Continue to Site

Welcome to MCAD Central

Join our MCAD Central community forums, the largest resource for MCAD (Mechanical Computer-Aided Design) professionals, including files, forums, jobs, articles, calendar, and more.

Register Log in

Book about calculating tolerances

Z

Zestje

New member
Hello,

I'm looking for a book on the correct methodology of calculating tolerances in parts and assemblies.
For the mechanical development of the devices we develop at work I must often calculate tolerances, which occur e.g. between parts in assemblies.
These are required to make sure the parts can be assembled, or minimum distances between parts are kept, to determine which requirement for the accuracy of a part is needed for the manfacturer and so on.

What I want to find out is if the methodology I use to gather and describe the dimensions and tolerances is correct.
I have not been able to find ANY book on this subject.

I know about tolerances and norms and so on, I'm specifically looking fo information on -how to- (methodology).

Does anyone know if such books even exist?







Edited by: Zestje
 
It's too broad an area for there to be any specific book on it. You could contact Brtish Standards Institute (BSI).


The general British Standard for limits and fits is outlined in BS EN 20286-1:1993.


However, the tolerances you use in your manufacturing base may or may not comply to the guidlines set out. Generally speaking, you should permit as large a tolerance as practically permissible to keep manufacturing costs low and production rates high. If tolerance deviation could be a problem for your assemblies then you really need to start performing 'Tolerance Analysis' whereby the consequences of tolerance ranges between components are investigated. There are software packages availablethat will do this for you.


Phil
Edited by: pjw
 
i think you need to look at the new Tolerance Analysis Extension (TAX!) in WF4. It's based on Ce-Tol and is very easy to use.
 
@all: First of all, thank you for you answers. I do want to state again, that I'm looking for information on the -methodology- of calculating tolerance stacks (tolerance analysis), not on information how to tolerance parts in drawings, tolerancing norms, hot to do geometric tolerancing and dimensioning (GD&T) and so on.
Although I have some (mostly self-learned) experience in calculating tolerance stacks, I want to verify if the methods I use are correct.

@JRay: Thanks, they have a book which seems to suit my needs: http://www.geotolmeadows.com/store/Book3.htm
I've also found these:
http://www.informaworld.com/smpp/title~content=t732535273~db =all
I'll see if I can found out if one of them is worth buying for our company.

@Blockandhole: My boss is currently getting a testing license for Ce-Tol, to compare it to the 2D tolerance calculation software we use now. Big question is if such 3D tolerance software actually allows for worst case tolerance calculation. Due to the way (most of?) these tools simulate the tolerances, they can NOT provide true worst case tolerance analysis, only statistical analysis, for which the order of coordinate directions the simulation follows has major influence on the outcome (this is the case with ViSVSA e.g.). I don't know about the costs of TAX, but CeTol is very expensive...

One more remark: I work in Germany, so we work according to European and German (DIN) norms, not American norms. The methodology of calculating tolerances must be almost identical of course.
 
It is actually just as easy to calculate worst case in Ce-tol as statistical. The question is why do you want to do worst case? The idea of statistical analysis is that the proportion of failed assemblies is minutely small for a given stack up than when compared with worst case.


Have you looked at any 6 sigma initiatives? This is the basis that Ce-tol was written for years ago.


The TAX model is Ce-tol built directly into WF4 and only caters for 2D stack ups. Full CE-tol can do much much more. Once the 'system' is modelled, you can quickly modify tolerances and Cp (Process capability) values to get to the best solution.


If you think Ce-tol is expensive, try going into production with something that doesn't fit together..... (TAX is approx $1500 for a locked license)
 
@blockandahole:
first of all, thanks for the background info on TAX.

Dependant on the customer's requirements we must often guarantee the functioning of a device in a worst-case scenario (e.g. for medical devices I believe this is very common). Statistical guarantee is only allowed in specific cases.
For 'general purpose' devices like e.g. mobile phones there is usually only a statistical requirement (Cp).
 
ReinhardN said:
@Zestje,
maybe a stupid question, cant you set your models to the desired tolerance and watch what happens ?
Click to expand...


How would you do this and what would you be looking for? Setting all tolerances to upper limit or lower limit for an assembly will have a different effect depending upon whether features are hole or shaft based. I.e. maximum metal condition on a hole will mean the lower end of the tolerances whereas the shaft will be at the upper tolerance. I think the aim of tolerances analysis is to study the effects of tlerances range on the functionality of the components. Stresses set up in components due to interference and so on.
 
in part models you can set the tolerance of any driving dimension to its desired value. But I agree, you have to know where to look.
 
In reality axis never actually line up. Specifically in Ce-tol you can set things like the way shafts go through holes to be left, right or 'floating' . That is it may be worst case if the shaft is at the extreme left or right of the hole but also statistically it would float around the central axis.
 
@ReinhardN: We must make a documentation of all important tolerance stacks including nominal values, tolerances, (estimated) Cp(k) and so on. So simply setting the models to the worst case tolerances is not an option.
Furthermore, this is technically impossible, since we mostly model plastic parts like e.g. phone housings, which rely on design surfaces which can't be tolerated like standard geometric features. Setting the tolerances of some dimensions in 3D would guarantee enless errors further down the model tree. Not to mention the amount of work it would be to change the geometry to the worst case tolerance for every important stackup...phew...

@billdaugherty: Although this is only general information about the tolerancing system (not methodology for stckups), it isquite interesting to see the differences between the metric an imperial tolerancing system
smiley32.gif
 
You must log in or register to reply here.

Sponsor

Articles From 3DCAD World

Back
Top