| T O P I C R E V I E W |
| DRockWilson |
Posted - 27 Jan 2016 : 16:33:52
I am having a strange problem. I am have an Mean-Range chart that has data that is deviation from an Aim. So the values are sometimes negative if the value is below the Aim. Whenever I create a chart like this there is some weird labels/values that show up on the chart all squished together on the zero y-axis value line and always 1st and 2nd x-axis tick marks. Have anyone seen anything like this before and how to resolve it?
The Mean-Range chart works as expected if I use the actual values and not deviation from the aim.
I can provide more information if needed.
|
| 4 L A T E S T R E P L I E S (Newest First) |
| DRockWilson |
Posted - 28 Jan 2016 : 11:57:24
I sent an email so I could provide a picture of what I am seeing. |
| quinncurtis |
Posted - 28 Jan 2016 : 10:35:27
Based on the information provided, we cannot reproduce a problem. We simulated data with a mean of -45 and standard deviation of 15 and the software works like it should. Please answer the following questions.
1. Which QCSPCChart software package are you using (.Net, WPF or Silverlight)
2. We do not know how to interpret your data. It is an absolute requirement that a Mean-Range (XBar-R) chart have multiple samples per sample interval, with a non-zero range. Otherwise a standard deviation for a sample interval cannot be calculated. Yet you are just showing a list of values.
3. You said your data had a zero mean, yet the data you show does not have anything like a zero mean.
4. How exactly are you setting the control limit (not specification limit) values?
My suggestion is that get rid of all specification limits, and auto-calculate the control limits. See the chart that produces. If it doesn't look right, either post an image in the forum reply, or send it to support@quinn-curtis.com.
It is up to you to reproduce the problem in a simple example program. Start with one of our XBar-R examples and modify it first by changing the data, and then adding whatever else you are doing. |
| DRockWilson |
Posted - 28 Jan 2016 : 08:09:10
It would be difficult to send my whole solution because it connects to a production database. I guess I should specify that I am doing all the calculations for deviation from aim before I add anything to the chart. I guess it is really more of a delta from aim. Can you see if you can reproduce my problem with the following values (you can just add sequential times to these values)
-29, -39, -48, -56, -55, -57, -55, -60, -60, -61, -48, -47, -45, -39, -32, -41, -74, -61, -43, -40, -41
Here is my function where I add the data.
chart.Variable.SampleSizeXBarR = 3
chart.Variable.PointsPerChart = 8
public void GetChartData(SPCBatchVariableControlChart spcChart, ChartObject chart)
{
DoubleArray samples = new DoubleArray(Convert.ToInt32(chart.Variable.SampleSizeXBarR));
int j = 0;
int batchCounter = 0;
try
{
for (int i = ((Convert.ToInt32(chart.Variable.SampleSizeXBarR) * Convert.ToInt32(chart.Variable.PointsPerChart)) - 1); i >= 0; i--)
{
samples[j] = Convert.ToDouble(chart.ChartData[i].Value);
if (i % chart.Variable.SampleSizeXBarR == 0)
{
ChartCalendar timeStamp = new ChartCalendar(chart.ChartData[i].Time);
spcChart.ChartData.AddNewSampleRecord(batchCounter, timeStamp, samples);
samples = new DoubleArray(Convert.ToInt32(chart.Variable.SampleSizeXBarR));
batchCounter++;
j = 0;
}
else
j++;
}
}
catch (Exception ex)
{
throw ex;
}
}
Here is how I add the Spec Limit Lines
spcChart.PrimaryChart.GetControlLimitData(SPCChartObjects.SPC_CONTROL_TARGET).LabelLimit = false;
SPCControlPlotObjectData aim = spcChart.PrimaryChart.AddSpecLimit(SPCChartObjects.SPC_UPPER_SPEC_LIMIT, 0, "Spec Aim", new ChartAttribute(Color.Orange, 3.0));
aim.ControlLimitRecord.AlarmEnable = false;
SPCControlPlotObjectData min = spcChart.PrimaryChart.AddSpecLimit(SPCChartObjects.SPC_LOWER_SPEC_LIMIT, -40, "Spec Min", new ChartAttribute(Color.Orange, 3.0));
min.ControlLimitRecord.AlarmEnable = false;
spcChart.PrimaryChart.GetControlLimitData(SPCChartObjects.SPC_LOWER_CONTROL_LIMIT_1).LabelLimit = false;
spcChart.PrimaryChart.GetControlLimitData(SPCChartObjects.SPC_LOWER_CONTROL_LIMIT_2).LabelLimit = false;
spcChart.PrimaryChart.GetControlLimitData(SPCChartObjects.SPC_LOWER_CONTROL_LIMIT_3).LabelLimit = false;
SPCControlPlotObjectData max = spcChart.PrimaryChart.AddSpecLimit(SPCChartObjects.SPC_UPPER_SPEC_LIMIT, 30, "Spec Max", new ChartAttribute(Color.Orange, 3.0));
max.ControlLimitRecord.AlarmEnable = false;
spcChart.PrimaryChart.GetControlLimitData(SPCChartObjects.SPC_UPPER_CONTROL_LIMIT_1).LabelLimit = false;
spcChart.PrimaryChart.GetControlLimitData(SPCChartObjects.SPC_UPPER_CONTROL_LIMIT_2).LabelLimit = false;
spcChart.PrimaryChart.GetControlLimitData(SPCChartObjects.SPC_UPPER_CONTROL_LIMIT_3).LabelLimit = false;
|
| quinncurtis |
Posted - 27 Jan 2016 : 16:59:45
It doesn't sound like software is designed to work with values you say are deviations for an Aim. I have zero knowledge of that being used in SPC charts in general, or in our charts in particular. I would assume that you are somehow mixing in data values, and control limit values of completely different dynamic ranges. This is forcing the chart to be auto-scaled to the larger values, while the data is represented by the smaller values. I think you should just stick to using the software the way we designed it, which is to be used with actual data values, rather than deviations.
It was easy to simulate data with a mean of approximately 0.0, and a standard deviation ranging to 0.1 to 5, and the software works as expected, without the anomalies you describe.
If you send us a sample project (to support@quinn-curtis.com) which reproduces the problem you describe, we will at least look at it.
|
|
|
