In contrast to recalibration samples, check spectrometers samples have composition corresponding to those of qualities to be analysed or have such low alloy concentration that they lie on the calibration curves produced with calibration samples.

Again in contrast to recalibration samples, check samples have analysis values which have been determined, using different procedures, in at least 5 laboratories producing such qualities. Their analyses are reliable within narrow limits.

  The high expenditure incurred for CRM is not necessary for check samples. In the case of CRM the main is not to bring as many as elements as possible into a few samples collected to form a set. When using such samples for high-speed steel only “auxiliary calibration curves” should be plotted, these being supplemented by means of calibration samples corresponding to the high speed steel qualities.

  Check samples can be used in various ways on spectrometers. After the spectrometer has been calibrated, the check samples are analysed a number of times in the recalibrated state. The spectrochemically determined values must agree with the given value within ±2SR of the calibration curve and both values must be reported. If there is no agreement, calibration or recalibration is insufficient. Correct calibration of the spectrometer can be tested by the user by means of the check samples:

  a) If the user does not have confidence in the spectrometric analysis of an unknown sample,

  b) at regular intervals in order to increase quality assurance

  c) to check the need for recalibration. In this case the sample can be used for type recalibration.

   Method is recommended particularly if relatively few samples per day are measured with the spectrometer and these samples belong to only a few quality groups.

   The individual elements in the check samples must be assigned sensible limits based on the spectrochemically determined value. If , at the time of time of the check procedure, all the values determined are within these ± tolerance nothing need be done. If they are outside, recalibration is necessary.

   The actual recalibration values determined with recalibration samples and the actual check values determined with check samples should be recorded and assessed by statistical methods. This leads to increased operational reliability of the spectrometer and provides information on actual stability and the frequency of recalibration. The last-mentioned also leads to saving in material for calibration samples.