Closed Loop Steppers - and 12mm lead screws

You are correct that new plates (plus possibly other modifications) would be required for 12mm lead screws. I'd leave that for a potential future upgrade and just go with the 8mm screws which will almost certainly work. They will be just as precise and accurate. The only difference is that the maximum speed of your machine will be lower.

The purpose of closed-loop systems is to provide more reliability, not to provide more accuracy. OTOH, buying from a random supplier on eBay and figuring out how to integrate the closed loop feedback into your electronics and controller will likely make the system hard to get going in the first place and reduce reliability. I'd recommend just buying steppers from the OBPS to start. They are pretty good about support and returns, so you will be able to get help if you have a problem with their hardware. In addition, you will have a working and relatively straightforward system. Later on, if you choose to upgrade, you will have a working system as your base and the experience of building your first CNC router under your belt.

Speaking for myself, when I built my first CNC router, I had no idea which upgrades would turn out to be essential to my workflow. It was only after I had a working system that I was able to figure out that I needed one thing or that another would be a waste of money for my workflow.

After building your machine, I think you will find that the most important factors for good accuracy are all about calibration and making things square. These tend to have much more impact than the particular drive mechanism or theoretical accuracy limits.

-D

 

I'd say before jumping on the latest and greatest motors, you need to take a step back and think through your spec requirements.

What woods are you machining? Any metals? How much power will the spindle be and what type? What level of precision and accuracy will you actually need for the materials and projects you're planning on doing? How much hand finishing are you anticipating, or will you be expecting ready-to-glue? Can you justify ballscrews or do you have to stick with leadscrews? Can your antibacklash nut cope with the forces you're applying? Is your backlash consistent enough across materials to compensate for? How much will the aluminum frame itself flex across the worst-case-scenario spans based on the forces of the machine?

A few years ago, I used to think I could potentially get a micron in steel from an aluminum extrusion machine and a decent router- the journey of knowledge from there is why I'm only just now preparing to build my first multi-axis machines. There's more to machine design than meets the eye, and a lot more than just adding feedback and hoping for magical accuracy gains. Just because the motor knows where the shaft is, doesn't mean the rest of the machine is where that implies!

Edit: There was a larger point that I'd intended to make that I drifted from a little. The ONLY purpose feedback serves is to let your controller know that the motor has moved to the place it told it to. That's it. Linear scales, if you have a controller that can cope with those, are more useful in terms of machine movement but again only tell you that the axis has moved, not that the tool head is exactly where it's supposed to be. The only advantage closed loop has over open loop is if you're not 100% sure that your drivetrain is capable of overpowering the material resistance. If you size your motors such that they will always maintain steps in the work materials used, and have a drive system that can't skip, feedback doesn't really tell you a whole lot.

Servos have the advantage of not cogging, and therefore don't transmit vibrations through the machine frame. A hybrid stepper is still a stepper, and so you're only deciding between feedback vs no feedback.

 

Sounds very similar to the machine I thought I was going to be building, before I got work that would require steel and aluminum capability. I had anticipated somewhere around a 1000x500x300 with a 4th axis, for wood, tooling foam, pre-poured urethane blanks, possibly the odd aluminum blade, etc.

The killer with aluminum is that while it's very rigid, it also transmits vibration extremely easily- and with a spindle that likely won't run out very much at all, that may be more noticeable in materials that can hold those tolerances. The woods you're using, you may be able to pretty much get away with a stock build and just tune the speeds and feeds- worst case scenario is likely only a slightly "smeary" look from the chatter, but that'd sand or scrape out. I was planning on adding a lot of gusseting, bolting the whole thing to a epoxy-leveled concrete base, and filling the extrusion with the finest sand I could find. Which, of course, wouldn't hurt on any other build. Massively overbuilding machine tools is always ideal- there's not really such a thing as "perfectly sized", it's more about what you can do in the limits of your budget, available space, and build capabilities. The more overbuilt it is, the better the finish you're gonna get, in most cases.

If you're not using square linear (THK/Hiwin) rail, there's probably going to be more flex than you were anticipating once you hang a big 80mm spindle off of the thing, particularly when it comes to the gantry and z-axis. Figuring out ways to eliminate flex there would go a long way, given the power your spindle is going to be capable of. The main thing is probably twist of the gantry itself, because as the tool plunges, or moves along the y axis under load, there's going to be a torque force about the x-axis due to the offset of the z axis. Minimising that offset is one option; making sure the z-axis isn't dangling out in free space and keeping it tight to the gantry, making the z travel as small as possible while still being useful, putting the gantry itself as physically low as possible to the table, getting the x-axis rails as far apart as possible, etc. Think of the tip of the toolbit as a lever with two horses harnessed to it, and you have to make sure they can't bend any of your machine whilst bolting in random directions.

Adding steel parts can help there, but only if you have the means to make them- steel is dirt cheap and allows for a lot of overbuilding (and iron has lower resonant properties than aluminum) but if you have to send out for laser/waterjet cutting, it gets just as expensive as alu. You can add a fair bit of weight to the gantry before NEMA 23s will run out of puff, you just have to be careful that it's the right kind of weight in the right places to be useful.

Making your gantry uprights ribbed torsion boxes instead of flat plate would help if you have the capacity to do so (which you could then fill with sand), joining them together underneath with more extrusion so that the gantry is a consistent travelling ring is another, and ensuring your y-axis rolling surfaces are as far apart as possible front-to-back would be an obvious third. Lots of methods of adding rigidity to the machine structure itself depending on how far you want to go. Once you've done that, then the differences from different propulsion systems upgrades should be noticable, but yeah, I'd go straight from basic steppers/leadscrews to DC servos/ballscrews in this case, purely for the speed and smoothness, since torque issues seem unlikely.

If you're stopping at 4 axes, a Gecko G540 seems like the obvious way to go, given the relatively light requirements for this machine.