Injection Moulded Parts - As Designed v As Moulded 2

[DJBLINX]

When designing injection moulded parts, tolerances are added to ensure the parts will go together (GD&T). When the parts are moulded they fit together, but do not necessarily meet the drawing specification. How do other companies manage this? Do you modify the drawings/models to match the actual parts (could be difficult for complex parts), insist the supplier modifies the mould tools (expensive), or do you do something completely different?

Thanks

David.

 

[fsincox]

Good question,
We have the same problem here with our castings. I assume you also need configuration control?
I am working on flushing out a concept I have seen from a big company that places a modification note on the current drawing dimensions that basically lists the engineering accepted deviations on another drawing/spreadsheet and it then becomes an “acceptable substitution approved by engineering for this part” list.
Frank

 

[pmarc]

DJBLINX,
In a company I worked for in a past (connectors design) fit and function was like Bible. That said, even if actual measurements of a component were showing some features out of spec., the part was not rejected as long as it was remaining 'functional'. This happened quite often since injection molded parts by nature are quite tolerant for dimensional unconformities.

We dealt with this by placing an extra chart on a print listing:
- in column #1 - all dimensions and geometrical tolerances that were out of spec;
- in column #2 - max or min values at which the dimensions/tolerances from column #1 were allowed to be in order to keep a component functional.

 

[fsincox]

pmarc,
That sound like it, was this company big and most commonly known by a (2) letter acronym by any chance?
Frank

 

[MintJulep]

The simple answer is that if your parts can be "out of tolerance" and still work then your tolerances are wrong.

 

[pmarc]

No Frank, but close.
It was tier 1 supplier for this 2-letter-acronym company.

 

[TheTick]

Some of column A, some of column B. We change models & drawings to match as-molded if discrepancies are acceptable. We thrash molders to in order to meet critical forms.

 

[pmarc]

MintJulep,
The answer is not that simple. It would be true for metal parts (in most cases) but for plastic components it is different.

Try to predict in which areas of a plastic component design the tolerances can be opened, even though theoretical stack-up calculations show that there is no way to let's say mate two parts together. One has to be really really experienced guy to be able to forsee this.

 

[fsincox]

pmarc,
Thank you,
Large complex castings and weldments in metal have these issues, too.

Mint,
I suspect you do not work shop that utilizes large castings or weldments?
Frank

 

[drawoh]

pmarc,

I agree with MintJulep. If out of tolerance parts work, your tolerances are too tight. Plastic is very much more flexible than metal, so you can hand-assemble interference fits, which may be why your parts still work. You should review the tolerances of your parts.

How do your parts work in the hands of your customers?

If it is too expensive to make interchangeable parts, you can always set up some sort of automated sorting process. Sorting may be cheaper than working at or above the limits of what your fabricator can do.

JHG

 

[ptruitt]

I agree with Mint, but the closer the properties are to rubber, it may be better to specify tolerances for values such as deflection force at a specific location, or volume, or mass, or durometer, or resilience, or function and limit the GD&T, explicitly identifying what it covers and does not cover. Fully utilizing 'profile of a surface' callouts is another option if you have the metrology and software and the part is rigid or can be restrained consistently.


Peter Truitt
Minnesota

 

[fsincox]

The other issue I had seen in my many years of experience is that this document also discusses is the concept that we do not want to slowly actually change the original nominal design criteria over time, that has been a sore spot for me in the past. This company has apparently had to live with the effects of that over a very long period of time and seen the value in it, where somewhere with less depth might tend to say, “Well that will be someone else’s problem”.
Frank

 

[MintJulep]

Perhaps I don't understand the nature of the question then.

By definition "tolerance" defines the allowable limits of a dimension within which individual parts can vary and still function per the design intent.

It must therefore follow that if parts are "out of tolerance" but still work the tolerances are wrong.

Of course since most people don't understand what tolerance really means there is often confusion associated with it.

So I'm suspecting that issue here may be that parts as a whole, may be out of tolerance, but that the important bits are in tolerance, which is why things work ok.

This could stem from inappropriate use of "title block" or general tolerances, or the choice of datum features that do not correctly reflect the required function of the part.

 

[fsincox]

Mint,
I tnink you do, I just suspect you do not know how much trouble it can be. Companies are in buisness, unfortunately, to make money not perfect drawings.
Frank

 

[dgallup]

In general I agree with MintJulep. However, there can be times that one dimension is out but because other dimensions are not at their limits the parts do work. Increasing the tolerance that is out could in the future allow parts that do not work to be produced. In cases like that we will sometimes allow a deviation "for the life of the tool" for that dimension. Ugly solution but sometimes there is no better one.

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

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.

 

[ewh]

The simple answer is that if your parts can be "out of tolerance" and still work then your tolerances are wrong.

Not necessarily... other parts may also be out of tolerance, accomodating the original out of tolerance part. You have a shaft with a diameter that is .005 too large (and out of tolerance), but still adequate if the hole it is to assemble in is also out of tolerance in the same direction and magnitude. No problem for that assembly, but probably not interchangable with parts from another assembly batch.

Technically, the glass is always full.

 

[fsincox]

dgallup,
Actually in general I agree with Mint too, I just can't keep up with 100 years of old drawings.
Have you also seen and actually worked with a system like I had described?
Frank

 

[MintJulep]

We dealt with this by placing an extra chart on a print listing:
- in column #1 - all dimensions and geometrical tolerances that were out of spec;
- in column #2 - max or min values at which the dimensions/tolerances from column #1 were allowed to be in order to keep a component functional.

Well, it's ugly.

But in effect what it does is document the "real tolerances" doesn't it?

 

[pmarc]

Guys,
Let me try to clarify my point with a help of very simple example.
Two components which are designed to mate together:
- solid cylinder with outside diameter [Ø]9.9[±]0.05 (male part);
- cylindrical housing with inner diameter [Ø]10.0[±]0.05 and 0.4mm wall thickness around the slot (female part).

Now, imagine these part made of some kind of rigid material (e.g. aluminum alloy). At a worst-case scenario, due to parts rigidity, it will not be possible to mate them together when the cylinder and the housing both violate their 9.95 envelopes, so designer at the design stage knows the functional limits right away.

Then, think of these parts as of plastic components. A tolerance analysis at the design stage shows nothing different to previous case. But after the items have been manufactured, due to a flexibility of 0.4mm collar, it may occur that the mating is possible even if the envelope of inner diameter is 9.90. Reading your comments to my earlier post, I understand that at this point you will say that if mating was achievable the tolerances were "wrong" (MintJulep) or "too tight" (drawoh). And okay, I understand this. But how can a designer know what is the limit below which the mating of two parts will not be possible? Is it 9.90? 9.80? 9.70? What if the wall thickness was 0.5 and not 0.4? Would the functional limit be 9.92 or 9.94?

And this was/is my point - when dealing with plastic components the answers to those question are not that simple as in case of rigid parts? Designers with 20 years+ of experience will not be able to give you clear answers, trust me. Plus, I am talking here about an extremely simplified case, which is very unlikely to happen in a company designing and producing sophisticated connector systems. So 'the extra table' method is nothing else than an attempt of saying to toolmaker: "You do not have to rework the tool every time. Even though you did not meet the spec, we (design) can live with that because the component still functions." Isn't it opening tolerances up?

And last but not least - I am not saying that this method is the one and only. It has some shortcomings too but I just wanted to reveal one of the possibilities. Nevertheless, I really appreciate your comments.

 

[fsincox]

pmarc,
Do you know why they insist that the casting tool be numbered and add the cast tool number and the dimensional variance? I would have thought just the dimensional deviations allowed would have been enough?
Frank