Values combined into classes result in a frequency diagram (histogram), i.e. the measurements values are classed, and the x-axis of the value plot has been divided into classes. The number of values within a class is referred to as class frequency. You may show this class frequency on the y-axis in the histogram either in absolute figures or relative to the subgroup size (percentage or ppm). The histogram is usually displayed together with the probability density function (PDF) of the fitted distribution.

This topic describes the special settings that only apply to this graphic. There are several variants of histogram graphics available. These are described separately.

1    Histogram

In the histogram of the individuals, the raw values are grouped into classes and displayed according to their absolute/relative frequency. By default, the x-axis represents the entire range of values and the frequencies are plotted as bars on the y-axis.
 

1.1     Distribution

This could be used to deactivate the distribution curve.
 

1.2     value plot

With this option, the value plot can also be displayed in the histogram. One of the most important functions of the value plot within the histogram is thus the visual representation of the resolution.

Sufficient resolution	Low resolution

1.3     Bar labelling

With the bar labelling, the individual bars of the histogram can be labelled with the number of values.
 
However, a more detailed representation is available in the graphic "Show classification".
 

1.4     Show quantiles

With the quantiles displayed, movable bars appear in the histogram, for which the proportion outside specification can be read in the upper part, and the "limit value" required for this in the lower part.

The option works together with the C-value function option described below.

1.5     C-value function

By activating the C-value function, sloping lines are shown as well as an additional y-axis of the "C-nominal values".
 
Together with the activation of the superimposed quantiles, the following possibility arises:
As an example, the "specification limits" are to be determined, which achieve a capability of 1.67 in the process. For this purpose, a characteristic is loaded as an example, which is not capable (Test_01.dfq).
After fading in the C-value lines, their intersection with the "1.67" line is searched for: As can be seen, these limits are still outside the visible range.

Now the quantiles are also displayed.

These lines can now be pushed "outwards". To put it somewhat technically: You push the held quantiles a little towards the edge:

Now the quantiles can be shifted to the intersection points of the C-value lines with the line of the C-value = 1.67 - line. In the lower area, the specification limits can now be read, which may ensure a C-value close to 1.67.

For cross-checking, these specification limits are entered on the characteristics mask:

This would roughly result in a potential index of 1.67.

As can be seen, this is not a 100% guarantee to exactly back-calculate the C-value. Rather, this is intended to be an aid in the discussion about specification limits and their possible extension.

Another possibility was realised in the standard report "1060_Minimal_Tolerance" with the associated output points. Here, however, the "target C-value" is not freely definable, but the target value which currently exists as a target value for this characteristic.

1.6     Show multiple distributions

In the discussion about which distribution should be taken, deviating from what the strategy has defined, several options are available.

In general, it should be mentioned in advance that precisely this question "which distribution should be taken" is the reason for the existence of the Q-DAS software. The evaluation strategy searches for the best-fitting distribution for the distribution time model. A manual search for other forms of distribution is nevertheless a frequently used option, because no strategy should be accepted as a "static unchangeable truth". With the change of processes, the improvement of measurement techniques, the change of structures for sorting the data (K-field workshop), new requirements always arise, also for the strategy. Here, these options for sifting through several distributions can be helpful.

After calling up the option, you can first select whether you want to work with or without offset. Then the desired distributions can be selected.
 
It should be noted that only distributions should be selected which - as far as the regression coefficients are concerned - have a high value.
As a bad example, the matching log-NV as well as the non-matching magnitude distribution are shown here.

Here it can then be seen in the representation that this does not fit (as an extreme example)

The following distributions could also be displayed as suitable distributions

With this data set, all 3 distributions would be "suitable" here:
 

2    Histogram in solara.MP

The classic histogram is of secondary importance in solara.MP. But as with the value chart, there are 2 special histograms for this method.

The technical explanation of what "part centre" or "reference" means is described in the value chart documentation.
As with the value chart, the overview of all measurements can be compared with the measurements of the individual operators.

Topic in PFD format

This topic is also available in PDF format.

English:PDF

German:PDF