Tolerance analyses rotary tooling 4

[Terry2001]

I’m a third year mechanical engineering student.
Currently I have an internship at a company that provides rotary converting solutions for the packaging industry.

My assignment is to re-evaluate the current tolerances on one of the machines and see if any tolerances can loosen to possibly save money and make rotary tooling more interesting for smaller volumes.

I’ve done some research and I haven’t found much useful information except the different kind of analyses and small examples of 1D problems.

I’m kinda stuck at this moment on what to do.

Can anyone help me out?

 

[MacGyverS2000]

Pick a toleranced part, then look at what will happen if you loosen that tolerance... will the machine fail if the part gets too small, will it seize up if the part gets too large, etc.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[Pud]

Investigate 'Statistical Tolerancing'.
It's a bit of work to do but well worth it - especially for complex assemblies.

Cheers


Politicians like to panic, they need activity. It is their substitute for achievement.

 

[3DDave]

Most always "loosen tolerances" means "accept greater variation" which means "accept greater uncertainty."

It's unclear why, but in talking with some it appears they think that "tolerance" is some in bucket to be ladled out and for no good reason engineers just don't want to.

The fact is that the greater the uncertainty the greater the difficulty in seeing that the product will function as expected and, when it fails to do so, the greater the difficulty in identifying where the problem originates.

Almost always better precision or cheaper manufacture requires a new concept.

For example, instead of increasing the precision of placing parts onto a conveyor belt in an automated assembly process, a great improvement is to use computer-vision and greater degree of freedom robots to handle misaligned parts. Done to the extreme, one can dump parts right from a box onto a conveyor without any alignment mechanism at all. Any items that are so mis-oriented the robot cannot pick them just get swept to a conveyor that dumps them at the start again where their new orientation might be better.

To do otherwise one needs to determine what variations are acceptable in the process. This includes how much parts can deflect in the machine and in the material being processed. Then look at how the full machine contributes to that variation and see if there are some parts of the machine that have little influence.

 

[IRstuff]

40 years ago, Japanese car manufacturers began their dominance of the auto market because US factories thought that tolerances were simply guidelines.

Ditto US semiconductor industry. We once second-sourced a Hitachi second source of a Motorola CRT controller chip. We got the mask set and to our surprise, the usual test devices we used to monitor our process weren't there. We asked we were supposed to check our process; they replied, "Don't worry, it'll be fine. Those pesky test patterns are usually located in the prime yield areas of the wafers and we don't like losing them."

They didn't provide a process recipe; we asked them what to run; they replied, "Don't worry, just run your usual process." OK, so we did. The first lot yielded twice as good as the best yielding of our own products, specifically designed to run on our process. The 2nd and 3rd lots yielded even better. The moral of the story is that tighter control and understanding of ones production process is usually better and results in lower overall life costs, better yield, etc.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm