| Author |
 Topic  |
|
|
sakamali
Turkey
33 Posts |
 Posted - 28 May 2013 : 14:07:46
|
Hi,
We are using your products and there is a problem in np-Chart Control Limits. I have these sample values :
{10, 12, 8, 9, 6, 11, 13, 10, 8, 9, 6, 19, 10, 7, 8, 4, 11, 10, 6, 7}
In this form, CL is calculated truly : 9,2
Also UCL and LCL are calculated : 9,2
However, UCL should be 18,159 (~18,16) and LCL should be 0,241 (~0,24) according to p (p should be 300 in our sample)
UCL = np + 3 * [(np * (1- np/p))^(1/2)]
UCL = 9,2 + 3 * [(9,2 * (1- 9,2/300))^(1/2)]
LCL = np - 3 * [(np * (1- np/p))^(1/2)]
LCL = 9,2 - 3 * [(9,2 * (1- 9,2/300))^(1/2)]
How can I set needed p value? So, if I run my example in C#, I get this chart :
 |
|
|
quinncurtis
1586 Posts |
Posted - 28 May 2013 : 16:05:06
|
| Since you show a NO. INSP. value of 1 in your table, I assume you did not properly set the value of the numsamplespersubgroup parameter in the call to the InitSPCBatchAttributeControlChart or InitSPCTimeAttributeControlChart . In your case that would be 300. |
 |
|
|
sakamali
Turkey
33 Posts |
Posted - 29 May 2013 : 11:50:49
|
Thanks for your help.
Regards. |
|
|
|
sakamali
Turkey
33 Posts |
Posted - 14 Jun 2013 : 10:23:05
|
Hi,
My sample values are :
{5, 4, 7, 6, 8, 5, 6, 5, 16, 10, 9, 7, 8, 11, 9, 5, 7, 6, 10, 8, 9, 9, 7, 5, 7}
What should be the other parameters to get below chart (u-Chart) and UCL, U and L values?
Thanks,
Regards. |
Edited by - sakamali on 14 Jun 2013 10:23:37 |
|
|
|
quinncurtis
1586 Posts |
Posted - 14 Jun 2013 : 11:10:46
|
Sorry, but I do not understand the question. You seem to be using a batch attribute control chart, so study the examples in the BatchAttributeControlChart example program, most specifically the NumberDefectsPerUnitControlChart chart, which is the u-chart. Base your own program on those examples, and frame any questions you have in terms of the code in those examples. Also, make sure you read the section Updating c- and u-charts (Fixed Sample Subgroup Size) in Chapter 5 of the manual.
By the way, your example chart does not look correct for a u-chart. Are you saying that the average number of defects per unit for your process is as high as 16, with an average of 8? Because that seems extremely high. Or are these supposed to be the number of defects for a sample of 300 parts per sample interval. In which case the chart is wrong for a u-chart, which should show the average defects per unit per sample interval, not the total number of defects per sample interval. This example: http://en.wikipedia.org/wiki/File:U_control_chart.svg is much more typical of a u-chart, which is similar to our example program.
The chart you show above looks to be a c-chart (number of defects) , not a u-chart. The NumberDefectsControlChart chart from the BatchAttributeControlChart example program represents the c-chart. So you need to decide what type of chart you want.
|
|
|
|
sakamali
Turkey
33 Posts |
Posted - 17 Jun 2013 : 04:08:52
|
Hi,
I want to use u-Chart and my parameters that I use are in below:
int charttype = SPCControlChartData.NUMBER_DEFECTS_PERUNIT_CHART;
int numsamplespersubgroup = 1;
int numsamplestotalgroup = 1;
Samples = {5, 4, 7, 6, 8, 5, 6, 5, 16, 10, 9, 7, 8, 11, 9, 5, 7, 6, 10, 8, 9, 9, 7, 5, 7};
And I call spcChart.ChartData.AddNewSampleRecord(timestamp, samples);
I get this chart:
Which chart do you get by these values?
|
|
|
|
quinncurtis
1586 Posts |
Posted - 17 Jun 2013 : 08:41:38
|
I still don't think you want a u-chart.
If you do, then please confirm that based on your data, your product has an average of 7.56 defects per unit. That seems like a really bad process which produces such an error rate. A more typical number of defects per unit for a u chart would be 0.01 to 0.1.
I do not understand the question: Which chart do you get by these values? Please rephrase it with more detail.
MORE: The graph that you show, with the UCL of 15.81, is a c-Chart (Number of Defects), NOT a u-chart. As stated previously, our example program for that is BatchAttributeControlChart.NumberDefectsControlChart. Please take the time to figure out which chart you are dealing with.
|
|
|
|
sakamali
Turkey
33 Posts |
Posted - 17 Jun 2013 : 09:35:55
|
Could you please send me a sample (not random) data (and parameters) for get a u-Chart?
Thanks. |
|
|
|
quinncurtis
1586 Posts |
Posted - 17 Jun 2013 : 11:01:07
|
Sorry, but we do not write custom example programs on demand. You must work with the examples we have already provided you with.
Modify the BatchAttributeControlChart.NumberDefectsControlChart by replacing the Simulate data method to use your own data.
All of our data is random, so use the data you already have. The result does not look your original chart, because your original chart is a c-chart, and has c-chart limits.
You should probably add the line:
this.EnableCategoryValues = false;
Since you only have one category of defect.
private void SimulateData()
{
if (this.IsDesignMode) return;
double[] data = { 5, 4, 7, 6, 8, 5, 6, 5, 16, 10, 9, 7, 8, 11, 9, 5, 7, 6, 10, 8, 9, 9, 7, 5, 7 };
for (int i = 0; i < data.Length; i++)
{
double batchnumber = batchCounter;
ChartCalendar timestamp = (ChartCalendar)startTime.Clone();
DoubleArray samples = new DoubleArray(1);
samples[0] = data[i];
// Add new sample record
this.ChartData.AddNewSampleRecord(batchnumber, timestamp, samples);
// Simulate timeincrementminutes minute passing
startTime.Add(ChartObj.MINUTE, timeincrementminutes);
batchCounter++;
}
}
Nothing else needs to be changed. The result is a u-chart. It assumes that the defect numbers you update the chart with are based on a 50 unit sample size per sample interval, the value of which can be changed using the numsamplespersubgroup value at the top of the program.
|
|
|
|
sakamali
Turkey
33 Posts |
Posted - 18 Jun 2013 : 04:25:50
|
Hi,
I got these values by data[] array. Are they true?
UCLP = 0,385
PBAR = 0,151
LCLP = 0,000 |
|
|
|
quinncurtis
1586 Posts |
Posted - 18 Jun 2013 : 08:39:32
|
| Yes, those are the values you obtain with that data, assuming those are the defect numbers for a sample size of 50 units per sample interval. |
|
|
|
sakamali
Turkey
33 Posts |
Posted - 18 Jun 2013 : 08:53:59
|
Thanks a lot.
Regards. |
|
|
|
sakamali
Turkey
33 Posts |
Posted - 20 Jun 2013 : 03:31:58
|
Hi,
There is a problem. I have an example for u-Chart from your examples and it has these samples (Generated randomly) and calculated values:
UCLP = 0.489 PBAR = 0.180 LCLP = 0.000
U-Chart Formulas are:
According to the formulas, calculated values must be below values:
UCLP = 0,18 + 3*[(0,18/50)^(1/2)] = 0,36
PBAR = 153/(17*50) = 0,18
LCLP = 0,18 + 3*[(0,18/50)^(1/2)] = 0
And this chart seen true
Where is the problem?
Thanks.
|
|
|
|
quinncurtis
1586 Posts |
Posted - 20 Jun 2013 : 09:49:15
|
| What you describe is a bug in the software for the Number Defects Per Unit chart. We have fixed the bug and you can download the corrected version of the software using your original download link. |
|
|
| |
Topic |
|
|
|
Control Limits calculate problem
