Dear folks,

I am a new fish here. :) I have a little question below, hope you can give some comments.

How to setup a control limits of SPC chart, e.g. Xbar-R chart, or C-chart based on historical data? and when is the right time to tighten the control limits of SPC to get the process continuous improved?:confused:

This will help you to understand some basics

http://www.itl.nist.gov/div898/handbook/pmc/section3/pmc31.htm

sathis :bigwave:

This will help you to understand some basics

http://www.itl.nist.gov/div898/handbook/pmc/section3/pmc31.htm

sathis :bigwave:
:thanx: Thanks, Sathis! This do give me a basic idea! So, what you mean is we should periodically calculate the Sigma and get the control limits(3 sigma) less and less. Is that right? But if we want to shift from 3 sigma to a 6 sigma process, do I need set up the control limit at 6 sigma ? :nope:

Here is one more link " Calculation of control of limits"

www.qualityadvisor.com/library/control_charts/recalculate.htm

sathis :)

:thanx: Thanks, Sathis! This do give me a basic idea! So, what you mean is we should periodically calculate the Sigma and get the control limits(3 sigma) less and less. Is that right? But if we want to shift from 3 sigma to a 6 sigma process, do I need set up the control limit at 6 sigma ? :nope:

hi Claymanz,
Im not an SPC expert but i know that Control Limits are always set-up at 3-sigma limits. Yes, this is true even if you are shifting to a 6sigma process.
The rationale behind is that a 3-sigma control limits can easily discrimate between assignable variation and random variation.
The term "Six-sigma Process" can be misleading for those unfamiliar yet with six-sigma. Six-sigma process does not mean that you will set-up your control limits at "mean +/- 6*sigma". doing this set-up will surely hide variation in your process and will not held you in your goal to reduce variation.
A Six Sigma Process means that your process variation is too small such that you Specification Limits are too far from the mean. But even at his situation, Control Limits of control charts are set-up using "mean +/- 3*sigma"

Hope that helps somehow in anwering you question:-)

Normally, I will use the Pre Control Chart which will help me to detect the shift in process parameters that have been statistically significant. As I believed, the pre control charts can only used when the process is within statistical control under X bar and R charts monitoring and is capable. After 6 months monitoring the average and range chart which I only satisfied to use the pre control charts are effective enough to control the process.

This will help you to understand some basics

http://www.itl.nist.gov/div898/handbook/pmc/section3/pmc31.htm

This link is very misleading. It is definitely not the way that Shewhart invented control charts, nor does it follow the thinking of Deming nor Wheeler. The following paper by Dr Don Wheeler gives an explanation.

http://www.spcpress.com/ink_pdfs/Wheeler%20Neave.pdf (http://www.spcpress.com/ink_pdfs/Wheeler%20Neave.pdf)

Normally, I will use the Pre Control Chart ... Yew,
Carrying out process adjustments as dictated by Pre Control Charts can actually increase variation.

But if we want to shift from 3 sigma to a 6 sigma process, do I need set up the control limit at 6 sigma ?
Clamanz,
6 sigma refers to specification limits, or a Cp=2. Process tolerances may be set wherever you wish. Bill Smith, the originator of the term "six sigma" achieved Cp=2 by both widening specification limits and reducing variation.
Control limits are the voice of the process. They are dictated by the process.

Dear folks,

I am a new fish here. :) I have a little question below, hope you can give some comments.

How to setup a control limits of SPC chart, e.g. Xbar-R chart, or C-chart based on historical data? and when is the right time to tighten the control limits of SPC to get the process continuous improved?:confused:

Please see http://www.hanford.gov/safety/vpp/spc.htm for some ideas on SPC. As commented by others, you never "tighten the control limits". The control limits are set by the data, and as Dr. Wheeler states "are the voice of the process". Now, if you improve the process such that you create a statitistically significant decrease in the variation (on a Range, Moving Range, or Sigma chart) then you would rebaseline the chart to reflect the new "voice of the process".

There is a cause and effect issue here - "tightening the control limits" would NOT result in "getting the process continuously improved". However, achieving process improvement would cause you to calculate new control limits, which could be tighter (the standard deviation is now smaller) than the original.