Minimum Samples for DFM discussion

supplier_quality56 · Jan 13, 2023

Hi All,

This is my first post here. I’m working with a supplier for DFM discussions where they are sampling 5 parts or less and proposing tolerance reliefs for some of the dimensions. My concern is 5 samples is not a statistically significant number to warrant tolerance relief discussions as the 5 samples may potentially represent the extreme ends of the process giving minimum info to accept or reject their proposal.
The supplier is proposing that during capability study they would run 30 samples which would give us enough info to tighten back tolerances that were relaxed initially.
Any thoughts on what would be a good strategy based on your experiences in similar situations?

Bev D · Jan 13, 2023

Specifications/tolerances should be engineered to ensure product functions as required. It is NOT a statistical exercise. You may use DoE to determine whatTolerances are needed to ensure function. Capability of a supplier to meet engineered specs is not grounds for negotiating changes to proper specifications.
The question you must answer is: are the specs correct or not? If not, figure out what the correct spec is. If so, help the supplier to be capable to the specs or find someone who is.

supplier_quality56 · Jan 13, 2023

@Bev D thank you for the response.
Ideally we should have done DOE to determine tolerances. We are at a stage of approving production parts. The engineering team confirmed that they designed very tight tolerances which could be difficult to manufacture but we have determined that there is room to relax them a bit. But my thoughts are, we would need enough data from supplier process to be able to approve some these relaxation requests. This is to ensure the supplier actually trying to engineer better processes to achieve the specific tolerances rather than just ask for relaxation.

Note: Some of the engineers are not well versed with right way of planning tolerances in their design (we are a new team and org)

Ed Panek · Jan 13, 2023

Regarding the supplier's proposal, it is a good practice to measure the capability of the process with a larger sample size before making any decisions about tolerance relief. This will help ensure that the proposed tolerances are appropriate, and will allow you to make more informed decisions about the design of the parts and the process.

Bev D · Jan 14, 2023

A few things to remember:

30 samples can be representative of actual capability IF they are randomly samples from a set of part made across the full range of variation of the inputs (and thus the outputs) of the process. This is rarely done in the preproduction phases.
If a biased sample (anything other than above) doesn't exhibit capability - the process isn't capable. Although you won't know what percentage may be out of spec until 1 above is done...
However the opposite is NOT true: a biased sample that is all in spec doesn't mean that the process is capable. If you haven't sampled from the extremes of the allowed inputs. It can tell you that the sampled input conditions are capable but that is usually a very restrictive input range.

supplier_quality56 · Jan 14, 2023

Bev D said:
A few things to remember:

30 samples can be representative of actual capability IF they are randomly samples from a set of part made across the full range of variation of the inputs (and thus the outputs) of the process. This is rarely done in the preproduction phases.

If a biased sample (anything other than above) doesn't exhibit capability - the process isn't capable. Although you won't know what percentage may be out of spec until 1 above is done...

However the opposite is NOT true: a biased sample that is all in spec doesn't mean that the process is capable. If you haven't sampled from the extremes of the allowed inputs. It can tell you that the sampled input conditions are capable but that is usually a very restrictive input range.

Thank you for your input. I’ll keep this in mind when writing SOPs to ensure supplier processes are properly tested across their full range to assess capability.

I’d want to know how DFM is handled when in developmental phase for supplies products.

Bev D · Jan 14, 2023

supplier_quality56 said:
I’d want to know how DFM is handled when in developmental phase for supplies products.

The same way it is for your manufacturing site. SCIENTIFIC and empirical knowledge of the basic material, process and equipment capability along with DoE to to determine the input parameter specifications for teh specified output characteristic tolerances…there is no shortcut. Physics is a b…ch

supplier_quality56 · Jan 14, 2023

Bev D said:
The same way it is for your manufacturing site. SCIENTIFIC and empirical knowledge of the basic material, process and equipment capability along with DoE to to determine the input parameter specifications for teh specified output characteristic tolerances…there is no shortcut. Physics is a b…ch

I have some DOE knowledge from my grad courses but have not had to use them till now. I need to figure out a way to bake this process into our procedures to set our developmental projects for success.

Wish there were example procedures I could take a look at to see how this has been done at other orgs.

Jim Wynne · Jan 14, 2023

In addition to all of the good technical advice that's been given, it seems to me that there's a glaring weakness here, that being the failure of the design engineers to make a conscientious effort to set tolerances at what they need to be. It's their job. Job shops will, if it's to their advantage, be dishonest about their capabilities just to get the work. This a solid "given" and should be anticipated. If the tolerances are what they need to be, there's no room for negotiation. If you have incompetent design engineers, it shouldn't be your responsibility to educate them.

Early in the development stage, design engineers should be working directly with suppliers (or potential suppliers). The more people that are added to the mix, the more likely that things like this will happen.

supplier_quality56 · Jan 14, 2023

Jim Wynne said:
In addition to all of the good technical advice that's been given, it seems to me that there's a glaring weakness here, that being the failure of the design engineers to make a conscientious effort to set tolerances at what they need to be. It's their job. Job shops will, if it's to their advantage, be dishonest about their capabilities just to get the work. This a solid "given" and should be anticipated. If the tolerances are what they need to be, there's no room for negotiation. If you have incompetent design engineers, it shouldn't be your responsibility to educate them.

Early in the development stage, design engineers should be working directly with suppliers (or potential suppliers). The more people that are added to the mix, the more likely that things like this will happen.

Design engineers have been working with suppliers but they do feel that their design spec tolerances were quite tight. They wanted to see if suppliers could achieve them and if not then maybe provide relief were unachievable.
With a new design team I felt this was a lessons learned for all. But I want to create a process flow SOP to prevent this for future products.

Minimum Samples for DFM discussion

supplier_quality56

Registered

Bev D

Heretical Statistician

supplier_quality56

Registered

Ed Panek

QA RA Small Med Dev Company

Bev D

Heretical Statistician

supplier_quality56

Registered

Bev D

Heretical Statistician

supplier_quality56

Registered

Jim Wynne

supplier_quality56

Registered

Similar threads