ATE (Automatic Test Equipment) "Gage" process for electronic modules?

I

IBATE

Newbie here. I am responsible for developing/maintaining ATE's for a number of complex electronic modules for product validation testing - not manufacturing. Little or no operator variation.

1) Is a gage R&R appropriate or would the process better be described as a Capability Analysis? ANOVA OK for this?

2) I cannot produce samples that consistently fall into lower/upper limits of the spec. range(s) due to cost, etc. How do I get around this?

3) Anyone have a process defined for this situation?

4) FMEA recommended to insure all out of spec. values are captured?

Appreciate any comments - I can't be the only one wrestling with this one.
 
S

Southern Girl

Can't address all of your questions, but I may have an answer for #2. I had the same problem with a test stand and trying to create samples that ran the spec range. What I did was to "fake out" the software by hooking my pressure calibrator to the transducer and applying whatever pressure I needed to test the software. The biggest thing we were trying to do was to make sure that the software would reject parts that fell outside the process parameters. We verified that the test stand, not only passed an R&R, but also would reject non-conforming parts. This worked great and our registrar (DNV) auditors loved it.


Since I don't know what you are testing, look to see if there is some way you can "generate" the parameters (reliably of course) you are testing for and see if that works.

Good Luck
 

Jim Wynne

Leader
Admin
IBATE said:
Newbie here. I am responsible for developing/maintaining ATE's for a number of complex electronic modules for product validation testing - not manufacturing. Little or no operator variation.

Welcome to the Cove :bigwave:

IBATE said:
Is a gage R&R appropriate or would the process better be described as a Capability Analysis? ANOVA OK for this?

It depends on what you're trying to find out. GR&R and capability analysis aren't either/or; tell us a little more about what you want to do.

IBATE said:
I cannot produce samples that consistently fall into lower/upper limits of the spec. range(s) due to cost, etc. How do I get around this?

If you want to do GR&R you need values (parts) that reflect the operating range of the process, not necessarily values at or near the spec limits. On the other hand, if you just want to find out how well the test will detect out-of-spec conditions, there methods other than GR&R that might be appropriate.

IBATE said:
FMEA recommended to insure all out of spec. values are captured?

Again, it's hard to say without knowing more about what you're doing. Are you recording/analyzing actual values (variables) or just pass/fail data? If you haven't already, you might want to avail yourself of the AIAG MSA (Measurement System Analysis) manual, which is available here.
 

Miner

Forum Moderator
Leader
Admin
IBATE said:
Newbie here. I am responsible for developing/maintaining ATE's for a number of complex electronic modules for product validation testing - not manufacturing. Little or no operator variation.

1) Is a gage R&R appropriate or would the process better be described as a Capability Analysis? ANOVA OK for this?

2) I cannot produce samples that consistently fall into lower/upper limits of the spec. range(s) due to cost, etc. How do I get around this?

3) Anyone have a process defined for this situation?

4) FMEA recommended to insure all out of spec. values are captured?

Appreciate any comments - I can't be the only one wrestling with this one.

1) Do you have multiple ATEs performing the same measurement? If you do, treat the ATE as an Operator in the Gage R&R study. If you do not, I would recommend performing a stability study per the AIAG MSA 3rd edition manual. The method shown in the MSA manual allows you to calculate a repeatability value from the same data.

2) This is no longer a requirement. The emphasis has changed to using samples that represent the full range of PROCESS variation. Therefore the standard deviation of the samples should be equivalent to the standard deviation obtained with a capability study.

3) No. Sorry.

4) Please clarify your question.
 
I

IBATE

More specific info.....

Thanks for all the feedback.
For this application, we have a single ATE that is set up to test a number of somewhat similar products. What the ATE is doing is simulating "events" and reading back both variable data and CAN message responses. There are literally hundreds of data points/messages in single functional test (both variable and pass/fail). I realize that ideally we should simulate the many possible failing responses/conditions but this in not practical due to sheer number of parameters we measure. We have a specific calibration "standard" that we use on this system. Would calibration alone be sufficient along with with a stability study as described in the MSA manual?
Also, we are not really interested in manufacturing process capability as we are interested in the capability of our test system (again, we are part of R&D here).
 

Miner

Forum Moderator
Leader
Admin
IBATE said:
Thanks for all the feedback.
For this application, we have a single ATE that is set up to test a number of somewhat similar products. What the ATE is doing is simulating "events" and reading back both variable data and CAN message responses. There are literally hundreds of data points/messages in single functional test (both variable and pass/fail). I realize that ideally we should simulate the many possible failing responses/conditions but this in not practical due to sheer number of parameters we measure. We have a specific calibration "standard" that we use on this system. Would calibration alone be sufficient along with with a stability study as described in the MSA manual?
Also, we are not really interested in manufacturing process capability as we are interested in the capability of our test system (again, we are part of R&D here).

Unless you have a customer demanding something specific, I believe that calibration and a stability study (with the associated repeatability) should be sufficient.
 
B

briggs_joe

I agree with Miner on this. A calibration and stability study should be sufficient for a test of a single ATE system. Make sure you are calibrating the ATE as an entire system, and not just the individual piece components alone.

I work in a medical device production house where our test engineering dept produces tests for everything from wafers to final assemblies, and we are only now beginning to see and understand the value that a good MSA will bring.

The biggest problem is identifying which measurements are critical for performing an MSA on in order to limit the amount of work you need to do and the amount of money you need to spend on DUTs. Attempting to achieve high orders of accuracy on every measurement you make is usually not justified (do you really need to perform an MSA just to make sure that you're making good contact with the DUT for low voltage/low current measurements; or if a lower level assembly tests is your "accuracy" test, does the higher level assembly test need to be just as "accurate", or only confirm that the assembly processes didn't change DUT behavior, etc). The FMEA will help you identify these critical measurements.
 
J

jeffrey_Chang

Newbie here. I am responsible for developing/maintaining ATE's for a number of complex electronic modules for product validation testing - not manufacturing. Little or no operator variation.

1) Is a gage R&R appropriate or would the process better be described as a Capability Analysis? ANOVA OK for this?

2) I cannot produce samples that consistently fall into lower/upper limits of the spec. range(s) due to cost, etc. How do I get around this?

3) Anyone have a process defined for this situation?

4) FMEA recommended to insure all out of spec. values are captured?

Appreciate any comments - I can't be the only one wrestling with this one.

Hi,
As a provider of microelectronic assembly and test services, we do perform GRR for our ATE. I believed we are facing the same situation here.

1) Yes, I would think that GRR is applicable. Under such condition, you only perform gage repeatability for that equipment as there'll not be any reproducibility error.

2) Do you have any golden units, you can use it for the study? Our golden units is also use for tester correlation purposes. You might want to consider doing a tester correlation study.

3) Yes, to summarize it briefly, the procedure states the following
- Determine the number of samples for the study, In our case we use the number of test sites per tester to derive the no. of samples to use.
- Determine the no. of trials
- Determine the test parameters to use, what is considered most important to your application. In our case, the test-suite to be used for the measurement is determine by our customers and it covers such parameters as Continuity DC test check, Leakage Test Measurement, and IDDQ Test Measurement and others.
- The calculation to be provided will be GR&R report, BIAS Report, Linearity Report and Attribute Gage Study Report.
- The rest of the procedure are basically administrative contents.

Hope this help. I'll try to post an example of the GRR done for one of our ATE soon.

thks.
jeffrey.
 
A

anurodh

Hi,
I want to understand how can we do Gauge R&R for variable data.

My problem is that Equipment on which i want to perform R&R study has no affect of appraiser on it, when measured on a single equipment

However there is a huge variation from equipment to equipment of same make & software version in terms of output signal.

But again it is not due to appraiser.

So how will Appraiser variation be addressed during Variable R&R study while calculating Appraiser Variation or while calculating X bar Difference.

Also, i;m not able to get D2* Table

Can you please help
 
M

maybie

Hi all,
I am a mechanical student and now undergoing my Final Year Project regarding on an Industrial Automated Test Equipment for product validation testing (current drawn from a vacuum cleaner product). Along the way, definitely I need to perform gauge R&R study to see how much the contribution of variation made by the measuring system. Providing that there is only a single equipment and little involvement of operator (mounting the part on the measurement equipment platform and the product handle to the robotic arm of the equipment ).

My concern here, assuming that the result depict gage R&R result with % Study variation and % tolerance below 10% as per MSA manual standard, should I need to improve the test variables even knowing that those variables are not giving significant effect on the variation? What should I do next? Do I need to improve the test method to control the test process so that the result for upcoming test will remain reliable just like the previous test? or should I just proceed to Design for Experiment (DOE)? Apology if my question sounds strange to you. But indeedly I really need your help.
Any response from anybody are greatly appreciated.

Thanks in Advance.
 
Top Bottom