Wide lowrider w/large Z

LowRider CNC Advice
1

Hello all,

Looking to build a highly customized lowrider and wanted to ask the experts a few questions. Basically an all new machine inspired by the lowrider 2. Maybe I’ll call it the high rider? :wink:

I’d like my CNC to handle a full 4x8 end-to-end (carving right to and over the edge). To allow for the full travel and some extra space I figure I need to make the X axis about 6ft wide. But I’ve seen recommendations of a 4 ft limit with the 1" OD 304 stainless and 0.065 walls. If I get thicker walls (o.083 or 0.095) will that sufficiently alleviate droop/flex/other concerns?

I’d also like to have a great deal of Z travel. It looks like the lowrider 2 has about 11 inches. I’d like to go crazy and support 4 ft, but still maintain high accuracy (I know crazy!). Let’s focus on physics instead of why though. With the low rider 2 the Z axis tubes move with the Z. As they move up the Z plane is farther from its support/pivot point leading to higher inaccuracy (wobble). So if you just made the tubes longer and moved further up your inaccuracy would sky rocket.

But what if you changed the design so the vertical Z rails didn’t move and were permanently fixed to the inside of the Y plates. Basically take the design of the X rails and tilt it up 90 degrees. The rails would now be supported by really tall Y plates at the top and bottom. The Z sliders would be trapped between them at all times and not moving away from their support/pivot. This should mostly eliminate wobble and inaccuracy, right?

With the now taller Y plates and the huge Z travel, would/should the roller blade wheels be placed further apart? This would probably be necessary to improve stability. How far apart would you recommend?

With the added weight of the thicker rails, more pipe for the taller Z and much taller & wider Y plates, I imagine it would add a great deal to the rolling weight in the Y axis. But as its only rolling weight would that be ok?

Looking forward to your feedback!

Thanks

Rob

2

Sorry it is just not the machine for you. There are practical limits for every machine.

 

What exactly do you plan on carving at 4’ high that a bar all the way across an axis will not ruin?

3

I don’t particularly plan on carving anything at 4’ high. I would probably never do it honestly. The low rider 2 looks awesome with the one limitation of not having a larger Z axis, so I was trying to noodle on a way to overcome that limitation. I’ve literally been up a couple nights just thinking on different ways to do it while trying to maintain accuracy. You’ve infected me Ryan! :wink:

The only real purposes I could think of are:

  • No need for adjustable height table if you want to work on thick items like foam (or whatever). Use the extra Z to handle it.
    • If Z accuracy can be maintained that would be a lot easier than re-leveling the table after adjusting the height.
  • Install a 3D printer head and you can print a huge (wide/tall) model.
    • One of the guys in this area is built this (https://www.thingiverse.com/thing:3324280).
    • It would be neat to have a printer that could do this in more dimensions.
  • The ability to add other tools on the bed without losing all the Z space. Like a DIY vacuum board for example. At 3-4 inches thick it would use up 40+ of your Z before adding anything on top to actually be worked on.
    • Though I will likely just build the vac board into the base table so it's a none issue.
So maybe 4' is extreme. But could we change the design of the low rider as I've described to gain more Z without losing accuracy?
4

Every axis you add after 2 gets exponentially less rigid and each of those is also less rigid linearly per length. That is why both my CNC’s have a low recommended Z axis.

 

Yes, a drop table is by far the best/least expensive way to handle it. Added with your only actual usable Z is dependent on your bit length, unless you are only ding some really 2 Dimensional carves.

The MPCNC geometry as a 3D printer has been pushed to 18" (or so) by dui. Same as as above any 2 axis can be giant, just not 3.

 

A drop table, you will always be limited more by the depth of your tools than anything else. Making a giant axis “just in case” is never a good idea. The other method is build two specific machines, one tall and one wide or whatever. Just because these machines can do it all does not mean they should do it all at once. They are inexpensive enough to have more than one. Maybe a good analogy is buying extra cab longed dualy just in case you need to buy a full sheet of plywood when 99.999% of the time you just sit in traffic and commute to work the better solution is a small car and a big truck (even though it initially costs more). Extra axis length is just like that.

 

5

Hey Ryan thanks for your help. I’m learning a lot. I hope you don’t mind if I continue to ask questions and pick your brain.

So, if every axis over 2 gets less rigid, how do other machines get around this? I’ve seen some machines that do x/z together and then have a separate platform underneath to do y.

Are these machines less prone to inaccuracy because they effectively behave like 2 separate machines that just happen to work together? (One ‘machine’ does the x/z (only 2 axis) and thus can remain rigid. The other ‘machine’ does only y (one plane) and can also remain rigid.)

Thanks again!

Rob

6

Is there any sort of test to determine the accuracy of a machine? Like repeated random movements and then checking that the return to a fixed point is within some acceptable range? Or moving among multiple fixed points repeatedly and verifying each time its within an acceptable range?

 

Thanks!

Rob

7

Not at all.

 

[quote=102762]how do other machines get around this?[/quote] Money. Giant bearings, solid cast steel, giant ways.

Machines that have separate axis tend to have concrete poured extremely thick then a very expensive technician comes out and calibrates it for the first use. Actually most bigger machines require you to have a technician set it up for the first shot. Hidden expense most people don't think about.

Any motion test serves a purpose, it really depends on what then end goal is. It gets extremely complicated the further away you get from the decimal point in accuracy. Temperature becomes a huge factor at some point as well.

This is all a very different playing field from any machine you can have in a home garage. Have a look at YouTube for highly accurate machining videos, they’ll probably blow your mind.

 

 

1 Like
8

I find it a fun thought experiment, even if it is impractical to build.

Your redesign addresses increasing stiffness at high z, but from what I gather, mainly in the y direction. Forces in the x direction (parallel to the horizontal tubes) can cause the machine to deflect into a parallelogram so you might need great big x/z gussets. And the gussets need to travel with z, so they need to be part of the z carriage that rides on your fixed z rails.

Just more fodder for you to think about.

1 Like
9

Jamie, yes of course exactly!

Basically a giant X is attached to the two vertical rails on the Y plate. The 4 left and right outside points of the X travel on the vertical rails and the center of the X supports the x axis rails. It’s something of a deformed X as the center would have to be wide enough to support and properly brace the x axis rails.

If the X brace were a foot tall on 5 foot vertical rails rails you would have 4 feet of z axis travel. Since the vertical rails are securely mounted and the X brace is large, parallelograming should be prevented. Ideally you maintain reasonable accuracy, because you maintain rigidity. (Or that was my thought).

10

If you’re going to all that trouble for something that might not work, why not just get a http://www.cncrouterparts.com/ machine? We know they work, Frank Howarth has one, and Brian Benchoff from hackaday. Also if you look at the construction of those machines, you can see how they are more rigid than the lowrider, but like Ryan said, money.

 

Actually just thought of something. I have a client that has a few cnc machines that have a little more than 6 feet of Z travel. You can literally fit a car in the build area, they make molds for car interiors. It cost several million per machine, and the floors under them are 6 foot thick reinforced concrete.

2 Likes
11

I’m still waiting on Ryan to macgyver the Mostly Printed Robot Arm

HTB1LAbgKpXXXXc8XXXXq6xXFXXX5.jpg1000×1000

3 Likes
12

Edited post and it seemed to disappear. Maybe it’s waiting for approval after the edit? Reposting so it doesn’t get lost. Please excuse the double post if that happens.

If I wanted a pre-packaged professional CNC I’d just buy one as you said. But that’s not my goal. My goals are to:

  • Have fun. Which includes:
    • figuring out how to make changes or try to improve on an already great design.
      • Ryan's done some awesome work and I'm aiming to help, not take away in any way.
    • designing and building accessories (laser, multi-pen potter with automated color change, vacuum table, lots more)
    • software improvements on the tool chain (TBD) but considering Marlin / Prusa firmware and Prusa slicer, etc.
      • I may back port some Prusa firmware changes to Marlin. I'm a software engineer by day.
    • which all requires a great community which Ryan has built here and I hope to contribute to.
  • Get the kids interested in fun things
    • STEM is all the rage you know
    • really these things are all just robots, so let's build a robot!
    • show the kids how hardware / software all interact
    • have some fun with the kids
  • Learn (that's happening right now ;) )
Even the CNC Router parts design for their pro level machine cautions against making the Z axis more than 18". So I've learned my lesson, too much Z -> bad. Thanks for the appropriate chastisement folks, you all rock.

Setting the large Z discussion aside, can we look at the CNC Router parts design and incorporate some of its features into a lower cost design. For example, they have 52" pro rack for $77.50 and the rollers for $50. The rack can be any length you want by placing them end to end. So 8’ of travel would be about $410 for the rack/pinion. (Additional cost of course for belts, etc). Is it worth the increased accuracy in the Y axis for that price? Perhaps not because the original price of a lowrider 2 is probably in the $500 range and I’ve just doubled the cost. But is it worth $200? Maybe ( Rack 2500mm/98" & 22T Pinion Gear Module 1.0 CNC KIT Router Plasma Laser Mill | eBay ).

Could we use the rack/pinion to improve more than just the Y axis? I’ve seen some discussion of the low-rider moving in the X axis by about a quarter inch if it makes many passes up/down the Y axis. Some have mentioned creating a grove for the roller blade wheels to ride in to resolve that. If we had the rack/pinion, mounted it sideways like they do on the CNC parts machine (teeth on the outside or inside, not on top), that would act as a guide to prevent X axis shift. Improving the design in 2 axis seems worthwhile for $200 to me.

Can we go even further? What if we added a second passive pinion about 12" away. Or we could even make it active by having a belt drive both from the stepper. Now we have 2 points of contact on the rack instead of one which I think would help prevent parallelograming in the X/Y axis, which I also know some have had an issue with from the forums. Is $200 worth improving in 3 ways? To me, yes.

With the less expensive rack/pinion we lose the gear ratio. The CNC Parts pinion had a 3.2:1 ratio. Obviously there is mechanical advantage (more force) to that, but is there an accuracy improvement because now each motor step is smaller linear motion? Probably (I now know you will all correct me if I’m wrong. Which I appreciate, thanks!). Is it worthwhile trying to find an inexpensive rack/pinion that provides this or not worth it? Are there any concerns of the Marlin firmware or slicer dealing with the altered travel per step? And where is that be controlled from, the marlin firmware or the slicer configuration when creating the G-code?

The kids and I will take that on next after we build this: ( https://inmoov.fr/build-yours/ ). Just waiting for the Prusa MK3S kit to arrive. No seriously, my youngest is excited to take this on.

Thanks for all your help and input folks!

13

I think we could get that down to just a few hundred thousand. Also, your job is awesome.

14

Somewhere someone made a lowrider-ish machine that ran on rails instead of the spoil board. They had something similar to the X carriage bearings running the length of the table. The issue with that is you can only go as far as you can get straight tube or conduit. I think the conduit comes in 10 foot sections, so at least you could make a machine that does a full sheet of plywood.

15

So I have this new thing I am working on it runs on supported rails (conduit or SS). The rails are kind of a pain in the but to get “perfect” but no hockey wheels. To get one performance level above where we are now…it takes a significant amount of work. We can squeeze out a bit more from what we have and maybe simplify the parts a bit but any large speed jumps are not looking very likely.

Is there any good software and use for an arm or are they still pretty much hand coded and hand taught?

16

If you use rails the length limitation would be due to the ‘bump’ where the pieces are mated or because you couldn’t ensure the two pieces are straight relative to each other?

Would using something like this help address the bump? https://www.amazon.com/gp/aw/d/B01M9G96S6 ideally the wheels are large enough that you no longer feel the bump.

17

I imagine the robot arm would need a controller specific to it, or close. Similar to Marlin, but it’s much mor complicated for arms. There are software packages for ROS to plan to a particular xyz, angles, and they need to have constraints defined. You wouldn’t want to carve through your work in order to get to the next position.

Then, there’s the trouble of creating the tool paths, which presumably has some software for all the arms that exist.

But I’m mostly just guessing here.