Separate names with a comma.
Some features disabled for guests. Register Today.
Discussion in 'CNC Mills/Routers' started by beazleybub, Apr 5, 2018.
Null & Void
Looks intriguing. I'm a fan of gear reductions on axes! Do you have a plan for the type of belt you're using to deal with all that power? Seems like an upgrade from GT belt might be in order, maybe XL belt or a metric equivalent might be the way to go.
Only general suggestion I'd make is to make sure you have enough adjustability on the motor mount to be able to tension the closed belt properly and get it on and off the pulleys effectively- mine had to be perfectly sized thanks to the overall mounting geometry, which resulted in the belt being a little tight for my liking- not really enough to warrant changing it, but it'd be nice to be able to adjust it some. I'm assuming it'll stretch out to size in time, though (another thing adjustability would be handy for).
I see what Rob and you are saying about the belt adjustment, however, I'd really like to see just a little bit more meat at the bottom, just below the slots.
Would it hurt to have, say, 10mm extra between the mounting holes and under the slots? It would add a little more strength to the stress points, and allow for a few more millimeters to be added to the slots if needed.
Just a suggestion and I'll be real interested in seeing the finished project.
What belt do you plan to use for the motion? The resolution increase is great, but I'm sure even the stepper by itself can shred belts. 1275oz of torque isn't going to be usable at all. Instead of 425oz steppers, you can drop it down to 269oz and still have more than enough.
If you plan on using wheels, the 2.2kW is way more than the system will be able to handle.
Overbuilding is nice, but useless and costly if you can't actually use it.
Seems inconvenient for quickly replacing a broken belt. I'd probably try and engineer a way around that. I know my lathe belts like to snap at the most inconvenient time, and if I had to disassemble half the head casting every time one of the belts broke, it'd be incredibly annoying. Imagine having to take apart a belt sander/grinder every time a belt broke! Belts are consumables, is basically the point I'm obnoxiously belabouring here. And yeah, regardless of material, you're going to have stretching. Even high tensile steel cable needs retensioning occasionally.
I'd guess 9mm GT3 belt could handle 1kW, I know the 170XL (~12.3mm W, 2.5mm pitch) on my lathe handles 1HP perfectly comfortably, but when you have monster steppers on one side, a substantial torque multiplication in the middle, and a very heavy spindle and gantry on the other side- and that's before even getting to 2.2kW of cutting force (admittedly torque may be relatively quite low at the higher speeds, but it's not a trim router!) trying to pull things around- I doubt any GT size belt is going to handle it. I'd look through the Gates catalogue at kevlar and urethane industrial and automotive grade timing belts.
The resolution is, I suspect, not going to be what you'd really like. With such high forces, a theoretical resolution of 0.01mm (roughly, I'd guess, with your setup and a touch of microstep) isn't going to happen with plastic belts. While that may be your repeatability at low speeds spindle-off, under acceleration with the spindle cutting heavily, depending on your machine overall dimensions, you might see real-world repeatability somewhere in the +/- 0.5-1mm range. There's a reason they don't really make anything starting with "micro-" with a belt. You really need screw drives. Specifically ballscrew, if you're going for real repeatability under real load- anti-backlash screws are for 3D printers and cutting MDF. You could even add gear reduction to that- and in fact, when using servo drive, that's actually best practice to avoid instability due to inertia as well as other concerns like cost and power efficiency.
This is a good call, I should've caught that! It looks like 3/16"-ish steel, and if it's chromoly it might just about cope, but if it's a mild steel or even intended for easy manufacture in aluminum, for sure, there's got to be a ton more material there. I doubt this could even be made from aluminum anyway with these forces on it- the mounting holes and slots would become deformed over time, you'd have to use inserts regardless. Titanium could technically be an option there, but it'd be super expensive.
Yep, all of this. Realistically this system has already been designed... It's called ballscrews and linear rail! (Which, now they're used for DIY more frequently, are relatively cheap even compared to aluminum extrusion and belt systems).
I can tell you from now that wheels can barely handle an 800W setup. any 20mm profile rail is going to be way stronger than any amount of wheels. After a certain point, you'll just be spending more on wheels than on rails.
Just like belt stretch not being a fault of belts, whip isn't really a fault of a screws. Stretch/whip is a machine design fault. You can't build a skyscraper out of paper, nor can you make a book with steel pages. Everything has it's place and using them appropriately is what defines a design as good. A buddy of mine has a 4x8 and a 5x12, both Thermwoods. The smaller uses something around 30mm ballscrews. The larger isn't much bigger.
As for the 2.2kW setup, I don't recall you mentioning a build size. If you plan on the typical 1m x 1m build, that spindle is going to cause the gantry to sag. Having excess power/torque isn't helpful if the machine can't hold it's shape. The two reasons why I keep stating that the linear rail and v-wheels can't handle the power is because of sag in the rail and load bearing on the wheels.
Agree for sure on the plate thickness part.
Have you considered a larger belt? I have my own design based on interlocking 15mm HTD belts that I hope to one day build. It's meant for a 1m x 1.5m build, but could easily go much larger. The compromise is that I woul never expect it to efficiently cut aluminum as it would be built with wheels. A much smaller machine is easier to build rigid than a larger one.
Woah, slow down there. I was only using the ball screw as a reference to a designed machine. That machine pushes a heck of a lot more weight around than our hobby machines. At no point did I say that's what you need to do.
I never made any assumptions. I'm simply voicing my opinions on particular aspects that have been done a million times. Many of which have ran into the exact issues I pointed out. You haven't shared a single thing about your design so it's hard to make a decent educated guess. Don't be mad when you give half of an equation and get half of an equation worth of a response. Now if you said, "I plan to use an i-beam setup based on two 2060 and one 2080," then others would have a better idea about the structure you plan on building off of. At this point, all you've said is that it's called the Brute. Some people call doubled up 2080 rail the ultimate machine, so you never know.
I also never said v-slot is weak. Everything has it's place. Most people here work well within the limitations on v-slot. Those who need more often times go with 8020 extrusion (or similar) for much larger sizes. Some even go on to build steel rigs. Everyone using v-slot builds within a certain set of parameters and also only demands so much.
I'm not a pro either and never claimed to be, but at this point, no one knows where their answers to these questions are going. Heck, we could all be batting strikes based off of the information given. Instead of ranting, challenge what's being said with some concrete concepts. We're all here to learn from each other.
Sounds quite inconvenient to me, but I suppose that depends on whether you're looking at productivity or more along the lines of capability at the hobby level. I'd like nothing to remove in order to put the belt on, and only a single screw to set/lock the tensioning level. But that's just me.
Whip is unfortunate, but could be damped with an appropriately designed dedicated damping system. Otherwise, you're looking to choose between whip and stretch, and either way you aren't going to get any of the performance targets you want. There is, in fact, a product designed specifically for what you're trying to do, relatively recent to the market- within the last decade or so, I think. It's called Roller Pinion System (at least, that's one company's trade name). I was looking into it when I was trying to get beyond the 40" mark. It's basically infinite length capability, no stretch, no whip, no backlash, high precision. I'm quite sure it's also incredibly expensive and only available through select industrial procurement processes, but might be worth making some phone calls about. Definitely something I want to play with one day.
The other thing you should look into is rotating nut systems. They're great for long runs because you're not turning the screw, and therefore there's no whip, which means you can also use a narrower screw based only on how much preload you need to apply. You can keep the precision of ballscrews and not worry about the torque forces because they're mostly hitting the screw, not the belt. Since you're already gearing the motor down, you're set up to do it. I just did it on my laser z-axis, but it's not a new idea:
(Obviously there's no antibacklash mechanism there because gravity is doing it for me)
Alternatively, and possibly better for disassembly, I'd look at dual rack and pinion systems, with the two racks and pinions per side rigidly preloaded against each other. I think that's going to be the only realistic solution to the levels of accuracy you want at the sheer size you want. Belts you might be able to engineer and tweak up to about 4ft, but I suspect even that's pushing it whilst aiming for decent accuracy- or rather, repeatability. Since that's the harder part here.
1/4" steel plates would probably be fine, but aluminum is mushy and I suspect it might not have the longevity at the torque levels you're looking at. The motor and carriage are gonna be trying to twist them off the frame. The meat around the holes obviously needs significant upgrades in any case, but you can throw the model into Fusion 360 and spit out FEA numbers for that. Whilst aluminum is rigid, it's not necessarily enormously strong, and we tend to conflate the two.
You're also seemingly trying to break the cost/speed/accuracy/weight/precision/power multi-dimensional triangles, though. We're naturally going to make follow-on assumptions about the machine's design based on what we know about basic machine design precepts, y'know? If you don't share, people assume. Just human nature. And techie nature to try to break down designs!
If you're willing to throw out cost, you might get somewhere. But you'd be hard-pressed to find a market who's going to say "I need to cut plywood to the micron level for the price of a used Mori!" And since we're talking about gear reduction drives, the end use is relevant, which is why we're talking about your machine. Personally, I just drive screws with them- I don't even begin to consider a belt a viable drive system for anything above a couple pounds over a few feet. A full plywood size machine I'd drive with chain, possibly, or double-racks as I mentioned above. It doesn't need to be accurate and repeatable beyond 0.1mm, since anything coming off of it is going to get sanded and finish work done to it anyway, and 0.1mm is fairly trivial to get through conventional means.
The 2000lb isn't for rigidity. You can get rigidity with a properly designed weldment or aluminum spaceframe. The 2000lb of cast iron is for damping. Surface finish is what's going to suffer in a lightweight machine. Aluminum is notoriously effective at vibration transmission. Of course, again, a 4x8 machine isn't for metals where surface finish really enormously matters, so you might still be fine anyway. If you think you're gonna be flycutting aluminum signage or something, I have bad news for you, but I'm assuming you don't.
Why do people use aluminum extrusion for machine building? It's a low technical barrier to entry, low shipping cost, high rigidity system for most general desktop-scale types of projects. There's nothing enormously special about it, but it's a great beginner, general purpose, low-spec and quick-prototype machine building material. It brings machine building to people who aren't going to read journal articles on the packing equations of fly ash particles in epoxy.
Four thou seems like kind of a lot to me, but of course it's only 0.1mm, perfectly adequate for a woodworking machine. Do you know how much downforce that spindle will generate using various standard/upcutting end mills on different materials at various speeds and feeds? Also consider vibration and resonance.