I’d think you’d need two spools with motors attached to each to manage the tension and at that point, you’re talking 8 motors… I don’t know, but it seems rather fishy I could see it working with chain though.
To quote myself:
I’m actually talking about only 4 total cable motors. Ideally the take-up spools could simply be spring tensioned, and one spool could take up for two motors. (take-up spool is the one with the arrow showing spring tension)
The take-up can be “dumb”, if using something like a capstan as the driver motors the take-up only needs to provide enough tension to produce enough friction for the capstan to function. Generally speaking capstans by their nature are usually designed so they can slip to allow a variable feed rate; for a Maslow a slipping driver motor obviously would be useless if counting rotations is how we are measuring distance.
However, I assume the cord being discussed here is braided, if we use cord with a single, high contrast, “flag” thread I wonder if the measurement could be done optically by watching the pattern of the cord as it is fed and taken up. Or, perhaps the cord can be measured with a pincher roller (like how they measure rope at the hardware store).
It might be better to let each component do its own job well instead of expecting one component to do several jobs. Instead of a driver motor that spools cord, measures distance, and drives let’s let the driver motors drive, let the measuring rollers measure, and let the take-up spool take-up.
Without having tested anything at all I would think that a capstan could be made to have no slip under a given tension (more wraps = more friction. Does it slip? Then add wraps.)
I can think of a few other methods of take-up but the one I just explained is probably the simplest.
I’ve never used a capstan so this is coming from a point of ignorance. It seems to me that when you are trying to feed cable only to motor A, you have to feed cable to motor B as well. If that’s the case, then wouldn’t there be zero tension on the cable to Motor B?
You’re talking about tension on the sled-side of the motors right?
In that case, yes, very good point. A single spring loaded take up is not realistic. Perhaps this is where my other take-up ideas come in… You could basically make a small block and tackle type take up that can be fed from either end. This is a couple more parts but they are still “dumb” and require no electricity. Originally I thought just a weight hanging on the loop but that would probably get in the way. A weight is nice because the force is linear, a spring is nice because it can be compact. Neither are perfect I suppose.
I should reiterate that this entire idea of mine might be a terrible one. I just wanted to get it out there to get more brains on it just in case it has any merit.
Or, you could still use the spring spool idea, you’d just have to have one for each motor. Still 4 motors, but instead of 2 total take-up spools you’d have to have 4. I think each would be responsible for around 13" of cord.
What’s what nice about the forum… we can trash each others terrible ideas I’d love it if we could solve the problem, I like the concept.
How would the cable wrap around the two spools?
But now that’s four cables and not two…
Perhaps something that can be considered is not to create a z-axis for a specific router, but rather a platform that different routers or tools can be mounted to (crudely shown below)
The top square is the c-beam or whatever and the ‘rectangle’ below it is a platform with shafts in each corner and the platform can raise/lower via linear motion bearings. I could see potential in this being used for something other than routing,(I’d love the ability to somehow fit in some sort of circular saw), but understand if focus needs to stay on one purpose.
This is my first post. I’ve read and read and read the forums every day for the past six months. What an amazing community! I ordered and received one of the last of the original kits in September (but I haven’t built my frame yet nor assembled the kit)…I’m still cleaning out my garage to make space for it.
As soon as I started seeing ideas around alternative approaches to the current design, I thought “what is already out there that can be used with, or modified for, a Maslow CNC setup? In other words, “don’t reinvent something that is already out there, or at least, let’s start with what others have already pioneered.” I started doing some research and found what is called “cable-driven parallel robotics” or CDPR.
On YouTube, search for “cable-driven parallel robot”. There are many, very impressive examples of what people have done with CDPRs.
There are three in-depth books related to CDPR from Springer (https://www.springer.com):
Cable-Driven Parallel Robots
Theory and Application
Author: Pott, Andreas
eBook: $139.00; ISBN13: 978-3-319-76138-1; 465 pages / 10 chapters; Pub: 2018
Design, Analysis and Control of Cable-Suspended Parallel Robots and Its Applications
Theory and Application
Authors: Zi, Bin, Qian, Sen
eBook: $129.00; ISBN13: 978-981-10-1753-7; 299 pages / 9 chapters; Pub: 2017
Cable-Driven Parallel Robots
Proceedings of the Third International Conference on Cable-Driven Parallel Robots
Editors: Gosselin, C., Cardou, P., Bruckmann, T., Pott, A. (Eds.)
eBook: $219.00; ISBN13: 978-3-319-61431-1; 416 pages / 34 chapters; Pub: 2018
Most of the current examples use a ‘payload’ hanging from cables within a large 3D space. My thought is…rather than having long suspended cables, why not do everything within 12 inches of the work surface. The cables could actually hold the sled firmly against the work surface.
There are also several other documents (e.g. 90 search results) on the Springer web site related to CDPR: https://link.springer.com/search?query=cdpr&facet-discipline="Engineering"
One document, which can be viewed online or downloaded (at no cost) is “ A Review on Cable-driven Parallel Robots ” https://link.springer.com/article/10.1186/s10033-018-0267-9. It’s a good introduction into CDPRs. Note the CDPR prototype in Figure 14. It looks like a Maslow CNC!
If we can glean from, and tap into, these resources, I think we can create a community-driven, open-source, cost-effective version that will amaze us. I’m not an engineer, so the mathematics behind CDPRs is way beyond my comprehension. But I can still dream, design, and hopefully contribute!
Very interesting concept, something I had never come across before. Here is a link to a short video demonstrating a CDPR with eight motors…obviously overkill and more complicated than required for future Maslow, but inspiring nonetheless.
My first post as well, though a long-time occasional lurker. Like many here, I’m inspired and invigorated by the great community and excellent collaborative problem solving that is in the water in these post threads!
I’m also invigorated and sent dreaming by @KenL’s discoveries of the CDPR community and knowledge base. There is without doubt something to be mined there that could be very powerful for Maslow—if not for the exciting next-generation version @bar is ginning up, then for some bleeding edge view of the future.
In that vein, I was captured by the paper Trajectory generation and tracking control of a multi-level hybrid support manipulator in FAST (https://www.researchgate.net/profile/Zhufeng_Shao/publication/259137294_Trajectory_generation_and_tracking_control_of_a_multilevel_hybrid_support_manipulator_in_FAST/links/59f08a09a6fdcc1dc7b52d53/Trajectory-generation-and-tracking-control-of-a-multi-level-hybrid-support-manipulator-in-FAST.pdf).
I stumbled on it while reading KenL’s linked CDPR review article — it describes a cable-driven robot proposed for use as the signal receiver on FAST, a giant Chinese mega-science telescope 1/4 mile (500m) in diameter. The paper describes a test they did with a scale model in which they derived positioning accuracies of up to 1.97mm across the scale model’s 40m range. In other words, their scale-model experiments using a CDPR achieved positioning error rates of 0.005%. Across the length of a 96-inch board, that would translate to accuracies of 0.005 inches. Putting that into perspective, that’s about the expected accuracy of my rail-mounted Shapeoko, 3x better accuracy than the original stated 1/64" goal for Maslow, maybe 2x better than the good accuracies some are tuning their machines to, and only 2x less accurate than a ShopBot that’s 40x the cost.
…the especially interesting thing about the CDPR in this telescope is that the “S” in FAST stands for “spherical.” In other words, the cable-driven robot isn’t moving on a horizontal plane, but rather along a spherical surface. As it does, it tilts to stay parallel to the surface. This is where we get bleeding edge, I know, but pause and recognize the ramifications for a second—the working CDPR they outline is, in effect, a 20-meter wide, 5-axis Maslow.
Certainly one for dreaming about (and worthy of a separate thread someday), but already practical at least in proof of concept: the model they describe could tilt 30º in any direction with a tilt accuracy of better than 0.3º. Translating that to router terms, a router mounted on the same cable robot for their 60-foot model could mill the top portion of a 5-inch sphere in place at tolerances of 0.004 inches.
Again, all wildly speculative and theoretical, with a zillion holes in the practical application (e.g., how would you weight things down to ensure enough normal force for cutting?) Still, it’s both fun and highly intriguing to think that while it’s already disrupting the regular CNC world some future version of Maslow might be able to disrupt the 5-axis CNC world as well!
A repost of the Mark Roper dartboard: https://www.youtube.com/watch?time_continue=2&v=MHTizZ_XcUM in case we are going back towards >2-cable designs
What if the motor was attached to a “capstan” but there’s another roller next to it applying pressure (maybe rubber?) Think that might apply enough friction?
I guess the bot is telling me I need to stop posting for a bit.
I love the idea if pulling from the world of cable driven robots. I wasn’t looking down that path and y’all are right. We can learn a lot from those resources.
What is the anticipated price for all these improvements? I see there are some 4x4 ft open builds type cnc machine kits on aliexpress for around $1000 using traditional aluminum extrusions and lead screws… One always has to keep in mind if it makes economical sense.
We can use something like that:
SPOIL CONSTANT FORCE SPRING.pdf (14.1 KB)
It’s possible to use the same motors like the maslow kit.
yes one can buy a 12’ tape measure and use the constant force spring inside of it to recoil the 9 feet of chain or rope. if you are using d#25 roller chain I think the spool would be over 6" in diameter. that is why people want to use rope because it is more compact.
While 2 to 4 take up spools might tape up a lot of room it is still doable with chain or rope, just neater with rope.
ok… it’s been three days since my last post.
Any idea on how to handle the motor controller? Obviously, there’s a couple ports lacking on our current motor shield designs. Is it feasible to just expand the ports on the motor shield or will a new microcontroller also be needed? I’ve talked to @blurfl about this and he suggested maybe slaving a second controller to the first controller (two megas connected via high speed serial interface).