If you want to get .0005" absolute position accuracy (or better) (forget backlash at the moment and load) across 20 inches. Do they make screws that accurate or better?

At what temperature? Steel has a thermal coefficient of expansion of 5 parts per million per degree F. A 20" long chunk of steel would expand or contract 0.0005" if the room temperature changed 5 degrees F.

Mariss

the nook website gives .0005 thou per foot - close. google turned up threadcraft (http://www.threadcraft.com/thread-grinding.html) who claims .0002 per foot, so that gets you there. ground screws aren't cheap mind you. iirc keeping that kind of precision over a length like that will mean paying attention to heat expansion - accuracy 1/10's is tough and the longer it is, the more thermal expansion there is to deal with.

If you making something that requires that level of positional accuracy, rather than rely on the screw's accuracy, consider the newal encoders. Their accuracy is measure in 1/100,000 inch's irrc - hey maybe we'll get Phil working on an open source closed loop servor encoder design :D

It is more a question of how did machinists get high accuracy decades ago?

^^ How did they measure it? Or maybe they thought they did?

i would say decades ago much like they do today, accuracy from high quality ground screws, on grinders driven by leadscrews.

the question remains, where did the first decent leadscrew come from. flats and squares can be generated with simple equipment to a very high degree of accuracy. Not sure how this is accomplished with screws, as the common machinery used to make them is of course driven by a screw. Before screw cutting lathes (ie that had a leadscrew) threads were laid out by hand believe it or not and cut with tools like a graver (possibly followed up with a chaser?)

keep in the mind that the leadscrew or feedscrew isn't used as an absolute reference that often over these distances, for example you wouldn't rely on the feed screw to hold a dimension over 20" on the lathe or mill for highly accurate work, you'd get close and then measure with a mic (who's accuracy is also based on a screw but is adjusted to a reference)

is this going anywhere close to helping? not really sure what you're after

PS, don't forget, the airplane you ride in, the nuclear reactor providing electricty and the the space shuttle are all more than 2 decades old - we've had the ability to do accurate work for a long time, heck, how long have jig borers been around? probably fair to say though technology through cnc has made accuracy a lot easier and accessible.

....the question remains, where did the first decent leadscrew come from. flats and squares can be generated with simple equipment to a very high degree of accuracy. Not sure how this is accomplished with screws, as the common machinery used to make them is of course driven by a screw. Before screw cutting lathes (ie that had a leadscrew) threads were laid out by hand believe it or not and cut with tools like a graver (possibly followed up with a chaser?)

At the place I served my apprenticeship in the early 1960's there was an old guy who demonstrated 'striking a thread' with a chaser in a brass bar. You just have a rest running the length of the bar and guess how fast you need to move the chaser sideways as you start taking a cut. If your guess is good when your cut comes around you have moved one pitch and the chaser now follows itself. With practice it is surprising how accurate the diameter and lead can be. I have struck threads and run on a standard nut with a fit of a few thou on 1 inch diameter. It is a bit like riding a bike or sharpening drills by hand; once you have the knack you can always resurrect it.

Regarding how screws where first machine made I recall seeing something in a museum that had a sort of winch drum driven from the spindle rotation pulling a tool along to cut a thread.

It's more a question of learning. I get hung up a times on "how'd they do that" and wondering in the early days of mills how the accomplished high accuracy.

Just a thought based on an old memory.
When building a crude wood carving machine to make rocking horse bodies on a copying lathe type set-up, we used a piece of hard wood as a lead screw nut.
The compressibility of the wood averaged out any localised innacuracies in the lead screw.
Perhaps you could bootstrap your leadscrew accuracy by a simillar method.

John

Edit - no backlash, either !

http://www.lathes.co.uk/latheparts/page6.html

This link has a great explanation of how screw cutting first started, and notice the cool info about the guy who built the first accurate screw, he made a nut about 12 inches long, to help compensate for pitch inaccuracies.

Interesting, I see that it was 50 TPI. So my next question is how fine of thread (High TPI) has been made?