Producing precision machines from imprecise beginnings

In the " self replicating machine" post, the author comments on the possibility of losing precision with subsequent generations.

But this gets at the more fundamental question of how precision is produced in the first place - how did we get from stone axes to precision lathes? So I thought I'd start another thread about this. A refresher might be in order for the difference between precision and accuracy, so [en.wikipedia.org] is a good place to start. Basically we create precise machines and then calibrate them to get accuracy. Calibration is a process you can look stuff up about easily (try "metrology"). So let's focus on precision.

I have not been able to find any info really on a high level view of how creating precision goes, though. Anyone have any info on this?

Obviously there are some instances where you can create a very flat surface from imprecise equipment, like melting tin and putting glass on top to produce float glass (or just let the tin freeze again). I bet you could create an object with a lathe that is more round than any of the parts of the lathe.

You can depend on the averaging out of errors - two flat plates that are not very flat can be rubbed together to make two flatter plates.... Hm. What else?

... a pantograph 'scales' mechanical movement up or down and decrease or increase accuracy - so with say a 10:1 downscaling will rise the resolution on the tip by x10! ... but if you get the same accuracy is depending on the stiffness of the frame and used joints ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

Hello,

Producing an accurate part, on a badly worn lathe, is quite possible and is done all the time. The key is that the error is repeatable. If you know that the lathe bed is worn such the lathe will produce a part with a taper of 0.005 over 6 inches, for example, you compensate for it by working small sections of the part at a time. In this example, you would cut the part down to slightly oversize and then finish an inch or so at a time. Then, you can produce a part that is under 0.001 from the ideal measurement. By polishing the part, on a lathe, you could get it to within a couple ten thousandths of the ideal dimension.

As far as producing a flat surface from nothing? You take 3 close-to-flat surfaces and use them to compare against each other. Then they can be hand-scraped down to as flat as you have the patience for. You would use marking fluid to find the high spots. Three surfaces are needed as if you scrape two surfaces together, one could become convex and the other concave.

A master square is nothing more than a cylinder on a flat surface. You make a very precise cylinder, on a lathe, and place it upright on a flat surface in order to form a 90 degree angle.

Regards,

Brian

Edited 1 time(s). Last edit at 08/03/2011 02:51PM by reifsnyderb.

GreenAtol - You're right; the fact that today's precision equipment was built, generation upon generation, from less precise machines and hand tools is the best argument against critics who claim that self-replicating machinery will lose precision with each generation.

Alex Slocum's lectures are good reading in general, and the third unit covers a couple of those methods.

One of the ways that precision machines can be made from hand tools is by relying on geometry instead of measurement. For example you can use a compass and straightedge to generate geometrically perfect squares, triangles, etc, without any kind of ruler. Or a straight-tooth hob in a metal shaping machine can create perfect involute gear tooth profiles, without computer numerical control. Division of a circle is another good example. I suppose the main idea for creating precise shapes with crude tools is to aim for a process that will naturally result in the shape you want based on the rules of pure geometry.

As mentioned above, too, averaging and mechanical reduction are both ways of creating a part that is more precise than the parts of the machine that builds it.

Quote
Terry Pratchett, The Truth
The only tools a dwarf needed were his ax and some means of making fire. That'd eventually get him a forge, and with that he could make simple tools, and with those he could make complex tools, and with complex tools a dwarf could more or less make anything.

---

Bob Morrison
Wörth am Rhein, Germany
"Luke, use the source!"
BLOG - PHOTOS - Thingiverse

All this is exceedingly interesting. I'll keep looking for some books and other material - it would be really nice to have an eagle's eye view of these types of techniques, organized and with all the different fundamental principles that are depended on, and that sort of stuff. I bet there is some sub-field specifically dedicated to this sort of thing, just a matter of finding it.

... it's sometimes the right idea, but mostly it's the right tool.

Last years i was busy with some 'low-cost-nanotech' and developing tools and complete workstations for microassembly.

For a special project i've built some XYZ-stages out of parts milled with a cheap CNC-mill and modified with manual tools.

The accuracy of the CNC-mill was around 25microns, the manual accuracy maybe 100 microns ... but the positioning accuracy of the stages was something around 10 nanometers!

So with the right concept you can achieve much more precision and accuracy than your tools have ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

As Jbayless mentioned, averaging is a good tool.

So for example, lead screws for lathes could be made on lathes using the same type of lead screw. By clamping the carriage to the lead screw using a simple box filled with wadding the new lead screw would be cut to an average of the previous one. You could then put the new lead screw into that same lathe and make an even more accurate one...

Obviously it is possible to make more accurate machines from less accurate ones or we would never have got here...

George

when i started i had a similar thought, how can you get precision from these printed gears, and the answer is, the stepper motors quantize everything, and then everything can be geared down or increased in accuracy. how was the first straight line made? someone took a vine and stretched it taught, then how did they get a length that is half, they folded the vine in half. now one thing that keeps our machines from degrading is the smooth rods, or rails, we usually have these machined or industrially produced. I haven't seen a 3d printer that uses printed rails yet, although its comepletely feaseable.

also have you seen the before and after pictures for the z couplings? have you noticed that the z couplings are always very wavy? then the prints after the upgrade they are not wavy at all, this is because the frame part quality does not have an effect on the parts produced unless they have something to do with the linear motion, like rails or gears.

Edited 1 time(s). Last edit at 10/24/2013 06:56PM by aduy.

Anyone intersted in this sort of thing will most likely enjoy "Build your own metal working shop from scrap" by Dave Gingery. I haven't got a copy myself yet (I'm not sure why), but it's on the shopping list. As soon as I get enough dough to by a house, the backyard will be put to good use . I think backyard foundry sites actually led me onto 3D printers from the start.

(No affiliation BTW, it just seems like an awesome series of books)
/Andreas

Ahh, the Gingery lathe..

---

_______________________________________
Waitaki 3D Printer