Document Tight Tolerances 1

[bpcash]

At our company, whenever a we have a problem with a supplier meeting our tolerance, there is grumbling from management, purchasing, etc., the gist of which is "@%$# engineers can't design something that anyone can actually build". While I'm sure there are cases where this is true, there are times when tight tolerances are justified. I think our problem is that we do a poor job of communicating *why* we need such tight tolerances. Our design process seems to just output the specs and drawings, but the reasoning exists only in the mind of the engineer who did the work.

Anyone else have this same problem? How do you document the decision to go with +/-.0002, or to use some exotic material?

 

[ctopher]

It's usually ignorance from others.

I once had a drawing of a part that was scaled 100:1. the dims and tol were in microns.
The purchasing idiots changed the tol from .00002 to .005 & .010. They said it was to save cost and make it ease manufacturing.
With this new tol, the part can be air, or much bigger.

Hold to your guns, tell them that's how it is. Show back up and a tol stack-up with the dwg package.

Chris
SolidWorks 10 SP4.0
ctopher's home
SolidWorks Legion

 

[KENAT]

Ha ha, I've been round the 'exotic material' loop lately. We use lots of Invar for thermal stability and purchasing kicked up a stink because it's expensive, only certain machine shops will work with it to the tolerances we need and it has long lead time.

Fortunately, for most of the high ticket items they flagged we had tried not using invar and had done testing that showed the difference. Once this was established they were more willing to accept our arguments on other parts where we hadn't tried not using Invar but were able to do some ball park calcs etc.

Finally, for the couple of parts we thought were worth trying non Invar version of (legacy parts copied onto a newer system because they were 'common' rather than based directly on performance), the cost savings they showed were too small to justify the amount of testing required. This one sucked a bit as I'd spent a bunch of time creating drawings for lower cost version but c'est la vie.

On the tolerance issue, baffle them with math. Some time back I posted a spreadsheet for the hole position tolerancing per ASME Y14.5-M-1994 appendix B(2?). Some folks questioned why have a spreadsheet for such simple calcs, well this here is one reason. Being able to show those calculations to folks in a well presented format seems to hold more sway than some chicken scratch 'on the back of a fag packet' (as we used to say in the UK).

Aerospace and the like tend to be better at keeping copies of design calcs than some other sectors - heck in Aerospace they're often compiled into formal documents. This may be excessive for you but at least if you do calcs in some kind of log book then you can go back and look at them.

Hopefully if you're able to show them the math, or the test results a few times then they'll gradually learn that most of the time you aren't making this stuff up and they won't question it.

However, keeping design records such as tolerance calculations etc. for things like this - especially critical or non standard things etc. may be a very good idea.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[drstrole]

One of my primary functions is checking. I use a spreadsheet and enter almost every dimension and tolerance and show clearances, presses and any other information. If one value is changed the values driven by the entered information changes, so you can see the effect of changing dmensions and tolerances on other values. It also gives me something to show my train of thought with the designer so we can figure out the solution if there is an issue.

Drstrole
GDTP - Senior Level

 

[Belanger]

The best way to achieve optimum tolerances: talk to the customer and see what the design requirements are (such as the allowable space that your package will fit into). This is one driver of tolerances, sometimes called the "top-down" approach.
Then talk to the manufacturing folks and discuss what is the capability of the machine... this is the other driver and is sometimes called the "bottom-up" approach.

Now work from both ends, getting to the middle to determine an appropriate tolerance. (This is a greatly simplified explanation, but that's the main idea I recommend.)

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

 

[drawoh]

bpcash,

My experience has been that a lot of people generate fabrication drawings in brain-off mode. Tight tolerances are thrown onto drawings primarly for the heck of it. Your fabricators may not be taking you seriously. If you specify loose tolerances when you can, they are more likely to believe you on the tight stuff. At least, you are showing them your priorities -- what to work hard on, and what can be ignored.

You do not need to communicate why you need tight tolerances. This is design information that, at worst, could be proprietary.

JHG

 

[KENAT]

drawoh, he's not talking about explaining it to vendors/machine shops etc. He's talking about justifying it to his managers, purchasing folks...

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[SDETERS]

Most of the time when we have tight tolerances We flag that dimension and tell the supplier that they have meet this dimension. Then have them show us the capability study on this dimension. This makes sure when the part is first quoted that they do not miss these key features that makes the part function.

 

[KENAT]

So SDETERS, they don't have to meet the other tolerances on your drawings?

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[looslib]

We tried to open up some tolerances on glass molds one time, saying it would allow us (the machine shop) to give the glass plant a mold that would still meet their spec, but at a lower cost. They rejected the marked up print and said 'This is what we need'.
The machine shop I worked in was owned by the glass plant corporation and our only customers were the glass plants of the parent corporation. So much for trying to saving the company money.


"Wildfires are dangerous, hard to control, and economically catastrophic."

Ben Loosli

 

[bpcash]

I guess what I interests me most is what other people are doing to document their design decisions internally, rather than how to interact with our suppliers. If we are designing parts that can't be made, that is a separate issue, IMO.

One idea that has been put forth is to require a written justification of tolerances lower than a certain amount, to be approved by a manager. Not sure what that would be, though.

 

[KENAT]

bpcash - one problem with that is it doesn't really take the fundamental process capabilities into account. While one process may easily hit +-.002, another process may not.

So you'd need to make allowance for the process to be used, however often there is more than one way to skin a cat and generally designers avoid specifying processes unless it materially effects end function.

Like I mention above, I've shared tolerance studies, sometimes formalized in spreadsheets or even reports when necessary. Though I prefer not to spend my time doing that.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[SteveMartin]

We don't generally document our choice of tolerances - not in an official or standardized way at least. Chosing to use an 'exotic' material is generally driven out of a component design review or FMEA activity, both of which have standard procedures for us and the resultant paperwork is saved in the relevant project folder.

I understand your frustration with the tolerance pushback. I'm 'discussing' proposed tolerance changes (that make the part potentially unmountable) with one of our suppliers right now. Not many people around here outside of engineering want to understand tolerance stackups until something doesn't work. The fact that things at technically too-loose tolerances work (due to luck and/or receiving better-than-spec parts) is often used to attempt to justify looser tolerances elsewhere.

It doesn't help that some of the components that go into the parts we design have uncomfortably loose tolerances (a pain in its own right when designing the mating parts) which make ours look quite small by comparison.

 

[SteveMartin]

One thing I didn't think to mention in my previous post: Our reasoning behind tolerance choice isn't necessarily recorded, but all of our drawings go through an approval process that includes manufacturing and quality representatives (though not purchasing). Part of their review process before signing the paperwork to send the drawing goes out to fabrication *should* include checking that the tolerances are appropriate/acceptable. Unfortunately, I know that is not always the case.

 

[ewh]

I'm at a loss as to how manufacturing and quality can decide what tolerances are appropriate/acceptable for a design. This is an engineering decision, and manufacturing and quality need to determine how/if it can be accomplished.

"Good to know you got shoes to wear when you find the floor." - [small]Robert Hunter[/small]

 

[MikeHalloran]

For most of my career, eyeballed tolerances based on manufacturing process capability were sufficient because of small quantities or tolerance-insensitive design, and/or I was 'God' anyway and could reapportion tolerances when someone could justify it to me.

Then I got sucked into cleaning up after someone else's redesign of tiny ball valves, all screw machine parts made in quantity. Based on my testing and analysis, nearly every design feature they changed made the product less satisfactory, but the worst part was the tolerances, which were apparently eyeballed by an old screw machine supervisor. Screws didn't seat properly into holes that were half a thread too shallow, o-ring grooves didn't always have enough volume, stuff like that.

So I set up a spreadsheet for the assembly, and a linked one for every component, with an entry for every stated dimension and its tolerance, and every stackup I could think of, and just beat on it until I could come up with a set of nominals and tolerances that always assembled and always worked. Then I marked up the affected prints and walked through the ECNs to fix the mess.

I also kept a chronological text file containing narratives about everything I knew, suspected, or found out about the product.

Reviewing the text file helped me make a strong argument for the changes with each of the ECN signatories.

The spreadsheet helped with the technical people.

The spreadsheet helped more with the production supervisors. If they made a counter-proposal to loosen a particular tolerance a bit, I could tell them and show them how they would then have to deal with tighter tolerances on other parts.

I would have killed for a software tool to make that job easier, but Excel was all I could find. In retrospect, I don't think such a tool could be made sufficiently flexible to be useful and sufficiently simple to be comprehensible, so Excel is still my tool of choice for such things.

Oh. The production scrap rate went way down and the product worked better.


Mike Halloran
Pembroke Pines, FL, USA

 

[drawoh]

MikeHalloran,

Was the spreadsheet a way of justifying things to management, or was it just your process of analyzing and making decisions? At some point, you need your bosses to have enough confidence in you to accept your judgements.

JHG

 

[MikeHalloran]

The spreadsheet was just a way of making sense of the stackups and interactions. It grew as needed, and was never made fit for presentation or anything like that.

Later, I made a derivative but simpler and better structured sheet for doing First Article Inspections. That was made so that other people could use it with minimal training, but everyone else claimed inability, so I did all the inspections myself. Small outfit.



Mike Halloran
Pembroke Pines, FL, USA

 

[ptruitt]

bpcash...

I sometimes have a chat with management and try to get a handle on the level of risk they are willing to accept. Let them know that with excellent metrology, full or partial design of experiments, process capability studies, you may indeed be able to open up tolerances. Whenever possible, I base my tight tolerance on design guides that I can show management. Ex: The bearing manufacturer might tell me to hold +/-.0002. If management wants something cheaper, ask them if they want to take the risk and ask for a deviation in writing. I don't mind taking some risks, but I try to inform them as much as I can.

Peter Truitt
Minnesota

 

[fcsuper]

In my humble experience, some companies and departments have a culture locked in we-must-be-wrong mode. Someone else says something is wrong with us, so therefore something is worng with us. This may be due to a lack of respect for the responsibilities and skillsets of fellow associates in other departments. The problem with this mode is that it is easy for vendors to work departments or even individuals against one another for their own advantage (i.e., getting a job they normally wouldn't qualify for because of their inability to meet the tight tolerences).

Different vendors have different skillsets. If there is an engineering requirement for a tight tolerance, and one vendor complains about it, then find another vendor how can do it (as long as you know it can be done and is necessary to the design intent).

Matt Lorono, CSWP
Lorono's SolidWorks Resources & SolidWorks Legion
Follow me on Twitter