Kinematic systems and linear guides July 08, 2020 04:50PM |
Registered: 4 years ago Posts: 8 |
Re: Kinematic systems and linear guides July 08, 2020 07:13PM |
Registered: 4 years ago Posts: 93 |
Re: Kinematic systems and linear guides July 08, 2020 10:50PM |
Registered: 12 years ago Posts: 5,794 |
Re: Kinematic systems and linear guides July 09, 2020 06:17AM |
Registered: 13 years ago Posts: 1,465 |
Re: Kinematic systems and linear guides July 09, 2020 07:15AM |
Registered: 4 years ago Posts: 8 |
Quote
dekutree64
I do like the look of that crossed rail XY system. The moving bed is good for milling where you need high rigidity and movement is slow enough that momentum is negligible, but it amazes me 3D printers can yank it back and forth as well as they do when printing narrow zigzags. Also not good for tall skinny things.
What I don't like about the Ultimaker design is the cantilevered bed, especially if you're going to be printing large heavy things. I would prefer something like the Voron1 bed carriage, with 4 rods and two leadscrews, although it does increase the width quite a bit, and the cost for twice the components.
Re: Kinematic systems and linear guides July 09, 2020 07:54AM |
Registered: 4 years ago Posts: 8 |
Quote
the_digital_dentist
I have used linear guides in a bed flinger and a corexy machine. I would not build using any other type of guidance. Any time you use any sort of end supported guidance, either round rails or wheels in T or V-slot you have to worry about aligning the rails to each other in 3 dimensions. T and V slot rails are extruded, not milled and ground, so you never really know how straight or flat they are. Many machines use them and work fine, but the alignment problems - i.e. putting them parallel or square to each other is not easy. With linear guides, you can mount them on a piece of tooling plate and know that they will be on the same plane. Yes, to achieve all the manufacturer's specs on the guide you must mount it in a ground slot with a reference to butt it up against, and set another parallel by mounting a dial gauge on the bearing block, but the guides perform just fine without doing all that. Then the only alignment issue is to get them parallel in the plane on which they are mounted which is easy- bolt one down tight, mount a spacer on a bearing block and use that to set the other rail parallel to the first. In a corexy machine you set the rails square to each other before you install the belts. After you install the belts you make sure the rails are square again by manipulating the belt tension and you're done. For the X axis a single linear guide works fine so there are no concerns about getting two guides parallel in X.
I buy high quality surplus linear guides via ebay. It is not unusual to find NOS guides by NSK, THK, IKO, etc., for little more than the price of crappy HiWin knockoffs. Here's an example If you're designing and building one printer, you can use just about any size rail that's available for a good price. The 12 mm linear guides seem to be most common in the surplus market. The rails can be cut to required length using a cutoff wheel on a grinder. Some people have the idea that their design should be duplicated by anyone who can read their BOM posted in Thingiverse. They buy crappy HiWin knockoffs because other people will be able to get them easily/cheaply and the dimensions will be approximately the same no matter who actually makes the rail. Great.
This is the XY stage on my printer:
There is a 4040 t-slot subframe to which two pieces of cast tooling plate are bolted. The 8x24 mm Y axis guide rails go on the cast tooling plate, and the single 8 x 24 mm X axis guide rail mounts between them. This type of assembly can be dropped into a machine with any length Z axis. I used aluminum tubing to support the pulley axles at top and bottom to prevent them from flexing inward under belt tension. The assembly was designed to be used in a heated printer, so the X axis guide rail has an extra bearing block that allows the guide rail to slide a little as the frame expands and the Y axis rails move apart so it never binds. I have seen some forum posts from people who built corexy mechanisms using linear guides that bind when the temperature changes by 20C. I used large diameter skate wheel bearings for long life, quiet operation, and no print artifacts from belt teeth riding on smooth pulley surfaces. It's not the lowest possible moving mass mechanism, but it reliably produces high quality prints. Notice the minimal use of printed plastic parts (orange and yellow).
In my corexy printer, the only linear guides that bolt to t-slot are the Z axis rails. The t-slot is the type with a concave surface that is used to act as a locking mechanism for bolts that are mounting stuff on the t-slot. It does not provide a stable surface to mount a linear guide. For those I milled a shallow channel along the slots so that the guides would have a flat surface to sit on. It works well. You can get t-slot that has flat surfaces. They are extruded, not milled and ground, but if you think they're acceptable for wheeled carriages, they're fine for mounting linear guides.
Any printer worth the effort to design and build is worth the effort to enclose. If you build with wheeled carriages riding on the printer's frame, it becomes much more difficult to enclose in an esthetically acceptable way. Sure, you can just build a big insulated box over the whole thing, and that's fine for a machine that sits in a shop or garage, but a lot people like to keep the printer in their home or office where it will be seen by others, including wives. If you build using linear guides you can usually just bolt panels to the printer's frame which increases its rigidity and looks a lot better than a big, ugly box built around the printer.
Finally, if you price out the hardware you'll find that wheeled carriages aren't much cheaper than surplus linear guides.
Re: Kinematic systems and linear guides July 09, 2020 08:05AM |
Registered: 4 years ago Posts: 93 |
Ah, that is a good setup. Five components instead of Voron1's six, and eliminates the need to have the bed floating on springs and screws.Quote
pandaym
I also absolutely agree on the cantilevered Ultimaker bed - this part of their design they can keep to themselves. On my corexy I have 3 Z motors with lead screws on flexible couplers. This along with 2 linear rails feels like a solid Z setup, that I will probably bring over to my next printer.
Generally I am a fan of constructions that have the correct number of constrains - no more, no less, which is why the 3x leadscrew on flex couplers with 2x linear rails appeals to me.
Re: Kinematic systems and linear guides July 09, 2020 01:09PM |
Registered: 4 years ago Posts: 8 |
Quote
leadinglights
Hi Pandaym,
Having 3 axes working independently does intuitively seem the way to go. For myself I would rather build my next 3D printer from cast iron; filed, hand-scraped, and, if necessary, chewed into perfect flatness and squareness. The numbers however tell a different story: Many Delta printers made by happy amateurs achieve accuracy throughout the build volume not too far from the cast-iron machines of a few decades ago. This is done by having the power of the computer to correct the geometry of the construction and the flatness of the bed. The computer, or more accurately, the microcontroller, can also address many of the errors which are found in any kinematic mechanism. I must acknowledge that this is an unsatisfying opinion as I would rather be flint knapping than bit twiddling but I think my own rearguard action against the silicon overlords has been lost already.
My present printer slings the bed in both the X and Y axes on something like your crossed gantry. The reason for adopting this is that I rarely need a build volume grater than 200mm cube but do need lots of printer heads and other heavy bits on the printing end. Will I do it this way if I build another printer? No. I would build a CoreXY with automatic tool head changing. Maybe I would build a modified Delta with 3 hotends on a tiltable effector allowing each nozzle to be moved into the datum position with the others lifted clear of the print.
Or maybe I will adopt flint knapping as my next hobby
Mike
Re: Kinematic systems and linear guides July 09, 2020 01:22PM |
Registered: 4 years ago Posts: 8 |
Quote
dekutree64
Ah, that is a good setup. Five components instead of Voron1's six, and eliminates the need to have the bed floating on springs and screws.Quote
pandaym
I also absolutely agree on the cantilevered Ultimaker bed - this part of their design they can keep to themselves. On my corexy I have 3 Z motors with lead screws on flexible couplers. This along with 2 linear rails feels like a solid Z setup, that I will probably bring over to my next printer.
Generally I am a fan of constructions that have the correct number of constrains - no more, no less, which is why the 3x leadscrew on flex couplers with 2x linear rails appeals to me.
I've been playing around with a SCARA design, but now you're tempting me to build this type instead (or both) It would be fun to make a tiny one for printing high precision gears and such.
Re: Kinematic systems and linear guides July 09, 2020 01:55PM |
Registered: 13 years ago Posts: 1,465 |
Quote
pandaym
...........
Do you have any video of that two axis bed slinger? Would like to see such a machine working.
................................
Re: Kinematic systems and linear guides July 09, 2020 01:59PM |
Admin Registered: 17 years ago Posts: 13,976 |
Re: Kinematic systems and linear guides July 09, 2020 02:03PM |
Registered: 12 years ago Posts: 5,794 |
Re: Kinematic systems and linear guides July 09, 2020 03:05PM |
Registered: 4 years ago Posts: 8 |
Quote
leadinglights
Quote
pandaym
...........
Do you have any video of that two axis bed slinger? Would like to see such a machine working.
................................
The XY mechanism being used for some light milling in [www.youtube.com]
The reason why I didn't want to move the hotend mechanism in [www.youtube.com]
At the moment it is stripped down to replace the printed gears shown in the second video as they kept breaking.
Mike
Re: Kinematic systems and linear guides July 09, 2020 03:09PM |
Registered: 4 years ago Posts: 8 |
Quote
VDX
... better than a "normal" SCARA is a "parallel SCARA"! - much better accuracy/precision across the complete working range and some funny options too, if optimizing the kinematics (and dynamics)
[www.youtube.com]
[www.youtube.com]
[www.youtube.com]
... and a totally different "parallel" 6DOF-kinematic - [www.youtube.com]
Here I've started a thread about my own P-SCARA assembly ... it's not this fast (have a bunch of other projects running in parallel), but will do with time
[reprap.org]
Re: Kinematic systems and linear guides July 09, 2020 03:31PM |
Registered: 4 years ago Posts: 8 |
Quote
the_digital_dentist
There's small and there's small. I think a lot of people try to use too small hardware in a misguided attempt to minimize moving mass and end up with a wobbly mechanism. In my printer's X axis the guide rail is 24 mm wide. I set it on a granite surface and found it was flat, so didn't see any need to bolt it to anything else. The two bearing races are 24 mm apart, so they are essentially two parallel guides that are far more accurately parallel than any two separate pieces of metal could ever be, and they stay parallel no matter what the mechanism is doing and no matter how the temperature changes. There is zero play in the bearing block, and a 24 mm wide rail doesn't twist any meaningful amount. It works extermely well. It's not very light weight, but doesn't seem to limit printing speed. Extrusion is the big problem at high print speeds, not moving/controlling the mechanism.
I have always wondered about the effect of varying dynamic tension on the accuracy of prints in corexy systems. The static belt tension is manipulated to square the X and Y axes. When the machine is running, especially while accelerating and decelerating, the relative belt tensions will vary and that should theoretically cause the X and Y square relationship to be changing continuously. It doesn't seem to have any effect on print quality that I can see.
I'm interested in seeing what an actual mechanical engineer comes up with.
Slightly off topic, but maybe not, I recently installed servomotors in my sand table and have routinely run that corexy mechanism at 1500 mm/sec with acceleration at 20k mm/sec^2, and even those settings are not at the limits of what the motors can do. I'll be testing those motors in my printer soon. The extruder will have no hope of keeping up with that sort of speed/acceleration, but the servos run very smoothly compared to steppers, even with 256:1 ustepping, so I'm hoping for quieter operation and possibly higher print resolution. Higher acceleration can be good for corner sharpness in prints, though it increases the noise level and probably ringing in the prints, too.
Re: Kinematic systems and linear guides July 09, 2020 04:08PM |
Registered: 12 years ago Posts: 5,794 |
Re: Kinematic systems and linear guides July 09, 2020 05:03PM |
Admin Registered: 17 years ago Posts: 13,976 |
Re: Kinematic systems and linear guides July 10, 2020 12:02AM |
Registered: 4 years ago Posts: 93 |
That is brilliant! I probably would have used screws through oversize holes or slots, with teflon washers under the head and belleville washer under the nut, so it's snug but not too tight to slide under pressure. But your way looks simpler and even more reliable.Quote
pandaym
SCARA would indeed be an interesting choice! How far did you get along with that?
Constraining the bed like I described has a disadvantage when it comes to thermal expansion. Therefore I constrained the bed frame that way, and designed a mounting option for the tooling plate to ensure that no stress or warping is introduced as the bed heats up. I used the constrain pattern from a kelvin kinematic coupling for it.
Probably because most people who have time to work on stuff have no money and thus have to see what's cheap and figure out how to make the best of it, while people who have money to buy optimal components have no time to build the darn thingQuote
pandaym
I agree that people tend to use too small hardware. It seems lots of people choose their hardware dimensions backwards, instead of configuring it according to what they want to do with it. It seems you thoroughly evaluated your linear rail setup, and it sounds like it is more than good enough. I think a lot of that comes from the fact that you used high quality rails, instead of the cheap stuff that is flooding the market now.
Re: Kinematic systems and linear guides July 10, 2020 12:50AM |
Registered: 12 years ago Posts: 5,794 |
Quote
pandaym
The changing bed tensions in corexy is also part of my concern. This I am sure will have a much bigger effect on setups with more play, so I think your setup is simply solid enough to tolerate it?
Being a mechanical engineer of course helps with some things. I do have various experience that has some sort of reference to what 3D printers do, and of course a general understanding of kinematics. That said, I am absolutely amazed by what hobbyists design and build, some of which is more advanced than I have ever done really. The thing is that for work, I have only ever had to do the machine elements, and electrical/automation engineers will then take over to add stuff like motors, sensors etc. Therefore I don’t have all that much experience in building complete systems, which would definitely have helped in this case. What I do have a lot of experience with is design optimization. So I know a lot about what to do for robust design, design for manufacturing and design for service. Apart from that I am experienced in analyzing designs for constrains, to compare the intended scheme to the actual and optimize. I guess this means I am better at taking something good and making it really good, than I am at designing from scratch. Also it is worth mentioning that a lot of what I know and how I think is based on theory, and when it gets put into practice, it often changes nature somewhat. So even though your corexy setup might not be ideal according to an initial analysis of the constrains, it still might end up being the ideal design after iterating the analysis with respect to practical properties. So all in all, I wouldn’t say I am more qualified to do an ideal printer design than non-engineers - but I will probably have a different approach.
Re: Kinematic systems and linear guides July 10, 2020 05:25AM |
Registered: 13 years ago Posts: 1,465 |
Re: Kinematic systems and linear guides July 10, 2020 01:44PM |
Registered: 10 years ago Posts: 14,684 |
Re: Kinematic systems and linear guides September 11, 2020 01:49PM |
Registered: 4 years ago Posts: 2 |