CNC 4th Axis ideas needed

[mc]

I'm looking for a 4th axis for my Denford Triac mill, and am considering building one, as the available options don't really tick the boxes I'd like.

6" is the largest size I think that will still be reasonably useable. 8" would work at a push, but it's starting to get into tool clearance issues on anything more than a couple inch diameter. I'd like something with as little backlash as possible, and with a centre hole. I've got some parts I'd like to make that would work well being able to feed a bit bar through the chuck, and using the 4th axis as more of positioner than a true 4th axis, but there are also a few parts where using it as a full 4th axis would be needed.

The options I've looked at are the Skyfire 6" which uses a harmonic drive (near zero backlash), but it's got no centre hole, and I'd really like the ability to feed a bit bar through the chuck.
Other option would be converting a typical 6" rotab, but worm and wheel doesn't reliably give low backlash, and I want something that isn't going to need continual fettling to keep running well.

In my mind (I've not got as far as doing any sketches yet!), I'm thinking a hollow spindle with suitable chuck mount, which runs in suitable bearings, then a harmonic coupling for a compact gear reduction, driven by either a stepper or small servo via a toothed belt.
However hollow harmonic couplings seem to be very rare, and from what I've found so far, the biggest throughole is 40mm and likely comes with an eyewatering price tag.
So I'm looking for possible suggestions, on how to get minimal backlash with a reasonable sized through hole (I'm thinking a minimum of 2", but nearer 3" would be better).

[awemawson]

I put one on my Triac more years ago than I care to remember. The original controller had died so I installed a PC running an early windows, probably XP, on which I ran TurboCNC which let me write the necessary code for the fourth axis as well as the tool turret

I went the conventional way standing a rotary table on its edge driven by a stepper

(Mach 3 wasn't running then!)

[mc]

Mach 3 was well established when I converted my first machine (a Conect lathe), although I've since ditched it in favour of Dynomotion's KMotionCNC. I got fed up with the random quirks which often resulted in crashes.


Anyway, I've just done a bit searching on Alibaba, and it looks like the biggest through bore on a harmonic drive is 36mm, and pricing seems to be around $5-600. That would be useable, but would limit what material I could fit through it.

It has got me thinking about what precision worms and wheels are available. I know the generic rotabs sold by the usual suspects use a basic helix machined into the cast iron (I've got a 6" one buried in a corner somewhere with a stripped wheel). A quick scan of HPC gears puts a 50 tooth MOD2 wheel at £142 plus the vodka and tonic, and a corresponding worm at £40-85 depending on material/type.

[awemawson]

For my Beaver Partsmaster 4th Axis I got hold of a Nikken positioner and repurposed it. That has a reasonable bore up the spout and anti backlash gearing and a brake


 https://madmodder.net/index.php/topic,9439.0.html

[djc]

A few thoughts...

You say 6 to 8 inches maximum. You talk of using a chuck to hold parts. You mention that the largest harmonic drive you can find has a through hole of 40mm. I think it might be difficult to find a chuck in the 6-8" range that has a through hole of the 2-3" you want.

Worm and wheel would be difficult to use as a fourth axis unless you want to spin it very slowly.

I would try to find a lathe headstock with suitable spindle nose and bore and use a good servo motor coupled to it with toothed belt. I have one off a Denford Easyturn I that I intend to use for this. Taper roller bearings, nice square casting, D1-3 nose (which matches everything else I have). I have a little experience of the Mitsubishi MR J2S units and they are very versatile. It is easy to switch between positioning and velocity mode. Other servos should do the same thing.

You could use a bicycle brake disk on the spindle itself to really lock it down tight. Cable or hydraulic operated via a servo, with position feedback to the control so it knows brake is off before trying to move and knows it is on before making any cutting moves.

It might even be possible to put an encoder on the spindle and feed it into either the servo (I think the MR J2S units can use an external encoder) or KFlop to deal with the small backlash. The secret with toothed belts is rigid setup and proper tensioning. 

[WeldingRod]

I needed one for a work project, and bought a not-working HAAS 5c.  The slo syn was full of coolant glop, but a modern stepper fitted right in.  I ran it for a few hours with spindle oil drizzled into the worm gear, then greased it repeatedly while running it.  It runs great, BUT super slow.  Don't expect real feed rates from a good worm!

Sent from my SAMSUNG-SM-G891A using Tapatalk

[mc]

djc, those are all valid comments, which have been bouncing around my head while thinking about the pro/cons of various options.

Chucks I'll admit I'm not entirely sure about. I'm actually thinking of something interchangeable, as I'd like to feed square/rectangular bar through the chuck, which would likely involve some kind of custom adaptable setup to allow for good flexibility, but then being able to fit a normal 3/4jaw chuck would be good for handling round bar. Some kind of D type chuck mount is a real possibility, although t-slots with a locating boss or two might be easier.

Running some figures from the top of my head, if I take a 50 tooth wheel with a single start worm, that's a 50:1 ratio. Couple it to a smallish servo (I'm thinking 2-400W) that should manage 3000RPM, that's 50RPS per second at the motor, and with the 50:1 ratio, gives 1 full turn of the 4th axis per second. Not exactly fast, but I would of thought perfectly adequate for occasional 4th axis machining.

If I'm being realistic, the majority of stuff I'll be doing will be more positioning than true 4th axis work, so I suspect a brake will end up part of the design just to eliminate backlash (I'm more concerned about chatter, than tolerances in that regard). And feeding an encoder to the KFlop to handle backlash is a possibility, but obvisouly minimising backlash to begin with is the better option.