MSA for Attribute data - Automatic Hy-pot test machine

D

Donald Duck

Hi folks,

After read all the threads for ATE MSA. I still have a trouble in my case.

We have an automatic electrical test equipment for both hi-pot and continuity test. It automatically sort the pass and fail to different bins. One day my boss' boss just took a part from failure bin and retest it ---- it passed! The result is: I am required to make an effective MSA study for this equipment.

I am think I should assess the Repeatability and bias, or linearity? Any suggestions? :thanx:
 
A

Atul Khandekar

Anyone with experience with this type of a situation?
 
C

chalapathi

In this case measurement system is doing go/no-go inspection and hence Attribute MSA study is applicable. Take 50 parts (some good and some rejected and repeat 5/6 times and estimate the Effectivess, Missrate, etc.
In the MSA 3'rd edition, it talks about 3 operators. In this case operator has no role. You can treat this as one operator and do the calculations.
 

Marc

Fully vaccinated are you?
Leader
Thanks for the help on this old one, chalapathi. We appreciate it!
 
D

Dave Dunn

Donald Duck said:
Hi folks,

After read all the threads for ATE MSA. I still have a trouble in my case.

We have an automatic electrical test equipment for both hi-pot and continuity test. It automatically sort the pass and fail to different bins. One day my boss' boss just took a part from failure bin and retest it ---- it passed! The result is: I am required to make an effective MSA study for this equipment.

I am think I should assess the Repeatability and bias, or linearity? Any suggestions? :thanx:

In addition to the MSA testing, it might be a good idea to set up periodic pass/failure testing with known samples to verify that there is no malfunction with the equipment. An MSA study might give you a basis to understand the reliability of the test, but might not catch the root cause of problems with the tester.

One question: was the part that passed after retesting actually defective, or actually acceptable? (assuming that a defect could be visually verified or tested in some other manner) Many things could cause false positives or negatives: sticky spring-loaded pins in the tester; metal shavings bridging part terminals, etc. We have similar testing on a part using inch-long metal inserts that are pressed through holes in the parts that need to be tested for presence, alignment, and damage (continuity) as well as for electrical bridging between each terminal (hi-pot).
 
E

Ederie - 2007

Gauge R&R

I agree with Gauge R&R senerio from Chalapathi, If that proves unacceptable, (failed) I think you will know more about this process than anyone in the shop!
 

Miner

Forum Moderator
Leader
Admin
This is a good example of not accepting attribute data at face value. While most hipot tests are performed as a pass/fail test, there is usually variable data to be found and used.

The hipot testers that I have experience with will provide a variable leakage current value. However, most users will set the tester to alarm at a threshold value (max leakage current) and never pay attention to the actual value. This leads them to treat it as attribute data, but the variable data is still available for a variable MSA study. Creation of a gage performance curve would show the zone that accepts bad/rejects good product and a guard band could be established to prevent acceptance of bad product.
 
Top Bottom