[Rob81]

Hi folks! Dropping a post to see if I could get some advice on a design concept I'm trying to sanity check. I'm in the process of designing my 2nd woodworking CNC machine, which will be dedicated to 4th axis carving. The wrinkle is that I want to use a common x/y/z to carve multiple rotary A axis at the same time. Which got me thinking about the best solution for spindles motors, and concluding that if I can make it work, designing my own system would be the way to go.

Reason is I want to keep the weight and size of the Z gantry down, since that's the axis that really does all the work and I need it to be as snappy as possible. And since each spindle motor doesnt need much power behind it, as I'm carving relatively small diameter stuff, I figured a rack of DIY spindles would be best - for weight, space, and maybe cost. Add to this that I'd really like two spindles per A axis, with different size end mills on them. So this is my current design, exploded view second image. The plan at the moment is to have two of the below assemblies, with a 1.5 or 2.2kw spindle motor between them, then run gt2 belts from this powered spindle to the first pulley on each side, and then daisy chain the rest together.


Only other things not shown are an end plate on the bottom to hold the double bearings in place (they butt against a lip inside the housing on the other side), the belts, and a crinkle spring washer I'll add between the top belt pulley and the top bearing, just to add a bit of backup preload (hopefully unnecessary, as I'll use a clamp collar to compress the inner rings of the double bearings together).

Machine both housings from alu, run water over the top of the front face of the lower one, possibly compress a copper pipe behind to cool from the other side, and run water over the top face of the upper housing if needed also. I'm thinking deep groove rather than angular contact, shielded not sealed as shown. The cutting forces will not be large, so I dont think angular contact are necessary and deep groove are typically listed as having slightly lower friction coefficient (unless perhaps under high load). These are ID8 OD28 thickness 9mm bearings, although I am considering bigger might be better - I cannot find confirmation on this, but it seems logical that larger balls mean less frictional heating, and of course more rugged. I've found a relatively cheap and decent quality supplier for the ER11 chuck and shaft, tested one out and it looks good, they have options for 8, 10 of 12mm shaft diameters, so I can scale up a bit if needed. Probably have to get someone to machine the housings for me, maybe give it a go myself, but no big deal.

Sooo... thats the plan! Hoping for around 18K rpm. Am I crazy?! I consider myself pretty well clued up on CNC design now, but this is outside my experience, and I cannot see tips from anyone who has tried something like this before, all of which make me nervous . Trying to consider all the things that could go wrong.... first the force from all the DIY spindles on each side will be going through the final belt, connecting to the motor, although only two each side will be cutting at any one time. I've just pencilled the belts in as cheapo 6mm gt2, but there is space for bigger. And I might need to add idlers above the belts to keep them tight. Bearings will need good alignment, and the housing tightly toleranced . Thermal expansion could cause headaches, although if I keep it all well cooled, maybe not. I kinda feel like it cannot be this simple however, so if any of you have suggestions for what I might be missing, I'd really appreciate your thoughts!!! Any other input from "looks good" to "bad idea" would be welcome. Thanks for reading!

[kuzma]

How are you going to change rpm of a single spindle?
Long story short, it is cheaper, easier and more reliable to buy 8 spindles/vfds, an idea like yours will only get you in a botomless money pit

[Rob81]

Hey, thanks for the reply - I have no need to change RPM of any single spindle, because they will all be doing the same thing, control of the speed for the entire set is via the VFD and powered spindle. Are there any specific failure modes you have in mind as to why it would be a bad idea??

[skrubol]

I don't see any of the pictures you referenced in the first post.
So you're thinking 4 rotaries and 8 spindles so you can have 2 tools loaded?
A couple issues I can think of right off (without having seen your drawings.) Belt selection. Most charts top out around 10k RPM. It may be manufacturers just being conservative or figuring there's not much interest past 10k according to this thread:
https://www.cnczone.com/forums/linea...eed-belts.html
Keeping everything aligned will be painful. This is true with any multi-spindle system, but moreso if you need multiple sets of spindles (each set for 1 diameter tool,) and keeping a belt driven spindle setup aligned will be even harder.
You will definitely want idlers to take up slack, as you won't be able to move the driven pulleys to tension belts as they will only be moved to align the spindles.
Also remember that getting a job up and running will be more challenging. When you have all 4 parts loaded and machining, an issue on one part may scrap all 4.

Is your reasoning for doing this to reduce cycle time or just to reduce operator time? Having 1 or 2 1.5-2.2 kw spindles that you can run harder and still have 4 rotaries, won't increase operator time and may not slow your part time down that much and it will reduce maintenance and setup time significantly. Remember with multiple parallel spindles when you replace a tool, you will need to get the z offset of that tool matched perfectly to the rest.

[Rob81]

Thanks for the reply! Strange the pictures have disappeared, let me link them instead:
Spindle1
Spindle2
You have the idea though, 4 rotary axis, with 2 spindles per axis, although the A axis are not shown in these pics. My testing so far is showing me how much of a headache removing all the eccentricities can be! I checked the maximum surface speed for belts, gt2 are listed as around 30-50m/s, and I'm well within that, although as you say idlers will be necessary, as I can see how much looseness there is on the slack side, even when I make sure the belt is properly tensioned before I start it running. A bigger problem for gt2 belts is turning out to be finding a reliable source for accurate, well aligned drive pulleys - a lot of the cheap chinese alu ones are just a mess, visibly wobbling as they rotate, with grub screws on one side pushing them off the shaft etc. And gates themselves are not easy to source from, though maybe Prusa or someone keep them in stock. I'm also testing resin printed pulleys, which are accurate enough but the plastics are quite soft, so I'm not hopeful they'll last more than a day or so, and as you say reliability will be critical.

For why I want to do this, well I run very long carving programs that can easily go to 3-4h, which really limits my capacity. I could make multiple machines, but there are number of other complexities I'm dealing with, so it seemed to make more sense to have multiple spindles being carved at the same time on one machine. And then I thought, since I dont need much power on each, and will only infrequently change the end mills, maybe I could make my own spindles. Multiple 1.5kw or even smaller spindles is an alternative for sure, but the most important thing for me is the ability to move the z axis as quickly, and as accurately as possible, which means reducing weight wherever I can. Plus it saves on cost, as it means one powered spindle, rather than 8+! But yes, it does increase complexity for sure. I dont "think" Z offset will be too much of an issue, if I have the end mill loose-ish against a touch plate and then tighten the nut, my rough tests so far tell me the tip barely moves, but if it does I'll have to separate each spindle assy out so they can be individually zero'd.

[skrubol]

Another thing to keep in mind is how much work will it be to replace a belt. You could deal with it with creatively routed belts and located idlers, but if you're planning on just having a belt go between each adjacent spindle, you'll have to take the whole assembly apart to replace a belt.

[Rob81]

I admit hadnt put much thought into belt failure, but yes, I'd have to take out both connected spindle assemblies if that happens. Not too big a deal, but worth designing in to make it as easy as possible, even if only to do as maintenance. They should be rated for thousands of hours, but likely less as these speeds.