How do you handle people who belittle GD&T and point to their success without it? 1

[Tunalover]

My boss today discredited my use of the fixed fastener formula of Y14.5. Since he had never heard of it (he's an EE) and probably because I'm a new guy at the company, he essentially discredited my use of the formula and said "we've made hundreds of thousands of parts here over the years I've been here and we've never experienced fit problems between our PCBs and their housings."

I tried to tell him that any process change, tooling change, or supplier change can cause holes to drift away from their true positions. If they do, and the parts still pass inspection, then the parts are still accepted because the formula guarantees that the parts will fit together as long as all the holes come in within their allowable size and positional tolerances. He summarily dismissed the entire subject before I had a chance to show him how the shape of the tolerance zone alone provides a 57% bigger area for the hole centers to land. He stated that GD&T should only be used for fit-critical situations because the symbols drive up the cost of the parts.

How can you talk sense into someone with this lack of knowledge and appreciation for GD&T?

Tunalover

 

[djhurayt]

Even the cheapest calculator requires manufacturing precision far beyond what goes into a drive train.

Really ??? , do you actually believe that statement? I'll have some of whatever your smoking

 

[3DDave]

ICs are now pushing 15nm. That means alignments for these chips on the 1-2 nanometer level. What are you smoking?

 

[SwinnyGG]

IC paths are fabricated at 15nm, sure- that doesn't mean the chips on a board in a $3 calculator are aligned to single-nanometer levels.

You're also not likely to find 14nm chips in 'the cheapest calculator'.

 

[dgallup]

Very little machining going on in microchips. The lithography machines required to make state of the art chips are huge investments that only a handful of companies in the world can afford. The fact that such things can be fabricated at all has no bearing on dimensioning and tolerance analysis for the rest of the world.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[3DDave]

The fact that such things can be fabricated at all has no bearing on dimensioning and tolerance analysis for the rest of the world.

Neither does the specialized development of high end, mass produced automobile transmissions, which only a few companies in the world do. I think the number of companies designing transmissions is about the same order of magnitude as those designing IC manufacturing machines and far smaller in number than those designing ICs.

Interesting the hostility to the idea that (generalizing) Manufacturing, which makes those transmissions, won't continue to be improved to an extent that most mechanical components will be made more precisely than their operating variation limits require, eliminating the need to specify those limits. Except, it seems inevitably, for 8 speed automatic transmissions, which will forever require a couple of experts to tell everyone else that it never will.

 

[dgallup]

Many products continuously push the current state of the art manufacturing capabilities, ensuring the continued need for concerns about dimensioning and tolerance analysis. Look at the aspheric lens in your cell phone, the DI fuel injectors in your car, the artificial valves about to be implanted in your heart. Your example of micro chip manufacturing requires great concerns about dimensioning and tolerance analysis to achieve acceptable production yields. The fact that a cheap chip can do 64 bit floating point calculations only makes it easier to carry out dimensioning and tolerance analysis, in no way does it eliminate the need for it. We have had a new CNC machine sitting at the supplier for 6 months because they have yet to meet the run off Cpk requirements. I see a long long future for improved production processes and concerns about dimensioning and tolerance analysis.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[SwinnyGG]

As processes improve tolerances improve. As tolerances improve, products improve. As products improve, design tolerances tighten. As design tolerances tighten, processes are pushed to their limits. As processes approach their limits, they must improve and the cycle starts over.

This will never end.

 

[Tunalover]

Oh I forgot to mention another reason why my boss thinks that all their hole positions have been trouble-free:

In most companies, the pass/fail inspection rate is typically kept by the QA Engineer who steps in any time parts are rejected because they are out-of-spec. He normally calls or emails the supplier directly to figure out what the problem is and see that a solution is implemented and providing the expected outcome to the buyer. Many parts are rejected and returned for a refund by the supplier without Engineering ever knowing about it. but that number is not shared with the design engineers if it's not deemed a design problem. So the engineering manager (my boss) never knows about it when parts are thrown out for not meeting the print. Of course by virtue of the shape of the bilateral tolerance zone being a square and the GD&T tolerance zone being circular, the GD&T tolerance zone will be 57% greater in area (putting aside additional tolerance if the hole departs from MMC).

Hell, multitudes of functioning parts have been tossed out in the past because their centerlines did not fit in the box-shaped tolerance zones from bilateral tolerancing. Common sense says that dimensional variances happen radially from the center and that, as long as the hole axis falls within a certain radius of the true center then the part will fit. With the bilateral method, if the tolerance is, say, ±x in both directions and the part comes in with a hole at coordinate (x,x) then the part just passes. If it comes in at (x,-x),
(-x,x), or (-x,-x) then the part also marginally passes inspection. Those coordinates are the corners of the square tolerance zone. The distance from the hole center in each case is x√2. But what if a hole came in at (0,x√2)? This is the same distance from the center but is outside the box. The part is thrown out. But we know the part will fit if the hole center falls in a circle of radius x√2 about the true position!

I think these concepts are too large for my boss to follow so I think I'll just have to give up trying to convince him he is wrong.



Tunalover

 

[drawoh]

...

Of course by virtue of the shape of the bilateral tolerance zone being a square and the GD&T tolerance zone being circular, the GD&T tolerance zone will be 57% greater in area (putting aside additional tolerance if the hole departs from MMC).

I don't generally push that advantage of true position tolerances. Only if your process is marginally capable of meeting requirements, does true position tolerance reduce your scrap rate. If my process were marginally capable of locating holes, I would enlarge the holes.

The thing I like about true position is that it applies the tolerance directly to the hole. This is as opposed to applying it to a line that passes through several holes, not all the same size, and not all the same tolerance requirement.

--
JHG