Piecewise calibration curves in OpenLab 2.7?

Does anyone know how to get the multi-level calibration curves to be simple piecewise functions between each pair of points?

Ordinary chromatography integrators as well as the regular Chemstation software had a selection for this most fundamental calibration curve, but I'm finding nothing in OpenLab that would allow it to replace older software in this regard.

Best fit is fine for many linear applications, other times piecewise is still needed even for linear responses.

Parents
  • Hello,

    Piecewise calibration is not available in CDS 2.x. We have had requests for piecewise and higher order regressions as calibration options but at present they are not available in any CDS 2.x version.

    Marty Adams

  • Thanks Marty, I was afraid of that.

    Too bad really since piecewise is the most conventional kind of multi-level calibration.

    I train using the example of a typical pH meter where they still naturally calibrate at pH 4, 7, & 10.  Therefore there are two slightly different slopes, one from 4- to 7 and the other from 7 to 10+.  If instead you use more advanced math to determine the closest linear fit among the three points, the pH 7 standard then often fails to show 7.0 as an SQC result.

    With the earliest chromatography integrators there was no regression so it was only piecewise.  After a few decades you get accustomed to having calib standards return more ideal results then merely the "closest fit".

    I expect we'll just have to wait.

    Interestingly, earlier today I was cleaning out a cabinet and found some old HP floppy disks from 1997 with firmware updates for the ALS at the time.

    The README says the firmware is to add petrochemical capabilities, that must have been forgotten about when the device was first issued.

    Things like alcohol or acetone cargoes are simple industrial chemical commodities, but the number of workarounds continues to increase in unexpected ways.

    Oh, well.

Reply
  • Thanks Marty, I was afraid of that.

    Too bad really since piecewise is the most conventional kind of multi-level calibration.

    I train using the example of a typical pH meter where they still naturally calibrate at pH 4, 7, & 10.  Therefore there are two slightly different slopes, one from 4- to 7 and the other from 7 to 10+.  If instead you use more advanced math to determine the closest linear fit among the three points, the pH 7 standard then often fails to show 7.0 as an SQC result.

    With the earliest chromatography integrators there was no regression so it was only piecewise.  After a few decades you get accustomed to having calib standards return more ideal results then merely the "closest fit".

    I expect we'll just have to wait.

    Interestingly, earlier today I was cleaning out a cabinet and found some old HP floppy disks from 1997 with firmware updates for the ALS at the time.

    The README says the firmware is to add petrochemical capabilities, that must have been forgotten about when the device was first issued.

    Things like alcohol or acetone cargoes are simple industrial chemical commodities, but the number of workarounds continues to increase in unexpected ways.

    Oh, well.

Children
  • Do you run your standards and samples in the same result sets? If so, it might be possible to do the calculation in CC.

    Marty Adams

  • That's what I'm beginning to think now.

    Kind of like multi bracketing.

    This would escalate things quite a bit it looks like.

    However the same result set could get a bit oversized before too long, I don't know how open-ended the result sets are either.

    With very straightforward FID work testing flammable chemicals, once the relative response factors of all the components relative to each other (or relative to internal standards) have been determined, they usually don't change as long as injection, flow, & detector conditions remain consistent.  So full calibration is not needed yearly at all, only daily verification to make sure things like the septum, syringe, and electronics have not been compromised.

    Routine method changes are basically just to occasionally update retention times as columns age.  On most methods everyday operators never calibrate or recalibrate response factors.

    The FID is more stable than the reference standards anyway.

    OTOH with the SCD there is a standard before the sample and one after, the original Sievers specification was for "6 full hours" of stability, but that was under ideal conditions.  This GC is manually corrected per sample.

    I suppose the calibration files could be copied over to each result set as needed, that might be an idea.

    Thanks again, I am just beginning to explore the additional possibilities that were not available in previous decades so it is interesting.

    Best Regards,

    Mike

Was this helpful?