PPM Prediction for a New Automotive Electronic Product

[Heintje]

Hello,
Thank you for reading this post.

I have following question;

We are developing a new automotive electronic product. (so we are still in development phase)
A potential customer now wants us to commit to certain PPM values for this upcoming product. They want us to commit for a certain max PPM values they have issued for eg. number of failures once the product is used in an automotive environment.

I'm not experienced in this matter and wonder how to tackle this. Since it is a new product, we have no data what so ever available. Also no historical data available.
How can we commit already to certain given PPM values? What is your advice in this?

Kind regards,
Heintje.

 

[Jim Wynne]

Re: PPM prediction,

You should tell your potential customer that reliability (not PPM) predictions are based on data, not speculation, and you have no data at present. You should also ask them to specify the conditions under which they expect the product will be expected to perform. Just saying that the thing will be used under "automotive conditions" isn't enough.

 

[Heintje]

Re: PPM prediction,

Ok, Thank you for your reply,
understand what you are saying.

However, it is possible to do all kind of things to increase reliability, like good design, extensive testing, Halt or Hass testing, controlled production and so on, we can create good reliable designs, but still, I don't know how this can be translated to PPM values... Is their a way to convert design into PPM values?

If my customer says, your design may not go above 500PPM on a year base, how can I prove this PPM value at the start and delivery of the first product?

Greetings.

 

[Marc]

<snip> A potential customer now wants us to commit to certain PPM values for this upcoming product. They want us to commit for a certain max PPM values they have issued for eg. number of failures once the product is used in an automotive environment.

I'm not experienced in this matter and wonder how to tackle this. Since it is a new product, we have no data what so ever available. Also no historical data available. <snip>

Key phrase: "...number of failures once the product is used in an automotive environment...."

Such a prediction would have to come from reliability testing (e.g.: thermal cycling {and possibly ESS (environmental stress screening)}, vibration, shock, temperature/humidity testing, etc.).

I used to write what were called Environmental Design Criteria Test Plans for high reliability aerospace electronics back in the 1980's, which is how I ended up in automotive way back when helping Ford with their car electronics (the early automotive "computers", audio electronics, etc.).

NOTE: The testing can be long and expensive and the plan should be designed by someone who is well versed in reliability. See Jim's post above.

 

[Jim Wynne]

Re: PPM prediction,

Ok, Thank you for your reply,
understand what you are saying.

However, it is possible to do all kind of things to increase reliability, like good design, extensive testing, Halt or Hass testing, controlled production and so on, we can create good reliable designs, but still, I don't know how this can be translated to PPM values... Is their a way to convert design into PPM values?

If my customer says, your design may not go above 500PPM on a year base, how can I prove this PPM value at the start and delivery of the first product?

Greetings.

What your customer is asking for can't be reasonably done. You shouldn't guarantee any PPM levels without data.

 

[Marc]

Re: PPM prediction,

<snip> You shouldn't guarantee any PPAP levels without data.

I.e.: Reliability testing since it's a new product and I *assume* they don't have a similar product from which to extrapolate data.

Back in the 1980's and 1990's there were some pretty standard automotive environmental reliability test scenarios which were designed from a lot of testing of vehicles on test tracks and actual data from road trips and such. It's been a long time but my bet is I'm sure it's much the same - Certain accelerated life tests, vibration test profiles, and such. That's why I suggest a certified reliability engineer be involved. Someone current in the field.

And Jim is right - The customer has to either provide you the test scenario/requirements or you should make them agree to your test plan should you write one.

For the heck of it I have attached some old graphics I used in some test plans in one way or another around 1988. They're not very good, but they were done in something MacPaint on a Mac. They may not help you but note aspects such as thermal profiles, vibration profiles and such.

As Jim says, you need data. Back when I was in reliability I got data for aerospace products from the DOD, which got it's data from the aircraft manufacturer which they got during their design and development process. In the automotive field I got data from the manufacturers.

An up-to-date automotive reliability engineer (even as a short term consultant) many be what you need.

 

[Bev D]

I think there is a difference between agreeing to a ppm level and knowing whether or not your design and processes are capable of meetign that level.

As a customer I alwys have ppm or quality adn reliability expectations for my suppliers. if they are not comfortable agreeign to them I will get a different supplier. Of ocurse this requiers (as Jim said) an understanding of the use conditions and solid specifications for function, properties and geometries...

 

[DarrellH]

Take a look at IPC-SM-785,
if you look at table 1 it gives data for passenger compartment & underhood environments.
You should insist on customer specification for environment as an absolute minimum.