RE: [CAD_CAM_EDM_DRO] Re: ball screw preferences

Hi Balendo



My APTC hand mill is 5 or 6 though backlash on Acme screws.

My recently acquired CNC mill is an Avon Circa 1989, built for an aerospace
group, with low hours on it. The ball screws that were several thousand
pounds each, the servo motors are 6KW, and fetch over £1000 second hand, the
servo cards are over £1200 each, it’s a big professional machine, and was
very expensive in it’s day and weighs several tons.

I bought it cheap with a faulty CNC controller and moved it over to mach3,
and a G100.

Without having stripped the screws, I had about 0.5 (yes 5/10) of a though
of backlash, on one axis, and something over a though on the other axis.

I worked on resetting the end thrust races and got that down to less than
0.5 though, about 0.3 most of the time, this varies a little along the
length of the screw.

This measurement is taken using a 1/10 of a thou per division large clock
face DTI, (1 thou is over an inch span on the dial), nudging up against a
block, then nudging off until the dti moves.

You can push and pull on the bed, and you get about the same result, though
a hard wack on the bed will bounce the dti about 1 though.



Now you say you normally find 2 to 3 though backlash. To me this seems a bit
large for a ball screw, an acme screw isn’t much more, and a Bridgeport
screws spec says better than half a thou, which makes my findings relative
to a Bridgeport.



Last night I rebuilt these preloaded anti backlash ball screws, I can place
a mag mount on the screw, up against the nut assembly, and if there is
movement on the dti as I push and pull on the nut, it is less than 1/10 of a
thou, and that may have been a little rotary action in the nut. The ball
nuts are not quite as I stated, now they are all cleaned, the tapers on the
C washer do not come up against the taper on the nuts, these are just there
to allow the C clip to spring open past the screw without damaging the
screw. The C washer is ground so that the two ball screws nip up against
each other when the impact on the C washer, then a connecting key joins the
two ball nuts. The same as you do to build an anti backlash kit on an acme
screw. The ball nuts do move on the screw, but they are not what free, there
is some resistance. The firm that made them say this is exactly normal, it
does raise slight concern whether this all gets tighter as things warm up,
but apparently thermal expansion has been allowed for in the allows of the
design. The only play I can find on this screw now is the end float on the
thrust races.

I looked at the other screw I have not stripped on my X axis, and this is
about the same.



I do agree that thermal expansion along the bed is going to move all this
out quite a bit.

These are the numbers and measurements that fell out, they don’t seem out of
the ordinary to me, but 2 to 3 thou which you consider normal is to me very
sloppy.



I also work on spark eroders, and wire machines where we have repeatedly cut
carbide and cover to 2 to 3 microns. These machines use antibacklash ball
screws, and are heated to control thermal expansion. I just put the
increased accuracy down to the fact the there is no machining load, no drag
from long thick gib strips etc, but the ball screws are very accurate, as
proven by the work which was measured in the national physics lab.



Dr. Mark Vaughan Ph'D. B.Eng. M0VAU

Managing Director

Vaughan Industries Ltd, reg in UK no 2561068

Water Care Technology Ltd, reg in UK no 4129351

Addr Unit3, Sydney House, Blackwater, Truro, Cornwall, TR4 8HH, UK.
Phone/Fax 44 1872 561288

_____

From: CAD_CAM_EDM_DRO@yahoogroups.com
[mailto:CAD_CAM_EDM_DRO@yahoogroups.com] On Behalf Of ballendo
Sent: 11 March 2007 10:21
To: CAD_CAM_EDM_DRO@yahoogroups.com
Subject: [CAD_CAM_EDM_DRO] Re: ball screw preferences



Hello,

Snips, inserts follow...

>In CCED, "Mark Vaughan" <mark@...> wrote:

<snip> on a mill a ballscrew without backlash adjustment is normally

>within about 0.5 thou of backlash, mine measure at about 0.2
>including the endfloat on the main screw which isn't bad.

If I'm understanding what you've written here, I find it a bit hard
to believe. You're saying that your NON anti-backlash
ballscrews/nuts have 5 tenths BL? And that the entire backlash of
this NON anti backlash axis is 2 tenths? No way.

IMO/E Most NON-backlash compensated ballscrews will have .002-.005"
backlash. Ground screws may be less; and a BL compensated GROUND BS
assy. may be non-obvious (unless you had to buy it new!<G>)

The good news is that if the original poster needs to use one nut;
he probably CAN get by with one Bnut IF he reloads that nut with
oversize balls. These may be had from Ball-tec in Los Angeles, CA,
USA. I've spoken on this subject in this forum many times in the
past; a search of the group archive should show up plenty of posts
with details. AIRC, Dan Mauch has used a supplier in Chicago to
reload BN's with oversize balls.

I have always found it "interesting" how many folks have (or claim
to have<G>) specs in the "tenths' (.0001"). Since this is a
beginners forum; I'm also always trying to get folks to understand
that specsmanship is often a path that need not be trod.

Real-world commercial machines (like bridgeport vertical mills) can
be flexed with one finger enough to deviate 2 or 3 thou... And the
simple thernal exapansion of the aluminum--or wood--that many DIY
CNC machines are made from quickly exceeds tenths iduring a "normal"
shop day! Measuring to tenths is not something done willy-nilly, the
heat from your body can change parts dimensions that much!

The really cool thing is that for a whole lot of REAL-world projects
and parts; such tight dimensional tolerances are NOT needed.

Which is a good thing, since many cannot--or will not-- actually be
getting them...

You CAN get relational fits that work, and +/- .005 between features
is a more realistic AND "normal" job shop accuracy
requirement.

>On a router though at least the wood ones, you aren't looking for
>that kind of accuracy. I would try using one and measuring the
>backlash.

Probably .003" IMO/E. The double-nut preloads are often even
sloppier than the ones designed as single nuts. Because they CAN be,
since the mfr. KNOWS they will be used in "opposing" pairs.

BTW, as a former furniture mfr.; I can say that woodworking machines
DO need decent accuracy as well. Relational fits and surface finish
are the reasons.

Summing up: It's FAR easier to have repeatability than Absolute
dimensional accuracy. Don't confuse resolution with achievable
precision/ location... And specsmanship is IMO a waste of time.

>Make sure there is some sort of wiper on the end to keep the dirt
>out. Mine don't have seal wipers between.

Good advice.

>If you building a small router for something like PCB work, then
>backlash might be more critical.

True.

>Since one normally pushes against the other with a set preload, I
>would have thought they may well be a lot more efficient (free)
>with only one ball nut.

Free, yes. AND full of backlash. UNLESS they're ground screws and
that fact isn't immediately obvious to a surplus purchaser with no
prior motion component experience.

Hope this helps,

Ballendo

>>WJS wrote:
>>Subject: [CAD_CAM_EDM_DRO] Re: ball screw preferences
>>My question is can I take off one of the nuts without "screwing"
>>things up? I need the extra travel. WJS

[Non-text portions of this message have been removed]