What is the Formula for Cpk in Excel?

[lee01]

Can someone inform what the Cpk formula I should be using in excel?

I'm currently using:



=IF(B11<B12,B11/3,B12/3)

Converted: If(0.351<2.921,0.351/3,2.921/3)


I'll tell you what, I'm taking a full sample of points and have these values, anyone want to give me an idea how to calculate the value?0.0166
0.0073
0.0003
0.0078
0.0039
0.0138
0.0033
0.0101
0.0008
0.0009
0.0114
0.0129
0.0021
0.0053
0.0173
0.0139
0.0021
0.0007
0.0118
0.0126
0.0019
0
0.0124
0.0123
-0.0036
0.0145
0.0088
0.0101
0.0068
0.0089
0.0178
0.0014
0.0136
0
0.0138
0.0141
0.0023
0.0143
0.0172
0.0038
0.0084
0.0234
0.0129
0
0.0011
0.0119
0.0111
0.0023
0.0039
0.0092
-0.0018
0.0073
0.0014
0.004
0.0116

Mnay many many thanks

Lee01

[Sam]

I don't recognize the equation. How did you get there?

[Craig H.]

Lee:

Why don't you put the data in an Excel file and attach it to a thread?

I assume the spec is .351 min, 2.921 max?

[lee01]

Sorry my friends I’ve got it know!

I have been creating the following statistics in the following way, please can you have a look at the formula’s to make sure I’m on the correct path.


Cell Statistic & Exactly what I’m
Number Result Entering in Excel


B1 USL =0.005 0.005 (From design)
B2 LSL = -0.005 -0.005 (From design)
B3 STD = 0.001 =STDEVP(B15:B63) (note: B15:B63 is the location of the data)
B4 Mean = 0.001 =AVERAGE(B15:B63)
B5 Min = -0.002 =MIN(B15:B63)
B6 Max = 0.004 =MAX(B15:B63)
B7 Range= 0.006 =MAX(B15:B63)-MIN(B15:B63)
B8 UCL = 0.005 =B4+(3*B3)
B9 LCL = -0.003 =B4-(3*B3)
B10 Cp = 1.25 =(B1-B2)/(6*B3)
B11 Cpu = 3.25 =(B1-B4)/(B3*3)
B12 Cpl = 4.251 =(B4-B2)/(B3*3)
B13 Cpk = 1.08 =MIN(B11:B12)

[lee01]

Sorry my friends I’ve got it know!

I have been creating the following statistics in the following way, please can you have a look at the formula’s to make sure I’m on the correct path.

Its in the attachment

Lee01

[Darius]

Wrong

You are using Total Variation estimate (STDEV) to estimate Control Limits, and you are using Total Variation estimate to calculate Cp, Cpk, Cpu, Cpl.

The index calcualted this way is pp, ppk, ppu, ppl

To calculate the Cp, Cpk, and Control Limits use Moving Range or Averages to estimate the variation.

The Total Variation estimate is affected by the variation between samples and within samples, the estimate for control limits is the within sample variation.

Sigma = MR_average / 1.128 (for individual)
or
Sigma = R_average /d2 for (subgroups <>1)

[Atul Khandekar]

Lee,

I still don't get it - or maybe I am missing something very obvious!

1.You have posted 55 values, yet the document shows AVERAGE(B15:B63) ie only 48 values.

2. Your computed values are rounded off to 3 decimal digits and the data has 4 decimal digits. There may be some errors due to rounding.

3. I did not try calculting with Excel, but you have so many values above USL (0.005) that Cpk of 1.08 seems unlikely. Then again MIN(3.25,4.251) can't be 1.08 !

4. If by "full sample of points" you mean the entire population, then STDEVP is okay. If not, use STDEV. and yes the index would be Ppk.

Can you please post your original Excel sheet?

[lee01]

Okay! Wait a minute.

I understand where your coming from but. . . . .

I'm trying to work out Cpk, & Cp in excel and I have and willl continue to measure the full spopulation (No sample measurement) so I think I'm using the correct Excel calculation.

Is this worng/ correct what should I do? This is being used and has been used for some time in this multinational organisation so are you sure and how will I inform everyone they have been doing it wrong and how they are doing it wrong and how to to do it correct.

Lee01