Supplier problem solving

[Graciel]

Hello!
I have one situation, where I'm asking your opinion and experience about it:
We had a quality problem detected by the customer recently. On the next day, we found one part with same fail mode during the assembly (our production line). We analysed the components and one of them was suspicious by the appearance aspect. We called the supplier, they took 3 samples to do an engineering analysis on them. So....today they delivered the report and according to it, the 3 samples are approved.
But...for our customer, we had already sent a presentation mentioning this steps...and that this component was being analysed.
So now...what could the next steps be? I was thinking about that what further tests could be done to the samples.
Have you guys been in a situation where the most suspicious cause of a fail was later discarded as the cause? And did you guys do?
Thank you.

 

[Jen Kirley]

You have one clear choice, at least:

a) Believe the supplier and wait to see if there is another failure.
b) Have the part analyzed by someone else.

A question (sorry if this is dense): did they analyze three parts from the same lot as this one came from?

Second dense question: did they analyze the failed part?

 

[ChrisM]

You mention "fail mode". Did the supplier disagree with your customer that this was not, in fact a "fail"? Does your organization agree that it is a "fail" or not? If you cannot come to the same decision as your supplier, there is definitely an issue in defining what is an acceptable feature of the item in question and you will need to work together to reach agreement, then incorporate it into the specification, and goods-in inspection criteria

 

[Graciel]

You have one clear choice, at least:

a) Believe the supplier and wait to see if there is another failure.
b) Have the part analyzed by someone else.

A question (sorry if this is dense): did they analyze three parts from the same lot as this one came from?

Second dense question: did they analyze the failed part?

Hi!
About the questions: the parts 3 analyzed are by visual criteria not approved. They were sort out from our warehouse after we detected same failure mode during our assembly process. And they were not from same lot.
The failed part (detected by the customer) is still at the customer. They are about to return it to us but still didn't.

 

[Graciel]

You mention "fail mode". Did the supplier disagree with your customer that this was not, in fact a "fail"? Does your organization agree that it is a "fail" or not? If you cannot come to the same decision as your supplier, there is definitely an issue in defining what is an acceptable feature of the item in question and you will need to work together to reach agreement, then incorporate it into the specification, and goods-in inspection criteria

Hi! The supplier and us agree that is a failure but...the supplier, since first day we notified them, already disagree that their component had the suspicious defect we mentioned to them . It was a visual criteria but...they said their engineering could analyse the material (it is a fabric/textil part) and check if it was ok. They came back like 4 days after saying that, to do more profound analysis it would be necessary to a bigger part of the fabric, that the part we use actually doesn't give to much data.
So, the data they came back with the results, is not even complete.
Since we are talking about fabric/textil part, my suggestion to them is to check they still have the original material (fabric roll) that they used to produce our parts. But I believe the answer will be "no".

 

[ChrisM]

You don't need an engineer to analyse a "visual failure", neither do you need a further large sample of the original material.
When this episode is over, you need to keep the "piece" in question and use it as one of many samples that are used for visual comparisons to determine what is, and is not, acceptable. It may be possible to provide a photographic guide too.

 

[Graciel]

You don't need an engineer to analyse a "visual failure", neither do you need a further large sample of the original material.
When this episode is over, you need to keep the "piece" in question and use it as one of many samples that are used for visual comparisons to determine what is, and is not, acceptable. It may be possible to provide a photographic guide too.

For this case, the visual characteristic we are analysing means something may be wrong with the "fabric" composition (I'm talking about a textil part). Further analysis is required to check it.

 

[Tidge]

This bit of history may not help this circumstance, but if anyone is looking for an example of an unresponsive supplier brushing off a customer, here goes. TLDR: We removed them as an approved supplier.

We use a very simple, common electrical component on many of our printed-circuit board assemblies. On several of our devices, this component started failing in a catastrophic (but not dramatic) manner. It was really easy to demonstrate the failure (think: using a digital multimeter). We were able to isolate an entire lot (and factory of origin) for the offending component.

Our concerns were brushed off by the manufacturer; even though they stood by the lot and agreed they had made/shipped the parts. An x-ray analysis showed that the components from that factory didn't have an actual die inside them! They were about as fake as fake can be! Per the manufacturers own datasheets and press notices, these parts came from a new factory (operations had just moved from Japan to China) and it was pretty clear that this manufacturer had some serious issues with the new factory. That manufacturer simply didn't care about our concerns, and refused to even make a cursory effort to 'play along' with our supplier quality team.

 

[John Predmore]

Have you guys been in a situation where the most suspicious cause of a fail was later discarded as the cause?

Definitely had this experience. This is one hallmark which defines a tough-to-solve problem. There are other categories of tough problems. Albert Einstein famously said, "The significant problems we face cannot be solved with the same thinking we used when we created the problem." If our engineers and experts were as smart as they think they are, we probably wouldn't have as many tough problems to start with. So listen to advice from subject matter experts with a grain of salt. Verify every fact you are told with observational data and deploy a high level of critical thinking.

You asked what did we do [in such situations]. I subscribe to Red X methodology. The most fruitful course of action is to compare parts with the problem you seek to understand, to parts without the problem.

This statement leads to a few profound corollary statements: 1) you have to describe (and eventually identify) the problem you are looking for, 2) you have to be able to sort parts with the problem from parts without, and 3) even more powerfully, sort degrees of the problem so to sort worst from merely bad. 1, 2 and 3 are not easy to do, but with clear thinking and logical problem-solving approach, it is possible to untangle the Gordian knot. With a suitable measurement, you select part(s) with the worst degree of the problem and parts from the other extreme. The other extreme doesn't mean good parts, it means parts from the same batch, alike in every way except the difference which matters, with the characteristic of focus at the other extreme. By comparing worst part(s) to the other extreme, you have the best chance of seeing the difference which matters.

You typically start with the assembly, and quickly move your focus to the component or the way the parts are assembled, and eventually move to a characteristic which matters, in the component or the process. There may be little to gain to compare random parts, or compare parts to the design specification, unless there is a proven correlation of the selected part(s) to the problem you care about. In this way, you cast aside theories and explanations that don't bear fruit, and make rapid progress to the factor(s) which matter. Working with a supplier or remote factory site, you definitely want your best problem-solver there in-person to see with his own eyes, and not be lulled by fancy talk and spreadsheets.