DFMEA Design Control response

[Casana]

Hello all,
One of the engineers in our group is saying that stating "Experimental measurements & experimental measurements" is sufficient as a Design Control for the DFMEA, and he's used this answer for practially all the entries o on the DFMEA.
I asked him for more detail but he feels this is sufficient, says otherwise he'd need to write tons of detail in there. I say there should be a middle ground (ie list of what experimental measurements were done). He says it would still be a long list and he's trying to avoid writing it.
Previously he's said that FMEA's are a useless paperwork exercise(says that its not applicable to the semiconductors industry since we're not making doors or fenders)
Oh, before anyone says anything, yes I know DFMEA's are supposed to be done by a team... but this particular engineer insists on doing the whole thing. I insisted he at least gather the team and review the DFMEA and do the ranking with them.
So, is he right or am I right or neither?
Thanks for any comments!

 

[Al Dyer]

I agree with you! Where did the engineer get his/her training, Acme College of Applied Process Control Maybe his prof was Wyly E. Coyote.

Seriously, all types of FMEA's are critical in identifying processes, their controls, and their areas for continuous improvement.

Can 1 person do the FMEA? Sure, I don't agree but at the very least it should be run through the applicable people in different departments to get an accurate view.

What industry are you in? I really get upset when people say that FMEA/Process documentation is a waste of time and paperwork.

Good Luck!

ASD...

 

[Marc]

-> all types of FMEA's are critical in
-> identifying processes, their controls, and their areas
-> for continuous improvement.

In part I disagree. A design FMEA does not typically address processes as I think you are using the word processes. A DFMEA addresses product function(s) (which I guess one could call a 'process' if rolling up a window, for example, is a process) of the design and related requirements.

-> stating "Experimental measurements & experimental
-> measurements" is sufficient as a Design Control for the
-> DFMEA

This is sorta correct. You have 2 main aspects to the 'normal' design process: Verification and Validation. Verification includes the "..."Experimental measurements..." These experiments are the control. "This is what we predict (hypothesis), do experiement, and results either prove or disprove the hypothesis". Remember - We're not talking about process controls in a DFMEA. We're talking about DESIGN Controls.

-> Previously he's said that FMEA's are a useless paperwork
-> exercise(says that its not applicable to the
-> semiconductors industry since we're not making doors or
-> fenders)

One graciously accepts ignorance on the part of another from time to time as a courtesy if nothing else. We all have opinions on what is and what isn't important.

-> I asked him for more detail but he feels this is
-> sufficient

The auditor will want to talk about experiemnts performed which address each DFMEA line item and will probably want to see the results of the experiements performed. There will have to be documented test plan(s), etc. with defined acceptance criteria and the results are, of course, quality records. If they're not referenced on the DFMEA then its a risk factor in linking a specific line item on the DFMEA to a specifec 'experiement'. However, many companies do put a generic comment like this on their DFMEAs. They keep the links and such in a project folder. To be honest, I wouldn't want to go through an audit without specific links defined somewhere.

-> I insisted he at least gather the team and review the
-> DFMEA and do the ranking with them.

You are correct. Both your company and the engineer are at risk (legally, financially and otherwise) if only 1 person participates in the DFMEA - especially if it is the 'original' FMEA.

I would give your problem engineer a copy of the AIAG's FMEA booklet with a page marker on page 7 - the start of the section on Design FMEAs.

I worked with Motorola's semi-conductor sector a few years back as principle QS-9000 implementation consultant in Guadalajara and then Phoenix (I think they're now called On-Semi). They had a very good DFMEA system well before QS-9000. DFMEAs are undoubtedly an asset to semi-conductor manufacturers.

 

[Al Dyer]

Marc,

I do agree with your post. My response was of general nature trying to take into account the level of knowledge of the poster. In truth, I have seen DFMEA's but in practice my forte is PFMEA's.

I think the AIAG FMEA manual is a good reference to differentiate between the two types.

ASD...

 

[Casana]

Thanks, Al & Marc for your comments.

Al - since you asked, we design & manufacture semiconductor devices.

Turns out I HAD given this engineer a copy of the AIAG FMEA book, but its not that great at explaining how to go about doing the DFMEA. It only served to make him feel justified because its examples are very "nuts & bolts" oriented.

I succeeded in getting him to be slightly more descriptive in his responses (at least he now references specific design models that have been done in our dept). It probably doesn't help that previous management thought QS was only good for getting our foot in the automtive industry door. Current management thinks differently (or I wouldn't be here!) but it will take a while to change the culture...
Thanks for the support!

[This message has been edited by Casana (edited 20 July 2001).]