【转帖】help on lead screw assy tolerance

yang686526

help on lead screw assy tolerance
sketch attatched:
we have a detail assy that鈥檚 a lead screw and its boss that we purchase from an external vendor.
apparently we鈥檙e having a problem that the 鈥榖all鈥?on the end of the screw is not very co-axial to the boss od as it moves.
i鈥檝e been asked to come up with a tolerance scheme to better control this.  i asked the stupid question of what co-axiality they need and of course they don鈥檛 know although they did seem to think that the .1mm they were seeing was too much.
they are treated as a matched pair so i鈥檓 thinking it may make more sense to apply the tolerance to the assembly than to directly tolerance the 2 piece parts.
any suggestions comments 鈥榗ause i鈥檓 way out of my league on this one.
i鈥檓 thinking maybe a tight surface profile on the region of the ball that mates with the v groove on mating part in operation.  this would be tied to the boss od as the datum for co-axiality.   how do i define that this is over the full range of movement, with a note or is there someway to almost apply a projected tolerance zone?  or maybe i should work back the over way.
i鈥檓 guessing though that at this time most sane forum members will be in bed or at least at home.  and as they鈥檒l want an answer first thing in the morning i鈥檒l get to make my best guess and regret it later!
this is real hot so i鈥檓 coming back in later, fun, fun, fun.
kenat, probably the least qualified checker you'll ever meet...
need more info to help you.  this is a simple vector loop if you have all the info.  i would venture to guess that the largest contributor to variation is the thread fit.
xplicator, my boss & i were discussing that very thing just before i left, thanks for reminding me * for you.
i don鈥檛 think i鈥檝e every had to look at the 鈥榳obble鈥?or i suppose 鈥榬un out鈥?of a screw in a nut/bushing before.
taking into account just the effect of the threads how do i calculate how much the diametrical location of the tip of the screw can vary.
i鈥檝e tried google searches of both this site and the whole web but haven鈥檛 found an answer.  i鈥檝e also looked in machineries and in precision machine design (alexander h slocum) but can鈥檛 seem to find it.
is it as simple as calculating the maximum clearance between pitch diameters and calculating what displacement this allows over the length of the bushing and then multiplying by the length of screw/bushing length?  does the below make sense or have i made a school boy error?  
sorry for the late reply, i checked back late last night but must sleep at some point.  the actual radial float you have identified is just a part of it, backlash would also be a contributor.  similar to runout on circular ball bearings is a contributor to total axial end play or "wobble".  there are multiple considerations when dealing with all the geometries and dynamics involved here.  i don't know for sure about your scenario, but in bearings i have found that axial end play can be as much as 8-10 times bearing runout.
when in these situations in the past i would yield to experts in their field.  contact the  suppliers of lead screws or motion control products and pick their brain a bit for some ideas.  also search for backlash on lead screws and solutions for, such as pre-loading.  this my not provide the exact answer your looking for but should point you in the right direction and will definitely educate you and your boss on what you are experiencing.
which by the way is not so much a gd&t/dim scheme issue but a design consideration that wasn't considered.  although, you may be able to limit or minimize it with robust gd&t, proved out by a stack analysis.  suggestion, i would use pd as datum instead of od of boss.
i really hope this helps get you going in the right direction.
xplicator, i was trying to do my own research that's why i didn't post again sooner.  thanks for checking back though, really appreciate it.
i think you're right that there's a design flaw and they're looking at adding gd&t as a band aid not a real fix.
i've got to step out for a moment but will look more closely at your suggestion when i get back.
thanks,
ken
kenat, probably the least qualified checker you'll ever meet...
well i have some more information.
the screws are used on a pseudo kinematic mount to adjust the height & angle of the sample under our measurement head (or they might adjust the head height not totally sure but same end effect).
one screw goes into a cone, one into a groove and one just onto a flat plate.
there is (or at lease was) sometimes visible play of the screw in the bushing.  originally the bushing was much shorter but was lengthened to reduce this play.
this is causing massive problems during assembly, not so much in operation by the customer.  it takes our assemblers a long time to find sets of parts that work and make all the adjustments etc to make it work.
so it鈥檚 a design problem but there鈥檚 no time/money for a good solution.
so plan a is just to get the vendor to effectively do some of this for us by making matched pairs of bushings & screws that have better performance/acceptable play.  we are willing to accept that this may mean significantly higher price of parts, we鈥檙e currently playing less than $x for a pair and even it they cost $3-4x if it saves significant assembly time it works for us apparently.
to support this i鈥檝e updated my sketch with a proposed tolerance scheme.  please let me know if you think this is of any use at all.  i鈥檓 still putting the bushing od as the datum as functionally this is also important.  even if the runout of the 鈥榖all鈥?was good to the pd but the pd wasn鈥檛 closely co-axial with bushing od it would cause a problem.
thanks.
kenat, probably the least qualified checker you'll ever meet...
oops, helps if i attatch the sketch.
kenat, probably the least qualified checker you'll ever meet...
why circular runout rather than total runout?
at any rate, regardless of tolerance scheme, the problem is, how to achieve it?
how about cutting the nut into two and putting a spring between the two sections.  similar to anti-backlash nuts used on power transmission screws?
seems like that could tighten up at least some of the play in the threads.
kenat,
to be honest i'm not exactly sure how this will help or hinder you at this point, since per your description there is so much more involved with this fixture than just a screw and bushing.  
one would have to analyze the build objective (height and angle of sample desired to the measuring head) and determine where and what the sensitivities are within the all the components features that are contributing.
having said this, i would say tightening up everything until your supplier screams is your best course of reaction to not fixing the design oversight.  just make sure this is documented as a costly workaround and not deemed fixed.
simplistically only one circular element should be in contact with the conic hole or v groove at any one time.  as such i was thinking circular runout would be adequate.
as to how to achieve it.  looking at the web site of the current supplier they claim "our manufacturing techniques result in tolerances 30% better than the tightest industry standard (class 3 fit)." which reduces the componenet of tolerance stack caused by thread play to around 脴.2mm worst case.  given that this is 'worst case' then i'd think that matched pairs might be able to do a lot better.
anti backlash screws do exist 'off the shelf' for 1/4-80 so i'd want this to be looked at longer term.  the ones i've found aren't long enough but i'd expect they may be able to make a special.
however they still haven't told me what value of runout they need is, looks like i may get given the whole sub-assy to analyze though which with the state of other drawings from this product that i've seen could be fun.
thanks for the input.
kenat, probably the least qualified checker you'll ever meet...
xplicator, your post hadn't shown up when i put the above, i think i pre-empted part of it with my last line though.
quote:
having said this, i would say tightening up everything until your supplier screams is your best course of reaction to not fixing the design oversight.  just make sure this is documented as a costly workaround and not deemed fixed.
couldn't agree more, this is standard op at this place!  well, except for actually documenting it as a work-around.
however, it sounds like managment have made some progress in this area, they've decided we definitely aren't going to get another checker in the short or medium term.  so i better ramp up the job search as the next step is probably getting rid of the one we have.
kenat, probably the least qualified checker you'll ever meet...