceptable speed/feeds.
The reliability will decrease as you end up with 10+ wires instead of 3 or 4.
Consider, maybe, 3 phase hybrid steppers. More than up to the job. Nema 34, for sure, direct drive on the Z screw.
Using belt drive will certainly decrease positional repeatability due to the springy nature of the belt(s).

My Syil has the Y offset from center a bit, allowing to drill OVER the edge of the table. A definite advantage.
You need a hole in the tower to clear the Y screw to get reasonable travel.
Uncovering the X ways by 30mm each end is definitely OK.

Make sure you include an oiler system, from the start.
With lots of the way exposed on Y axis, you need to be able to grind the gibs every 200 hours.
Consider linear guides. That would be a much better solution.

At the end of the Y screw, a crude circlip is fine. My Y screw runs inside an aluminium torch housing. Yellow is pretty.
Z axis? Direct drive underneath, or on top of the column. Again, avoid a belt if you can.
Z axis really needs linear guides, as the overhung design will always have spring back/head nod.
Stability of Z column?. That is a REAL problem. Mine is attached to a BRICK wall at the top. Best thing I ever did.

Z gas strut? Don't over compensate. 2 struts, 20KG loading will be perfect. Make sure Z can travel ALL THE WAY to the top.
You can't have too much Z clearance for longish drills.
A 50-60mm thick machined plate under the column, to make room for Z motor is a good move.
Lower part of existing Z travel is never used, unless you rotate the head (which is a pain to get back to vertical again).
Consider being able to mount a small 3 jaw chuck. It makes a great CNC lathe, whether vertical or horizontal.

Spindle speed needs to be at least 3000. Consider pulleys to achieve this.

Direct drive on motors has a definite advantage. Use couplers, for sure.
Side/bending loads on stepper shafts will fatigue them, and they will break!!
Waste the space, and put a window in your enclosure.
I just hang heavy rubber mats, by magnets, from head.
No enclosure. How do you put something even slightly oversize on the table, let alone look behind it?
A big splash tray, with removable pieces.
These are my observations, after using my machine for over 2000 hours, at almost flat out speeds.

Preloaded ball screw nuts, for sure. Limit switches? I used 2 on Z. I can reference on lower one saving travel time to top one when referencing.
All opto slots, anywhere in the travel, that is convenient.
X and Y are under the table and are surprisingly reliable. My Y way just has e piece of rubber sheet held in place by magnets.
Magnets get covered in chips, but saves them going on the floor. Keep it simple.

I can work to 0.002" X and Y, and even better in Z.
The quill is the most useful part of the machine. Don't disable it, as some have. Why would you?
Add a small removable stepper to the quill drive, and you almost have something that can be used as CNC EDM with almost no mods.
The quill drive would need limit switches to make EDM very simple.

[Hamzter]

Great post! Given me a lot to think about...

I should explain why I'm after higher feed rates, I intend on fitting an additional high speed spindle to do some fine machining/engraving work, which means higher feed rates than a direct drive can provide. For that reason, I think I'm going to continue with a belt system, but after your comments I'll plan to make it easy to change over to direct drive if I ever feel the need. Big pulleys and wider belts might be in order to minimise stretch.

I'd not considered an oiler system yet, primarily because I don't have access to another mill. Has anyone ever come up with a way of incorporating one without milling oil grooves in?

Unfortunately I don't have a brick wall to bolt to (yet), I was thinking about bolting additional steel plate inside the column, I might do some calcs/FEA to see what kinda improvement I could get. Linear rail Z axis will definitely be a future upgrade.

I'll look into the fatigue issue on the steppers, i found some load ratings that suggested they should be fine for this purpose but might consider an additional support bearing if it looks like I might have issues.

For the enclosure, I was thinking of something with quick detach panels to help with cleaning and oversize workpieces, but it'll be a while till I get onto building that so I'll see how I'm doing without one first.

Brilliant idea on the EDM mod too, i can see that being a future project

[TangentAudio]

I have a ProMiCA conversion kit on my Franken-X3, and all axes are driven with MXL timing belts. The drive system has been completely reliable for me for over 4 years. I think that commentary like "side loads on stepper shafts will fatigue them" is way off the mark. Steppers are used in belt drive arrangements all the time, in all sorts of machines and systems. With timing belts you do not need extreme amounts of side-load to tension the belts, since the meshing teeth are used for transferring torque. Just enough tension to take up the slack, and you have a backlash-free and reliable drive system.

It does makes for a very compact package, and like you say, you can easily adjust your drive ratios. I believe my entire machine is 2:1 ratio (motor:leadscrew). I also use NEMA23 motors on all 3 axes - STP-MTRH-23079 | Stepper Motor: NEMA 23, 286 oz-in hold torque, 5.6 A/phase

I believe ProMiCA has since gone out of business, but their web site is still up. Take a look at the photos of the conversion to get some ideas. IMO he did a good job with the design, which is why I chose it. PROMiCA - CNCKits MX3 Kits for the X3 Mill

[Brett248Vista]

Tangent,

Thanks for sharing this info. The CNC Fusion kit uses a timing belt for the Z and I didn't think it would be an issue or CNC Fusion would have changed it.

I do like Neil's suggestion about using a pair of Gas Struts instead of one.. I will either do that, or a counterbalanced system.

[Hamzter]

A what balance system?!? I hope that's a typo

Thanks for pointing out the ProMICA kits, that gives me a bit of confidence, interesting the put the x motor on the rear side of the table, I found it would crash into the column if I did that. I think I'm going to get hold of some belts and do some stretch measurements, I'll report back when I have some results...

[Brett248Vista]

Fixed my post.. geesh of all typos to make...

[HawkJET]

Using belt drive will certainly decrease positional repeatability due to the springy nature of the belt(s).

This is NOT accurate! If the belts are "springy" enough to "decrease positional repeatability", the system is not designed properly. A HUGE number of "real" CNC machines use belts. My Bridgeport BOSS uses belts on all axes. Accuracy is excellent. Besides, You also recommend using couplings on direct drive installations. A poorly designed coupling can introduce "springy" nature also. Belts give more flexibility to the design and allow motors to be tucked into places to reduce footprint. Plus, in applications using servo motors, belts are usually REQUIRED for speed/torque matching.

[Hamzter]

Long time no post Been too busy with the conversion to actually document the thing!

Well I received all the electronics and mechanicals about 3 weeks ago after some very lengthy shipment from the US (I'm in the UK), after checking out a number of other builds I came to the conclusion that direct drive would give me sufficient speed on the X and Y so this is a much more standard conversion now. Here's the final spec list I went with:

Screws - CNCFusion Standard Ballscrew Kit (#2) with preloaded ballnuts and the Y screw extended 1.75"
X Axis Motor - NEMA23 381ozin (Automation Technologies)
Y Axis Motor - NEMA23 381ozin (Automation Technologies)
Z Axis Motor - NEMA34 906ozin (Automation Technologies) - Detuned to approx 520ozin, running through a 24:15 pulley, equivalent of approx 832ozin
Motor Driver - Gecko G540 (Automation Technologies)
Power Supply - 48v 12.5A Switching PSU (Automation Technologies)
Couplings - Heli Cal Couplings (CNCFusion)
Limit Switches - Honeywell SZL-VL-A
Gas Strut (CNCFusion)

I've now got all 3 axes working, and set up limit switches on the X and Y axes...

Attachment 225634

And here's my first test part...

Attachment 225636

I've uploaded a video of that part being machined here (please ignore the bad workholding practice, I was impatient to get started )...

SX3 - First CNC Part - YouTube

After adjusting the the X and Y limit switches and soft limits, I've ended up with the following travels...

X - 391.5mm (standard 360mm)
Y - 187.0mm (standard 145mm)

I could probably squeeze a few more mm out of those, but it's nice to have a bit of a safety margin.

I've yet to finish aligning everything, tuning motors, grinding ways, and tightening everything up, so performance isn't perfect yet, but I've got it reliably running with 1200mmpm rapids and backlash of approximately 30um (approximately 0.001" for those that prefer old fashioned units ). I've still got a few things to do before I can call it complete, should have all this lot done by the end of next week...

• Fit Z axis limit switch
• Link the EStops together (currently I have one to stop the spindle, and another to stop the motors)
• Sort out all the wiring
• Finish replacing the incredibly noisy 60mm PSU fan with a silent 120mm one (nearly finished this but accidentally wired the fan backwards and fried it, doh!)
• Clean up the ways
• Shim the column, tighten everything up and calibrate backlash

More updates soon

[Hamzter]

Oh, and a couple more parts that I've CNC'd since, first up I modified the PSU cover to suit a 120mm fan...

Attachment 225640

Then I made this little guy, internet cookie for anyone that can identify it

Attachment 225644

Managed to break 3 cutters making that one! First one broke because I set the Z height incorrectly, 2nd one I missed an M0 code to allow me to do a toolchange, and 3rd one went because I forgot to retighten the gibs which overloaded the cutter as it changed direction (only a 1.0mm diameter). We learn from our mistakes though (I hope...).

I got a load of ball nose mills yesterday so I'll be trying some proper surfacing stuff tomorrow, watch this space!

[ilovestuffforre]

Companion cube!!! Macadamia please!

Excellent build BTW.