Printed Circuit Board Assembly & PCB Design Forum

SMT electronics assembly manufacturing forum.

# Accuracy of placement +/-100um @ 5 sigma

tvd911

Accuracy of placement +/-100um @ 5 sigma | 19 November, 2003

Hi everyone, is someone in this forum familiar with the accuracy of placement. What is that number tell us. Appreciate for your attention.

Accuracy of placement +/-100um @ 5 sigma | 20 November, 2003

Well, let me take a shot...

+/-100um = 0.1mm 5 sigma = is the number of acceptable defects (or placements outside of that accuracy (or tolerance).

I believe 5 sigma is 233 defects out of 1 million, but I could be wrong on that.

So basically all placements will be within +/- 0.1mm (or less) of center and out of 1 million placements there maybe up to 233 placements that are placed outside of the +/- 0.1mm accuracy.

I hope that helps.

Accuracy of placement +/-100um @ 5 sigma | 20 November, 2003

Oh, and we boil 5-sigma down to 2 out of 10,000 placements as a starting point. I don't know of a single machine that can do 1 million placements in one program. 10K is doable.

If we fail there then we start looking at the bigger picture; either run the test a few times and average the results, or start calibrating the machine or feeders, checkout the quality of the parts, and such.

Vincent

Accuracy of placement +/-100um @ 5 sigma | 21 November, 2003

Thanks for your reply. I'm wondering if We have 2 machines, one can do +/-100um @ 3sigma and the other is +/-100 @ 4sigma. And how U come up the solution. thanks

Vincent

Accuracy of placement +/-100um @ 5 sigma | 21 November, 2003

Thanks for your reply. I'm wondering if We have 2 machines, one can do +/-100um @ 3sigma and the other is +/-100 @ 4sigma. And how U come up the solution. thanks

Process Engineer

Accuracy of placement +/-100um @ 5 sigma | 24 November, 2003

I am not sure where Frank got his numbers but here is what Sigma really means.

1 Sigma is one Standard Deviation, 2 Sigma is two Standard Deviations...(anyone take a Stats class in college?)

I am not going to bore you with all the stat-mumbo-gumbo: http://mathworld.wolfram.com/StandardDeviation.html

To borrow a phrase "It boils down to..." 1 Sigma = 68.27% 2 Sigma = 95.45% 3 Sigma = 99.73% 4 Sigma = 99.99% 5 Sigma = 99.9999%

What this means is that if a PnP machine is rated at 1 Sigma, then 68.27% of all placements will be within the required tolerance. Then that means 31.73% will be outside of that spec. 1-Sigma is not that good.

Most PnP machines that I have seen are rated at 3-Sigma and a few at 4-Sigma. I haven't seen one rated at 5-Sigma, that would be quite impressive. Although, I think some screen printers are rated at 5-Sigma.

So, Vincent, use "all placements are to be within +/- 100um at 99.73% (3-Sigma) and/or 99.99% (4-Sigma) of the time" as your goals.

tvd911

Accuracy of placement +/-100um @ 5 sigma | 25 November, 2003

Thanks all U guy for input. This helps.

Accuracy of placement +/-100um @ 5 sigma | 1 December, 2003

I see where I went wrong.

The numbers provided by "Process Engineer" are correct.

My company is trying to adhere to the Six Sigma code of business and the managament here falsely assumed that all of our equipment can do that as well. After double checking with my PnP supplier, I can finally stop pulling my hair out trying to get unrealistic results.

Overall the end product can follow the Six Sigma rules, but my process needs to be within the specs that the machines were manufactured to do.