Calculation of Upper and Lower Control Limits for Control Chart?

S

samsung

Can anyone please let me know if the upper and lower control limits calculated in the attached sheet are OK and under what circumstances these formulae are applicable?

So far I knew that these limits are Avg +/- 3 sigma. The attached template is from Microsoft.

Your inputs are appreciated.

Thanks
 

Attachments

  • Control chart.xls
    29.5 KB · Views: 441

Steve Prevette

Deming Disciple
Leader
Super Moderator
What it appears is being done is that the data are organized into subgroups of 5. The statistical standard deviation of each subgroup is being calculated, then a grand average of the subgroup standard deviations is calculated. So far, unconventional but might not be too bad. Then that average standard deviation is added and subtracted from the grand average. I don't see a calculation of 3 standard deviations, though you'd have to divide that by the square root of 5, so someone may have figured that was close enough.

Especially if you are just starting in SPC, I would not support using this spreadsheet. Generally, the traditional method is to go with "xbar-R", and that includes two charts - a range chart (which must be in control first) and then a xbar chart.
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
Seems to me, you have an average number of of defects and should be using a U-chart.
 

Attachments

  • printing defect.docx
    148.6 KB · Views: 195
Last edited:
S

samsung

Infact I wish to apply SPC to the data given in the attached sheet. The measurements are taken daily on the average sample (composite day sample). Based on the attached data what would be the most appropriate type of chart that I should be using. Could you also refer a good excel template for the type of chart I'm looking for.

I'm in the process of procuring QI Macros but before it's delivered, I want to learn the basics of SPC.

Thanks.
 

Attachments

  • Data for control chart.xls
    76 KB · Views: 303

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
Infact I wish to apply SPC to the data given in the attached sheet. The measurements are taken daily on the average sample (composite day sample). Based on the attached data what would be the most appropriate type of chart that I should be using. Could you also refer a good excel template for the type of chart I'm looking for.

It appears to me with a quick look at the data that X-MR is probably a good place to start. BUT....I will emphasize that there needs to be documentation of any changes in the process - start up, raw materials, operator, etc. Also, a confidence that you measurement system is adequate. I see some variation that might be explained by these types of things - which is OK...if you have the documentation to verify it. The control chart does not show an issue, but the distribution does.

I'm in the process of procuring QI Macros but before it's delivered, I want to learn the basics of SPC.

It is a handy tool! I use it and Distribution Analyzer a lot for these evaluations
 

Attachments

  • printing 2.docx
    231.1 KB · Views: 173
S

samsung

It appears to me with a quick look at the data that X-MR is probably a good place to start. BUT....I will emphasize that there needs to be documentation of any changes in the process - start up, raw materials, operator, etc. Also, a confidence that you measurement system is adequate. I see some variation that might be explained by these types of things - which is OK...if you have the documentation to verify it. The control chart does not show an issue, but the distribution does.



It is a handy tool! I use it and Distribution Analyzer a lot for these evaluations

Thanks Bob for a quick analysis. I agree there must be some documentation of changes in man, machine, material, method etc. The variation appears to be somewhat high and there's scope for minimizing it through control on inputs.

As suggested, I'll start with X-MR.

Thanks
 
S

samsung

Hi Bob,

One thing I would like to know is why you have calculated ppk instead of cpk in the above post. Is ppk a better indicator in this case and what is the optimum level where it should be maintained? 1.33 or so?
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
One thing I would like to know is why you have calculated ppk instead of cpk in the above post. Is ppk a better indicator in this case and what is the optimum level where it should be maintained? 1.33 or so?

That is the software's preference - it defaults to ppk, which is also preferred for many non-normal distributions. Some customers are also shifting from their preference of Cpk to Ppk. As far as what should be maintained - generally customers look for between 1.33 and 1.67 as a minimum. BUT - that depends on whether going to 1.66 is simply an adjustment of the mean, and improvement of the process (both of which are OK) or more frequent adjustment (or, over-adjustment, if you will, which is a bad thing because the adjustment becomes the process).
 
Top Bottom