Machine Capability Study - Cpk (Cmk) of an R/F testing machine

[qamobile]

Hello! I have been lurking around these forums for the past few days and I have to say, wow, lots of great info!

If this thread is in the wrong location, I apologize.

Ok, so here is my question:

My company uses R/F boxes and shields to verify a cell phones R/F capabilities. However, I am skeptical of the R/F boxes capabilities. The reason why is two fold:

1. The technicians due various things (or tricks) to get the testing boxes to pass a phone during the R/F testing phases. They start the R/F testing and wait until a certain phase of testing and then put the phone in the box and close it. Others put the phone in the box, wait with the box open until a certain phase and then then close it. Yet others have their own "special" tricks to get it to pass. (I know, I know. Please spare me the inconsistancies in this little gem I walked into.)

2. A phone can fail the R/F testing in one box, be repeatable in failing at that testing box and then the phone can be taken to another box and pass with flying colors and vice versa.

Ok, please disregard all the problems we already see with the above statements. Focus mostly on number two.

So, what I would like to start with is testing the R/F boxes capabilities first. I already know that the technicians are having an issue. I get that, I see it, I understand, I know etc. My problem, however, sits at the moment that even when following the R/F box manufacturers testing process, the same phone can vary so much between different R/F boxes of the same model that it can yield a pass/fail between the two. I am wondering if this is why the technicians started finding "other methods" to get these phones to pass the R/F testing phases.

So, my first steps, I think, is to test one phone, with the same jig, battery, R/F sim, etc. n number of times and get the results of the R/F testing output. However, is this correct?

Will this give me a Cpk (Cmk) of the R/F testing machine?
How many times should I test a phone on the machine to get a good look at the variance? (I am thinking 10?)
Afterwards I should do a GR&R between the technicians? On each R/F box?
How would you approach this situation?

Thanks for any replies!

 

[Stijloor]

Re: Machine Capability Study

Welcome to The Cove Forums!

What is an R/F Box?

Stijloor.

 

[qamobile]

Re: Machine Capability Study

Thanks for the welcome!

R/F is Radio Frequency.

 

[Miner]

Re: MSA - R/F Testing of Cell Phones

Wow. You have a lot of issues going on.

Please clarify: Is the R/F box the measurement device, or does it set up the test conditions?

What does closing the box accomplish? Does this shield the device under test or provide a stronger signal?

 

[qamobile]

Re: MSA - R/F Testing of Cell Phones

Wow. You have a lot of issues going on.

I know.

Please clarify: Is the R/F box the measurement device, or does it set up the test conditions?

The R/F box is the measurement device. It tests each phone for various results based off off different cell phone frequencies bands. (e.g. GSM, GPRS, UMTS, CDMA etc.) Even more specific, it is testing these frequency bands for peak power, call loop-back (it calls the phone you accept the call and verify the audio loop-back), rms and peak phase error and frequency errors.

What does closing the box accomplish? Does this shield the device under test or provide a stronger signal?

The shield does both. It shields outside interference and gives a stronger signal inside the box for the testing. It makes a closed controlled environment for the testing. Kinda like if you were standing next to a cell tower as opposed to out in the middle of nowhere. This also assures the testing frequencies that the box is sending/receiving is targeted to that phone and that phone only. If you leave the box open and have another cell phone (of the same make and model) turned on and close to the box, that other phone will also pick up the testing. This will skew the testing results obviously because more than one phone signal is being read.

Hope this answers your questions!

 

[Miner]

I would recommend starting with the basics:

1. Establish a standard test method
2. Obtain variable data if possible, not pass/fail.
3. Verify the calibrations on the R/F boxes.
4. Perform a stability study on each R/F box. This is similar to what you proposed (i.e., one phone, same jig, battery, etc.). Keep measuring the same phone over an extended period of time. Plot on a control chart. This will provide a measure of repeatability and demonstrate whether the measurement process is in control.
5. Measure multiple phones on each R/F box, then plot each phone on a Youden Plot (aka Iso-Plot). You can search this site for Youden to find details, but the x-axis is one R/F box, the y-axis is the other R/F box and each plotted point is one phone. All points should ideally fall on a 45 degree line that intersects zero on both axes. This will help identify differences between R/F boxes.
6. Perform a gage R&R study on one of the R/F boxes using the non-standard test methods.
7. Repeat using the standard test method. Any differences should show up in reproducibility.

This is more involved than is usually necessary, buth you have a thornier problem than most.