Empower 3 ROUND discrepancies

I'll try to make this as simple as possible.

I have 3 injections for each sample, i need the individual result for each injection and then an average result of the 3 injections (possibly 2 samples making 6 total injections, both samples sharing the same label)

I have CF1, that calculates results for each injection, the math is not so important, works fine. Since we view this as an intermediate result, we report it with an extra numeral (e.g. 99,59) . The CF2 for average result, we report with the required rounding. (e.g. 99,6)

My CFs are set up like this:
CF1 = ROUND(*the mathematical expression*,-2)
The final result-
CF2= ROUND(SAME.%.SAME.AVE(ROUND(*the mathematical expression*),-2),-1)

Should be fine,  and for the most part it works perfectly, except for a specific exception. Just now i got these 3 results: 98,45   98,66   98,84
Calculator tells me, the average is 98,65000(0), which rounded should be 98.7.
My CF2 reports 98,7 on the first injection, however on the other two it shows me 98,6.

Something like this seems to occur every time Emp3 has to round an average, that ends in 5.

CF is Peak, Real, Result set only, All or nothing. Any help is appreciated

Answers

  • Hi Paul_3,

    I'm not sure if this article will answer your question (it kind of describes the behavior that you suggests is correct) but one of the attachments in the Additional Information section does have a lot of detail on rounding.  Have a look! 

    Maybe this one too....

    Good luck!
  • Waters is processing my support request, hopefully i'll get some answers.

    The basic priciple of rounding in Emp3 seems clear enough, what i don't understand is how can it report two different values by rounding what should be the same number?
  • Another thing to check is to make sure that the second custom field used to round the first custom field is named alphabetically higher than the first example name the first cf Calculated_Result and the second cf Rounded_Average. 

    When cfs depend on each other, they need to be named alphabetically so that A is calculated first. B can then be calculated, followed by C etc. You can get odd results when this rule isnt followed.
  • >When cfs depend on each other, they need to be named alphabetically so that A is calculated first.

    The thing is, my CFs aren't dependant on eachother, the second one, instead of referring to the first one, actually contains the entire formula of CF1 (otherwise it doesn't do intersample summary correctly)

    I just reprocessed the data with rounding precision set to -10, the rounded average is the same for all 3 injections. We have noticed this before, always when the roundable digit is a 5.