Tolerance Stacking

Thank you very much for your response.

In my research for this question I came across this resource: http://www.stat.washington.edu/fritz/DATAFILES498B2008/isstech-95-030.pdf

On page 11 can you explain to me what the alpha variable is in the worst case scenario?

Thanks again for your insight.

To be on safe side it is better to add up tolerance in the conventional way ,but check the difference in value of RSS type with the conventional to see that it creates problem in our case the (0.026-0.0138)

Nelson

I am also having the same issue, We are also facing lot of problem while doing parts assembly throg cad madel vs physical model. Can any one of you suggest suitable CAD software to verify the upper as weel as lower tolerance which should refelct the Physical assembly

reg

VCK. Swamy

Although I'm not an engineer, I've done a lot of parts and assemblies, and have found that life is fart easier if you use geometric tolerances, making stacking almost negligible. With digital machining it is so, so easy to get great performance, and any mainline CAD/CAM package that has 3D can be used.

I've worked as a contract machine/mechanical designer since the old Applicon 770 system days, and AutoCad 1.1, and have found that tolerancing correctly can make or break a products profitability. Anyone can drill +/-.005 on one hole, but stacking gets out of hand with more than 3 points, and if nonplanar it is incredible how rapidly things get out of hand. One of the first things I was told by a machinist was that many machinists get frustrated, and even give up on, close tolerance items with stacked tolerances, so I looked around until I found Geo in the early '80s.

Working on civil and structural engineering projects, I have a tendency to use arithmetically stacked tolerances for details and RMS tolerance for the overall object where many tolerances on individual elements would not make sense to simply add them together as it would produce a very large overall tolerance.

This is reflected in the specification of the tolerances where it gives tolerances per element and tolerances overall.

I will read your other references with interest.

The designer usually places an overall dimension that can't be violated, even though the individual stacked dimensions may exceed this overall tolerance. For simplicity we will state that this is a linear dimension and you would add them. For example: In connectors you may have centerline to centerline tolerancing of +/- 0.0005 and on a 50 position connector you may have an overall of +/- 0.010. If you stacked the tolerances to their extremes then you could violate the overall considerably. So, all though you stack the extremes for each individual stack, in linear dimensioning you must adhere to the confining of the overall tolerances. In this case you would need a +/- 0.0001" stack-up to not violate the 0.010" overall.

To go one step further: most good designers (I won't get into different designing methods here) will start at a datum point and dimension the stack-ups as they go out. For example: the first position in the connector on a 0.1250" CL may be from 0.1255"/0.1245"; the second position of 0.2500" (from the datum point) would then be 0.2505"/0.2495". You can see that if each of these 2 CL's fall within their individual tolerance from it's datum point then they would be in tolerance; but if you had an overall dimension stating that each CL position from one CL to the next can not exceed the +/- 0.0005" tolerance then placing just the first and second positions at their extremes will bring you out of tolerance. Therefore, even though you could stack tolerances you may be bound by the overalls.

This has been one of my pet peeves for years. I constantly see drawings which for instance will dimension from one edge to a hole, to the next hole and to the far edge AND also show a dimension from edge to edge - all to either 1/16" or 0.XXX". I have tried to explain why either the last hole to edge dimension has to go or the edge to edge dimension has to go but I only get blank looks.

Sadly few people who are designing parts and structures have little (and more commonly) no experience machining parts or cutting and fitting parts to be welded. These days we have cad operators who do not understand that you can't build everything you draw (or model - we use AutoCad, Rhino and Solid Works) and most don't have the faintest idea how the guy on the floor will go about building what they draw.

Years ago when I owned a machine shop (I had a number of W&S turret lathes, a Birdgeport mill and an old Van Norman knee mill, an old J&L ATL and a few screw machines. I did a run of 100 parts which were dimensioned out to 4 places (and not just 0.0005, but 0.0003, 0.0006, 0.0008....). I called and was given the kid who designed the part who assured me that every dimension was good. They were floored when they got the bill. They didn't want to pay and I explained to them that those zero's cost money. Then they insisted on mic'ing every part (talk about a pissing contest). Of the 100 parts every single one was right on target.

Turns out the parts were a handle that attached to a lever and the operator wrapped their hand around it.

As they say - There is no such thing as a cheap education.

Travis

Well said, I work with a lot of physicists & electronic bods who will come up with schemes for building equipment. They are often stumped when I get to the part about 'how would you make that shape you've just sketched'.