Level specific dilution factors in the compound table of the calibration [OpenLab 2.5]

Isn't it simpler to calculate the concentrations of each substance at each level and use these values in the Calibration table?

Andy

Hello Andy, thanks for your answer. It would be possible, however we would like to automate our analysis as much as possible. A manual calculation of the concentrations, depending on the initial weight, would make us go a step in the wrong direction.

In the meantime I have rebuilt the calculations of the calibration within the custom calculation editor and it works so far. Only the linking of the results with the corresponding samples still fails. I cannot include the calculated slope and the intercept of the reference solutions in the calculation of the sample concentration.

Thanks
coh

I guess that you would have to use the calculated slope and intercept in another custom calculation to calculate the results for the samples too.

To calculate the slope and the intercept we used the formulas that can be found in the OpenLab documentation "CDS_DataAnalysisReferenceGuide" under the item 'Calibration Curve Calculation > Linear Fit'.
Now we noticed that the results are similar to those we obtained for the calculation in 'Origin mode': Origin included via OpenLab and Excel.

I hope someone has an idea for this.

Here are our formulas from the custom calculation:

Slope:



Intercept:


The custom calculations 'CalibrationAmountLvlX_E20' within the formula are the weights of the reference levels 1-3 and the custom calculations 'CalibrationAreaLvlX_Dyn' are the areas of the reference levels 1-3. So our RF ist defined as Amount per response.

Hello,

I am not sure I understand the question you are asking. I have done the regression in CC in the past for a couple of reasons and it can be done to match the standard OpenLab calculations. My custom calculations are more generic and can be used for any number of standard levels and compounds.

Hello,

I am not sure I understand the question you are asking. I have done the regression in CC in the past for a couple of reasons and it can be done to match the standard OpenLab calculations. My custom calculations are more generic and can be used for any number of standard levels and compounds.

Hello Martin,
Thanks for your reply. Unfortunately I asked my question unclearly, that was my mistake.

In our Excel spreadsheet for data verification of OpenLab results, we compared the Custom Calculation calculations for slope and intercept to the results in Excel and the in-program calculations in OpenLab.

We noticed that the results based on the slope and intercept from the Custom Calculation are very similar to the results we get in Origin mode "Include" and "Force".
As we understand it, we should get results from Origin mode 'Ignore' with our calculations.

In the screenshot, you can see our comparison calculations where the mean of 'f(x) manual CC' is similar to those from the program's internal 'f(x) transfer OpenLab OL O. Force' and 'f(x) transfer OpenLab OL O. Incl.' are very similar. 'f(x) transfer OpenLab' names the values from the program-internal 'Origin Ignore' calculation.

Additionally, I have attached a screenshot comparing the slope and intercept from Open Lab CC with those from Excel. Theoretically, these values should be the same or at least similar.
Again, only three amount and response values were taken for the calculation.

Is there an explanation for this? The calculations in the Custom calculation seems to be correct.

The selection of the Origin mode under 'Processingmethod>Compounds>Calibration>Compound Table' should have no influence on the Custom Calculation, right? Currently we have selected 'Included' there.

Hello,

From your calculation it looks like the weighting terms are missing. Those terms in that calculation replace the N term in the more standard least squares regression equations when no weighting is done. Below I did the calculation for some test data in excel with the Agilent calculation and a standard non weighted equation and got the same results for all. I have the weight included but all set to 1 for a non weighted regression. If you are not weighting the levels you could use the standard calculation shown below with the simple N value in your case 3. 

Marty Adams

Hello Marty,

the missing factors were indeed the problem. Now I get identical results with the CC compared to Excel. Thanks a lot for the great help!

coh