Custom Field Input

Hi, 
I would like to calculate [sample component area/(average of 6 standard injection component area -sample component area)]. How do I write the formula in custom field? Is it possible to let empower show me the value directly in reviewing window without processing the sample with summarize custom field? Thank you 

Best Answer

  • Empower2019
    Accepted Answer
    Label your standard injection as S01,Then create a peak, real cf with search order result set only and leave the sample and peak type to All. The formula is:
    Area/((S01.%..AVE(Area))-Area)
    Then process the sample set with summarise custom fields to generate a result set with the answers. If you want to see the result in review before processing, you need 2 cfs. First is a peak real cf search order result set first and formula is:
    (S01.1.(Area)+S01.2.(Area)+S01.3.(Area)+S01.4.(Area)+S01.5.(Area)+S01.6.(Area))/6 and call this cf Standard_Mean. 
    Then create a second peak real cf search order result set first and again leave the sample and peak type to All. Formula is:
    Area/(Standard_Mean-Area) and this should give you a result in review windown as long as you first cycle through the 6 injections of standard to generate a result for Standard_Mean then view your sample injections. Good luck.

Answers

  • Thank you so much. it works for me!! Now I have another question. Empower automatically does sample amount calculation based on (area-std area)*std amount.  I want to treat Area/(Standard_Mean-Area) as a correction factor and calculate true amount of the sample. How can I write it in CF, so empower can show me the result in review window? Thank you again for the help
  • You can't. Summary functions as required to generate the Standard_mean value only calculate via the Summarize Custom Fields function in a sample set. 
  • Try putting the cf for Area/(Standard_Mean-Area) as the Y value for your components in the "Y Value" column of your processing method, save and again cycle through your standard injections to generate a value for this cf. Test your sample amount, or process sample set and see if the result set gives you corrected amounts.
    Im not sure this will work though as Y value cfs are usually just an adjustment of the response eg Area*10 or Height/20. Your cf incorporates the unknown area divided by standard mean and then subtract area. If this is your Y value each standard injection will generate a result by dividing the area of component in standard by: mean area of 6 stds minus the same area of component in that standard, not sample. Then a calibration curve will be built for the standards based on these 6 values..
    Your sample amount will be calculated and compared to this curve, is that really what you are looking for?
    A better way is another peak, real cf, result set first all sample and peak types. Formula:
    CF2/((Mean_Standard)*Value*Dilution))/Sampleweight and see if that is better.
  • This seems like a very convoluted process for something you probably shouldn't be doing in the first place ... at least not in a regulated environment. 

    Just my $0.02....
  • Try putting the cf for Area/(Standard_Mean-Area) as the Y value for your components in the "Y Value" column of your processing method, save and again cycle through your standard injections to generate a value for this cf. Test your sample amount, or process sample set and see if the result set gives you corrected amounts.
    Im not sure this will work though as Y value cfs are usually just an adjustment of the response eg Area*10 or Height/20. Your cf incorporates the unknown area divided by standard mean and then subtract area. If this is your Y value each standard injection will generate a result by dividing the area of component in standard by: mean area of 6 stds minus the same area of component in that standard, not sample. Then a calibration curve will be built for the standards based on these 6 values..
    Your sample amount will be calculated and compared to this curve, is that really what you are looking for?
    A better way is another peak, real cf, result set first all sample and peak types. Formula:
    CF2/((Mean_Standard)*Value*Dilution))/Sampleweight and see if that is better.
    Thank you for the help and sorry for the late reply. I got very busy with other stuff at the year end.  I was trying to adopt a new USP method for Isopropyl alcohol, but I'm completely new to its calculation and I'm not sure how it can be done in empower.  The UPS formula is  Result (%v/v)= {[Ku/(Ks-Ku)] x Ck}/10000.   Ku= peak area of individual known impurity in sample, Ks= peak area of individual known impurity in standard, Ck= concentration of spiked individual known impurity in standard.(ul/L). I can always do manual calculation on it, but I think it would be more compliant in a regulated environment if the calculation was done by empower. 
    As for putting CF in Y value, it didn't work for me, and I'm not quite following the other formula since my test involves volumes only, no sample weight or dilution involved. Would you be able to help me more on this matter? Thank you and happy new year!
  • If the CF didnt work in the Y Value, you will have to try that manually, I cant see another way around that. As for your first question, create a peak, real CF search order Result Set First, All Sample and Peak types, and formula of (Area/((S1.%.(Area)-Area)))*CConst1

    Make sure to label the standard as S1 and put the Ck value into CConst1 of the processing method used to process the data, so each component can have its own Ck value example Caffeine has a Ck of 12.456 and Acetone has a value of 123.897- put the values into CConst1, save the PM and process some raw data. I checked it on some test data and it worked fine for me using several unknown samples. You get different values for each unknown sample, as expected. Hope that helps. 
  • If the CF didnt work in the Y Value, you will have to try that manually, I cant see another way around that. As for your first question, create a peak, real CF search order Result Set First, All Sample and Peak types, and formula of (Area/((S1.%.(Area)-Area)))*CConst1

    Make sure to label the standard as S1 and put the Ck value into CConst1 of the processing method used to process the data, so each component can have its own Ck value example Caffeine has a Ck of 12.456 and Acetone has a value of 123.897- put the values into CConst1, save the PM and process some raw data. I checked it on some test data and it worked fine for me using several unknown samples. You get different values for each unknown sample, as expected. Hope that helps. 
    Thank you very much, it works for me. How did you learn empower custom fields? I'm planning to take some training courses from Waters. Based on the course descriptions, I feel like they are quite generic. I'm not sure if I will be able to address specific needs after taking them. Do you have any recommendations on them, or any books? Thank you again for the help!
  • If you google Empower 3 custom fields you can get a huge amount of articles, some are good others not so much, and you can also get great help if you google Empower 3 tip of the week, i think Neil Landers covers cfs in detail from tip 80 or 82 onwards. It takes about a year or two to become proficient in the practical use of cfs. You need lots of practice using all the different types of cfs like peak, result, bool etc and using different formulas across multiple labels, channels etc, to get a sense of the best way to use them in your lab.
    If you really want to invest in the functionality and several examples of cfs  including intersample cfs, then consider buying the Empower 3 custom fields tutorial book from Beyontics, thats an excellent resource for self-starters and for anyone who wishes to apply cfs for all types of calculations.
  • Hi...  just saying, if you have multiple stds at one level, or multiple stds at multiple levels, the "Average by Level" setting in your processing method actually uses the 'Average Std area' as the denominator in the default calculation anyhow.... no need for a custom field or doing it all manually.