Scraping a Bed of a lathe

[Richard King 2]

I was doing some research for a question on Warner $ Swasey Turret lathes and found this on Vintage Machinery. It's a bit boring, but it teaches how to scrape a lathe.
I also like how the guy talks about 3 points at the end of the article. I say using 3 points is also a lost art as scraping used to be.

http://vintagemachinery.org/pubs/2261/25873.pdf

 

[Rex Walters]

Nice find, Richard!

I also like how the guy talks about 3 points at the end of the article

Pages 8 and 9, and figure 7 for those reading along.

I've been goofing around by building 3D printers recently, and it's astonishing how few of the twenty to thirty-somethings involved in the hobby understand the rather basic idea behind 3-point supports.

There are lots of designs for 3D printer kinematics, but pretty much all require that the toolhead moves in a plane that's parallel to a flat build surface ("the bed"). You're unlikely to ever hear it described that way, though.

All the online 3D printing guides talk about "leveling" the bed even though nobody uses a level to establish a reference plane (like you do with a lathe). What they are really doing is tramming, of course. Worse, most designs capture and support the bed on all four corners, so an incredible number of people end up warping/bending their beds out of flat as they try to "level" it.

Higher-end 3D printers use cast and ground aluminum beds supported on three points, so you can be pretty sure the bed is flat. They also have probes and software to automate the process of "leveling" (tramming) the bed and even allow you to compensate in software for non-flat build surfaces (creating a "bed mesh" of Z-height offsets to apply automatically). Even so, you'll invariably get better results with two flat planes parallel to one another and thus no compensation. I constantly see people complaining about a warped bed because the 3D graph of the probed surface doesn't look flat.

You can almost see the lightbulbs appear above their heads when I point out that the probe and toolhead might not be moving in a flat plane, so what they are seeing might have nothing to do with the flatness of the bed. Ensuring the toolhead only moves in a flat plane is a much more complex task than just tramming a flat bed. The motion systems on higher-end printers often use multiple linear bearing rails: all of those must also be co-planar or perfectly orthogonal to each other (which is non-trivial to accomplish). Almost everyone seems to believe that bolted-together saw-cut aluminum extrusions automatically ensure this!

Anyway, Rich, the stuff you taught me is applicable to all sorts of stuff!

 

[Weldingrod1]

On my printer I went to a 3/8" Aluminum plate for the bed that was machinist level flat. That clarified the fact that my motion system was flexing maybe +-0.1mm over its span. My "levelling" produces compensating kinematics that.cause the head to move in the same plane as the bed.

On the 3 points and exact constraint front, you might want to look up my 3d printer on thingiverse; lots of effort towards only constraining what's needed. Little to no tweaking to get get it moving smoothly, unlike most muli-post designs.

https://www.thingiverse.com/thing:981664

Sent from my SM-G892A using Tapatalk

 

[Ianagos]

Been wanting to build a decent 3d printer and it’s very obvious how most people building these printers don’t really understand a lot of these concepts.

I’m debating if it’s worth the effort to try to design and build something a bit more rigid.

Most of these printers also have pretty slow rapids which probably contributes greatly to the long build times.

 

[Rex Walters]

I’m debating if it’s worth the effort to try to design and build something a bit more rigid.

I know that as a machinist, I too was unimpressed with the design and rigidity of the 3D printers I saw. But mass/inertia, stability through thermal cycles, and heating time are usually more of a concern than rigidity (especially when printing ABS or other plastics that need higher chamber temperatures). I don't expect to see much cast iron on 3D printers any time soon. The tolerances required for extruding molten plastic just aren't as severe as those for machining with rotating cutters.

IMO, there are several good designs out there that are sufficiently rigid for quality parts. "Core-XY" designs have proven to be pretty quick and reliable for modest-sized parts, but splooging about thin layers of molten plastic ("glue guns on a gantry") will never be fast enough for anything other than one-off prototypes. The crossover to where injection molding or subtractive machining makes more sense comes pretty quickly.

That said, I have cut my print times to just a fraction of what they used to be. Most real-world structural parts still take me an hour or three to print, but 7,000 mm/s^3 acceleration and 200 mm/s travel speeds are easily achieved on a smaller Core-XY printer and still produce excellent quality parts — much faster than I could even think about on my first printer.

For what it's worth, building an enclosed-chamber printer so I can print ABS plastic easily, and learning to design with heat-set inserts in mind has made 3DP much, much more useful in my shop. PLA is just too flimsy for shop use in my opinion. Bolted-together assemblies allow me to overcome most layer separation weaknesses (as well as eliminating the need for supports).

I was shocked the first time I held a complex machine built almost entirely out of printed ABS parts. 3D printing is definitely not just for cheap plastic trinkets anymore.