Printed Circuit Board Assembly & PCB Design Forum

SMT electronics assembly manufacturing forum.

# Capability Process CPk

ricardof

Capability Process CPk | 8 January, 2003

I would like to know how to implement CPk study on SMT line At this time we have the application to calculate CPk on Chip Placers for X and Y axes repeatability, but I'm trying to find what should I calculate on my Oven and paste printer in a simple way.

Jon Fox

Capability Process CPk | 8 January, 2003

1. Do you have the necessary equipment to take on such a task and 2. Why? My only guess is that you are looking to find the probability or likelihood of a particular observation and / or a "confidence" score on your other equipment. This is just a measure of the likelihood that a particular distribution beyound your standard deviation will occur at a certain stage of the process. For your screen printers, you would have to measure your paste alignment and thickness and know your high anf low limit of acceptability. For your reflow, (this won't be easy), you would need an AOI pre-reflow and an AOI post-reflow and be able to compare X, Y, and theta for component placement. Crunch the numbers much in the same way you would your pick and place only instead of an actual measurement, you will be doing a comparison. I don't see doing it at this level can really do anything for you as you are battling the forces of surface tension and thermodynamics. There are books written on this stuff already (check out the SMT magazine article on tombstoning, WOW!) The one useful result would be determining the probability that a certain part will "fix" itself in reflow based on its initial and post-reflow X, Y and theta position. I used to calculate standard deviations and Cpk values on a daily basis on the testing of new pick and place equipment at my last job, I just find it hard to see a use for it at the printing and reflow stage. Just my opionion, and you probably know how the saying goes. If you need some help with the mathematics of it all, let me know. I have a few Excel macros for doing almost everything automatically that I would be willing to share.

ricardof

Capability Process CPk | 16 January, 2003

Thank you Jon I'm very interested in finding some more info on what to get and measure on the Paste Printer and the Oven (DEKs and Vitronics)

MA/NY DDave

Capability Process CPk | 16 January, 2003

Hi

Well I don't agree with Mr Fox.

Cp and Cpk i.e Capabilitiy studies always yield good results and add vigor to the engineer and the process.

You might want to hire a consultant or go to some classes to make sure you do this correctly for your situation.

The Vitronics people should also be helpful (I know at least one here in MA who worked at V then BTU then back again a V) and the DEK people should also be helpful.

Good Luck, Cp studies are a great process engineering education.

YiE, MA/NY DDave

This message was posted the Electronics Forum @

iman

Capability Process CPk | 17 January, 2003

A) Paste Printer :

A1) identify the critical SMDs' pads. (choice should go to the BGA, CSP, MLF types...RES,CAP last in line...)

A2) get a Paste measurement machine that measures readings of paste height (or better - volume) you are paste printing onto the PCB.

A3) implement the Paste measurement machine into the production line, taking readings from the earlier identified critical/choice pads. Plot into a X-bar Y-chart graphical format. From here you can calculate and have fun drawing UCL/LCL, and formulating USL/LSL for the production guys to tip-toe.

A4) make whatever adjustments to the Paste Printing process that gives optimal level UCL/LCL established, to be performed by the Line Technician/Engineer?

B) Reflow Oven :

B1) still banging my head to figure out a good method that gives fair deal to all involved.

B2) appreciate any ideas anyone has done and proven to give a fair deal to all involved.

Brian W.

Capability Process CPk | 17 January, 2003

You can Cpk numbers for the DEK using the QcCalc software. Simply connect a computer to the RS232 port on the DEK and follow the instructions. The program can give you Cp, Cpk numbers for a number of parameters, including X alignment, Y alignement, Theta, pressure, speed etc. In another lifetime we used it periodically to monitor the process and help troubleshoot the printers. QcCalc is available at http://www.prolinksoftware.com.

Hope this helps, Brian W.

MA/NY DDave

Capability Process CPk | 17 January, 2003

Hi

I haven't seen the DEK software you describe in action or studied it, yet I will tell you to be careful of some of the software on the market even from good companies.

Spend the time needed to understand how the software makes the calculations, how your process runs, and your products needs for paste and soldering/reflow.

Some S/W makes a lot of assumptions that may not be appropriate for your paste, your machine, the settings you actually use on your machine, or your product's particular needs.

For example in reflow one product had a dense field in the middle that required more heat to obtain good reflow while the edges easily reflowed. The S/W gave great (unrealistic)results yet the process needed to be tuned to actually have the reflow across the product to be satisfactory each and every time. Only with thermocouples applied on the product appropriately was the process for that product really understood and tuned.

YiE, MA/NY DDave

This message was posted the Electronics Forum @

Capability Process CPk | 4 March, 2003

Hello ricardof, To implement Cpk studies on SMT production equipment, you will require highly accurate glass plates, glass component slugs, accurately manufactured stencils and a measurement system to perform X, Y and Theta positional deviations from a desired set point. To diagnose capability of equipment you must first remove the variability in materials that will mislead you in obtaining accurate Cp and Cpk performance values, hence the above required items. To measure printer Cp/Cpk, you'll need to statistically evaluate two major machine functions: 1)vision alignment, and 2) print process accuracy and repeatability. Since different operating specifications are required in each of the two tasks, they must be run separately. In each test, you'll have to diagnose each functional axis of the machine since each one can have an effect on accuracy and repeatability. This becomes very difficult when testing each axis and collecting data for analysis. To measure ovens, you'll want to pick at least one critical profile attribute which I believe is peak reflow temperature and design a test to measure this over a period of machine cycles. The machine will have to be thermal cycled a number of times which includes nearly full cool down after complete heat up. The test becomes very lengthy (3-4+ hours) as you can imagine. You'll also need a thermal profile tool to collect thermal data for analysis. Let me know if you require additional assistance, we provide measurement services for this type of work. Thanks!

Neil

Capability Process CPk | 7 March, 2003

With any kind of statistical study you must have some limits to which you can measure against. The whole purpose of generating a CPK is to ensure that the equipment under observation can meet these limits when used under normal operating conditions. We have just carried out a capability study of screen printers in our process, and for the purpose of our exercise we set a limit of 0.1mm deviation in both X & Y. We then measured against this and conclude the capability index of each machine. Depending on the level of accuracy you need, you will need to define your limits. Measuring for deviation can be accomplished using a jeweler�s microscope or the placement machine in tech mode and calculate the drift.

Good Luck

This message was posted the Electronics Forum @

Dean

Capability Process CPk | 9 March, 2003

For the reflow process check out KIC THERMAL at kicthermal.com. They make a product called 24/7 which can collects data for process capability. I beta tested the unit about 2 years ago....interesting stuff.

eagle-eyed-one

Capability Process CPk | 10 March, 2003

Hello Neil, I absolutely agree with the conclusion of limits as you have stated. There are 2 sets of limits in which machines can be measured against for compliant performance. First and most critical being the quality performance specification limits set by the manufacturer (ie: +/- 25 micron @ 6 Sigma). If machines cannot operate under these conditions with acceptable results then they won't be able to soundly operate within realistic process specification limits. We must be able to depend on machines operating at or better than designed specification before we can set process specification limits. Only then can we set the appropriate limits based on the process (including technology requirements, ie: 0201's or QFP's) and desired output. For example, EAGLE-EYED ONE was recently contracted to measure the performance of a company's printer, we were able to first measure the machine under its own load (vision alignment accuracy and repeatability) with a quality performance specification of +/- 0.001" @ 6 Sigma as stated by the manufacturer. Once our measurement sytem and software determined it's compliance with acceptable Cp and Cpk values for X, Y and Theta axes, we then looked at establishing process specification limits for print accuracy and repeatability of +/- 0.002" at 6 Sigma. We made numerous prints on a glass plate, using a corrected stencil, and were able to measure and optimize the positional print accuracy of the machine under dynamic print process loads as usually found in production. The printer achieved acceptable Cp and Cpk results all above 2.0. We then provided a comprehensive report certifying its current level of performance as measured against each specification limit. I disagree with your comment on using a jewelers microscope to make deviation measurements. It adds too much subjectivity and variability to the measurement process. Only automation can be designed to accurately, repeatably and reliably perform such measurements. Our capability measurement system uses highly accurate optics and software algorithms to automatically measure deviations and plot them against the spec limits. Let me know if you require additional assistance. Thanks!

This message was posted the Electronics Forum @