View Full Version : leadscrew accuracy

03-05-2003, 04:36 PM
All the discussion of backlash, accurate measurements, ballscrews, etc. I am wondering, how accurately are leadscrews made? It seems to me that in order to make, or machine a leadscrew, there has to be a considerably accurate machine involved, even for a relatively cheap leadscrew. Assuming taking all plays into account, backlash, gibs, and eccentric indicator dials, would not the leadscrew run the table very consistently over it's length? Or is there apt to be spots where, say, 1 exact turn of the leadscrew is not an equal distance to another exact turn at another spot on the leadscrew? How does one go about selecting a leadscrew as a replacement, in order to increase accuracy, insuring exact table displacement per degree of turn of the screw. Are we screwed on that account with our off-shore machines?

03-05-2003, 05:05 PM
Below is a link to a web page that gives the accuracy of two grades of leadscrew. The "standard" grade has an accuracy of 0.010"/ft and the "precision" grade is 0.003"/ft.

Another manufacturer describes their precision ground leadscrews as having 0.0004"/ft. accuracy.


03-05-2003, 05:26 PM
You can certainly replace the lead screw in your import machines with a higher quality but there's no sense in doing that if the tolerance on the way, spindle, and tailstock are not up to snuff.

How accurate can they make lead screws? Well, think about the "screw" in your micrometer. It's an inch long and you'll be luck if it's accurate to +/- 1/10 thou. You do the math for a longer length.

Even if you're leadscrew is not accurate, you can still compensate by tweaking the gear ratio when cutting screw. This way, you're finished product is more accurate than your machine.


Rich Carlstedt
03-05-2003, 06:25 PM
Well yesterday, I was in Chicago at the big show there ( Plant Engineering) and talked about almost the same subject with Rockford Ball Screw Staff.
The .003 " per foot spec mentioned above is the "standard" for a precise "Rolled Thread" screw.
.010" is pretty crummy specs or in other words not a Leadscrew at all in my opinion.

Neither of them is a "Precision" screw !

When you want better, you must go to a "Ground Thread Leadscrew" whether it is a ballscrew or an ordinary acme thread, or a 60 degree V as in the case of a mike.

Rockford had a New replacement set of rolled ballscrews and nuts for a BP at the show and it cost about 650 as I recall. they were accurate to .003 inches per foot....OK if you have digitals in my opinion.
The same set "ground" would cost about $3,000 in my opinion. A ground set would get you .0001" per ft accuracy

Twenty years ago, I was at a machine tool show, and BridgePort was fighting for its life with all the imports on display.
The BP salesman when asked "Why pay more" by a visitor put this challenge to him, and I was amazed at the results, having watched it done several times with the BP getting it right every time. The BP had a Digital Readout on it and was a stock manual mill.
The salesman would move the table in one direction and remove all backlash going that way (.004 to .006 as I reacall). Then he would zero out the digital and the handwheel, and push the digital display so he could not see it. Then he asked the customer to give him a dimension like 2.065 or something like that. He would turn the handwheel and get it dead-nuts !! Then to show the acuracy and repeatablity of the BP, would return it to Zero (again passing Zero and coming back to remove backlash) ALL without him looking at the DRO,Which we saw plainly.
It was amazing....then the challenge...he said "go try it on another mill and you will have the answer!"
The reason is the precision that BP used in making their screws. They were all Precision Ground.I would guess to .0001 per inch based on the results
The "Imposters" did not match it even close. I know I tried. the best I found was still off .0015

Was BP correct in their approach ,Yes and NO.
As a Manufacturing Engineer, I appreciated the precision and thats what we paid for...however, do you need such precision is the question?.
If you have a mill with a Digital Readout, then I can say NO, you don't need it. The Readout with its precise scales gives you the correct answer, and without the concern for thread slop (lash). So BP lost on this point and spent unnecessary money. Now if you say ,"hey ,I might loose my DRO", then you may want to keep the precision.
Thats how foreign builders get their foot in the door.
If your CNC mill uses readouts for positioning , then a "Ground Screw" would add unnecesssary cost, but if the mill uses rotary resolvers (like my Hurco) then a .0001 Ballscrew is manditory, as that is what the control uses as a criteria.

Lastly, we built dies where I worked, and I came up with this formula for our engineers and accountants to estimate costs.
If a part weighed one pound say, we looked at raw material cost...say $1
If used as forged or cast then add the material cost again so our cost would be $2 total.
If machined to +/-.005 then add 5$
To go to =/-.001 then add another 5$
To Go to .0005 TIR(some dies had it)add another 5$
You get IT ?
1$ to $2 to $7 to $12 to $17
all based on the accuracy requested....
So it is imperative that you address how much accuracy you require.

Most Home Shop guys go way overboard here...and I am one of them too !
If you intend to machine parts to .0005, then go for it. If you are fixing the old farm tractor, then realise it...saves money

03-05-2003, 09:46 PM
Thanx for the input, guys. I don't intend to waste that much money to get virtually nothing in return. I'm intrigued that a dro would give better accuracy, isn't it limited by the measuring device it uses to assess movement- and what do they use to measure table position, anyway? I can understand if they are reading turns of the leadscrew, and fractions thereof, and of course, that just makes the dro an easier way to read the positioning. How does a dro bypass the other inaccuracies, such as table rocking from loose gibs, worse as it comes from 2 axis. Is it probable that on a cheaper, or import machine, that the dro would just be a convenience feature, and not really help you to get better accuracy? The last question (for now) ! is , what would a hobby guy use to measure a precise distance of, for instance, 14.508 inches, within a couple of thou? A 16 inch micrometer, readable in ten-thou, but out possibly by 3 thou over a foot?

[This message has been edited by darryl (edited 03-05-2003).]

03-05-2003, 11:49 PM
Hey I can tell you that a dro's sole purpose in life is to keep track of table movement I have seen them used on machines with rack and pinion table feeds worked just as good.They come in handy on old machines more than new. The debate about import-vs-U.S.has and will go on for ever I say leadscrew- shmeadscrew if you can't lash out the nut it doesn't matter if its high dollar toler. or hardwarestore althread if you can't lash out the nut you got junk!I have a Canadian Excello b-bport that has precision leadscrews and a zero-lash nut I adjusted it once and three years later I still have .0001 or less backlash.As for import mills you can pay a little more and get a Taiwan unit with factory installed ballscrews made to iso 9002 standards.But also remember accuracy still hangs on rigidity most b-ports ain't got it foriegn or domestic.It also hangs even more on the guy in front of the machine.Anybody can make a precision part with a new machine with tight tolerences it takes a machinist to do it on junk.As for the length measurement problem if you are fortunate enough to have a dro use it with an edge finder.My Accurite is close to.001"on 15"And if you look at it this way even if you have a large mike or caliper the minute you pick it up with your warm hand you ruin the accuracy.They also have to be calibrated more often and require a ring gauge and a temperature controlled room to work properly.With out these any claims of .0005 +/- in the home shop or the runof the mill job shop are b.s. you can claim that on part repeatability but not accuracy.

Forrest Addy
03-06-2003, 12:21 AM
Wierdscience wrote: "Anybody can make a precision part with a new machine with tight tolerences it takes a machinist to do it on junk."

Maybe I am a machinist after all. The only two new machine tools I ever ran I bought for myself. All the rest were old when I started my time.

I once walked across the table of an old 4" Lucas horizontal. The must have been balanced pretty close on the saddle convexity because when I got to the edge to jump down the table made a slurping noise and settle with a thump beneath my feet.

Some of those old Bullard saddles were shaped like rocking chairs but if you shove the right shims under the ends you can bore straight holes.

I once ran a turret lathe that pooled coolant in the hex turret so long as the slide was fully retracted but it ran out of the tooling hole when you extended the slide.

An old guy in our shop went to swing the aarm of a 11 ft radial arm drill clear of the work so the crane could get to the part. He stabbed at the clamp button and missed and the arm swung over 90 degrees and banged against the wall mashing a big hole in it.

When I was an apprentice newly transferred to the engine lathes Milton Brown, a classmate, called me over. "Listen", he said and ran his lathe spindle at a hundred RPM. The jaws of the empty chuck clattered in their clearances like four sets of castenets. It was the same with my machine.

Nope. New tight machine tools are for sissies. Where do I sign up for another one?

03-06-2003, 12:48 AM
Ultra precise ball screws (beyond .ooo1" accuracy/foot) are made on special grinders and lathes with laser interferometers accurate to a millionth of an inch.

As Rich has pointed out you pay for what you ask for. Thompson/Saginaw use the above equipent to produce ultra-precise and space grade screw and linear splines. It is worth noting that many ball nuts have excessive wear because the balls have no clearance space and tend to rub and the preload is set wrong. Thompsons ballscrews are ground so pressure is on opposite sides of the balls and they do not drag in the groove. Lower friction - they have better seals too.

03-07-2003, 12:29 AM
Heck those machines are in great shape Forrest,we got a lathe that the latest date on it is 1853!Told a new man who drew work on it not to panic if the headstock shaft floats up and down an 1/8"or so!Also ran a Landis threader that pukked more oil on the operator than the part! Once had a b-port mill that cut an .090"crown over 16"travel table shook like a leaf in a breeze.But an old timer I learned a lot from told me though "Son if you can manage first class work on one of these old battle axes imagine what you can do on a virgin"one more thing I found to be true. http://bbs.homeshopmachinist.net//biggrin.gif

03-07-2003, 01:23 AM
Searched around and found references to glass encoders and chromed glass scales. This seems to be the standard input device for dro's. I imagine they are etched with very fine and consistent pattern which the dro electronics reads from. One spec says 10 um or .0005 accuracy, can that be right? Half-thou? I suppose that's good if it's over the length of the scale. Anyway, that's suggesting that glass is more dimensionally stable than other materials, metals being the alternative. Not bad for a liquid!

03-07-2003, 01:27 PM
Darryl: I wish you had not mentioned glass being a liquid!

Your turn Dave (thrud) http://bbs.homeshopmachinist.net//smile.gif

03-07-2003, 02:09 PM

Yes, glass is more dimensionally stable than steel in relation to temperature but only in the lower temperature range.

There's an on going myth that glass flows because it's technically a liquid. I won't go into the disucssion of whether it's liquid or solid since even the scientists argue over that one, but there is no imperial evidence that glass flows at room temperature. The fact that glass windows from centries back look distorted is because it was that way the day it was put in, however the tour guides at old castles and churches will always tell you otherwise, and the so the myth continues.

I think some vendors muddle the defintions of ACCURACY, PRECISION, RESOLUTION, and REPEATABILITY. You'll notice that some digital caliper manufacturers don't even specify the accuracy. I can only say, buyers be aware.


[This message has been edited by Rotate (edited 03-07-2003).]

03-07-2003, 02:56 PM
I have some old cheap drinking glasses (ie. water glasses) about 35yrs old, that I know have become distorted over time. They no longer stand straight, nor are they as round as they once were. If I'd known that was gonna happen I'd have measured and documented their condition so I'd have some specific data to inject into this discussion. Hmmm. Maybe I'll get some new ones for that purpose. You folks check back here in March, 2038 and I'll have the results posted. http://bbs.homeshopmachinist.net//smile.gif

03-07-2003, 03:34 PM
2038? Sorry, I'll be 118 by then. Tell you what, I'll come back and haunt you. Dial in crypto on a ouija board.


03-07-2003, 09:01 PM
In one of the Lindsey books it talks about making master leadscrews in the early 1900's, late 1800's. They measured the point-to-point inaccuracies in the existing master screw then made a grooved plate as long as the master screw to mount in front of the leadscrew nut. The nut was mounted so that it would track to the curve of the groove. The curve of the groove would cause the nut to rotate slightly up and down effectively advancing and retarding the leadscrew to remove the perodic errors.
It was supposed to be accurate to 0.00004" over 4' or maybe it was 0.000004" over 4', I don't have the book here, but it wouldn't surprise me a bit if they did it.

I was thinking that if you had a truely accurate measuring device (you define it) you could make a table of the errors in the leadscrews of a homemade cnc machine and let the software adjust travel for the errors. Providing of course that you worked in a reasonably stable environment. This should make a so-so leadscrew into a "higher" accuracy one - if you can't tell she's faking it, who cares... oops, who let Thrud in.


03-08-2003, 12:28 AM
Darryl: I think your qquestion about DRO accuracy wnet unanswered; The DRO has two pieces, one mounts on the piecethat moves. the other mounts on the peice that dont move. The DRO tracks how much they move in relation to each other. You could mount a DRO and use a hammer or ccrow bar to movethings. Might have trouble positioning to where you want it, but you WILL know where its at http://bbs.homeshopmachinist.net//smile.gif.

I think,for hobby purposes that most precision over say two /three inches is mainly for getting a warm fuzzy feeling. There is seldom need for great accuracy on big stuff. You may need a shaft 36" long and 1" dia. but in use it can probably be 36 plus minus a lot. and it needs to be 1" plus minus a little in two or three spots. The rest can be maybe smaller by a lot. Its niceto beable to make 6 holes equaly spaced, but in most cases there is no need for them to be exact- template and drill where the holes match. Gear is different matter, pulley can be way off except the anglle for belt needs to be fairly close. I tried to leave tolerance to max that would do the job for one off stuff.

For mass production thinngs change- parts must interchange- that needs precision especially when tolerance stack up.

The "Arkansas ruler" should be used when possible.

03-08-2003, 03:38 AM
I get a warm fuzzy feeling just seeing all the responses people get to their inquiries on this forum. It's great. By the way, Lynnl, by the year 2038, I'll probably be dirt. If not, I probably won't know what glass is, or anything else, for that matter. If I'm still around, I'll drink to your health from one of those glasses, if they're still standing. I won't be.
Interesting stuff about making master leadscrews in the days of yore, Kevin. I'm always amazed at the ingenuity of our ancestors.

03-08-2003, 02:21 PM

If you get the "Foundations of Mechanical Accuracy" book from Moore you can get answers to all your questions. The standard spec that Moore used to quote for their lead screws was 0.000,001"per inch. I bought machines that had straightness of travel of about 0.000,010" peak to valley over 18" of travel. Straightness of travel roll, pitch, yaw of course encompasses more than the lead screw. Moore had a very precise and very special tread grinding machine in a very special temperature controlled room for grinding the lead screws. The checked the quality of screws against a master leadscrew. They lapped the nut to the screw as a matched set and tested the assembly against a master step standard for accuracy with hand lapping correction to get the delivered accuracy of the assembly. This is takes care of the point to point "bump" or other error you asked about. Very expensive but you get what you pay for. Now days Moore has stopped making those Acme thread lead screws and uses precision ball screws and relies on laser interferometer position readouts or precision scale position feedback for positioning. Ball screws are cheaper and with proper preload have less back lash and they also are more efficient for faster slide travel which is a modern priority.

Ragarsed Raglan
03-10-2003, 12:49 PM
An interesting thread this one. I like Forrests comments about old worn machinery making a better machinist out of folk who have to produce accurate components from them. During my apprenticeship the use of old worn out machinery was an essential part of the 'training' - and my time was well spent on a 1916 Herbert #16 Universal mill, and a 1941 Milwaukee Vertical.

The first machine was part of an order from the Czarist government of Russia, it had been rejected by their inspectors based at the factory - so was put to use in the same factory! The Milwaukee came as Lend Lease aid during the early part of the war and was removed from the freighter that delivered it to Liverpool docks from under water. The freighter had been sunk at the dockside in a bombing raid. The only way to dial in an accurate table movement, on either machine, was to use dial indictors on the axis of movement. I remember having to cut a slot to a Newall A limit (something like +0.0005"/-0.00025") in a 'fitters return', using dial and gauge blocks to get the table movement across the slot. I got a commendation off the foreman for this job on my indentures - apparently they didn't think it was possible to do this job and were ultimately prepared to strip the job down and replace the component that had the skipped operation on. For some reason the shop only gave 'fitters returns' to the senior apprentices - probably as they were an open ticket they could make more money on them by giving them to the 'cheapest' labour (i.e. the Apprentices!!).

As for leadscrews I remember seeing the leadscrews made by Matthew Bolton, for the Wall lathe he built at the Soho works around 1760. This machine is now housed in the Birmingham Science Museum in the West Midlands, UK, the leadscrew was made by marking out, and then hand chiselling. Three screws were produced and then each was corrected on the lathe against the two others to produce three accurate screws (in a similar fashion to producing a flat surface by hand scrapping three plates against one another). Parts of the marking out (by centre pops) is still visible on the screws at the unworn ends.


03-17-2003, 05:32 PM
I know on my Smithy 1220xl, I have just gotten used to using indicators on evrything because the dials are just not accurate enough.I hope to install sony/red-lion digital readouts on it soon. (as soon as i can cabbage enough parts from work)..
When i read post like this and the ones about the temerature of the steel creating errors etc. I wonder how we manage to make anything fit.........matt