XY axis calibration, when axis are compressing

[BCDueholm]

Hello,

I have an issues with XY plots like the one below. As can be seen the axis are compressing. So for example Delta_x=10 are a different amount of pixels in each end of the x-axis. Same is true for the logarithmic y-axis.

Is there any way to take this axis compression into account?

Best regards

[nbehrnd]

It looks like the scale's intervals (how many pixels for a Delta of 10 deg Celsius) changes three times yet remains linear along the abscissa within the intervals [-40...40], [40...100], and [100...140].

Thus, instead a one single definition of the abscissa extending all along the horizontal dimension of the plot, process the illustration three times for each of these intervals individually. I.e. define the points about the minimum/maximum at the limits of these intervals as «linear» 2D plot axis. (And for the ordinate, don't forget to enable the logarithmic scaling during the axes' calibration.)

[BCDueholm]

Hi,

Thanks for the answer.

I'm not quite sure what the «linear» 2D option is. What I did is the following. Add calibration for 2D(X-Y) plot, with the following points: Input the values of points: If I then put my cursor in [0,200], it gives approximately [3,225]. Am I missing something here?

To me it looks like scale's intervals are linearly decreasing in the direction of growing X/Y. For example the first and last box in [-40,40] interval are not the same size. Can this be taken into account, except doing the alignment in a lot of very small intervals?

[nbehrnd]

After measuring the dimensions of some of the rectangles (inkscape, activation by key m) of the illustration imported, the plot looks distorted. It neither is a clear semilog plot, nor an obvious double logarithm plot. Tentative simultaneous import of the .png and of a semi-log paper (e.g., papersnake.com) to adjust the shearing against the .png against this reference did not improve the situation for a better reading by the digitizer.

Thus I think the options available to the user of the diagram are limited to

• visual read-out of values close enough to the temperature of interest,
• selecting a number of visual (manual) read-outs around the temperature of interest and subsequent interpolation. Without access to the actual raw data used to draw the diagram, quite a number of variations could be submitted to a fit the values and to sort out some of them as «not good enough» by statistic tests. Perhaps the relationship of viscosity varying temperature is more complex than «only» by plotting the logarithm(s), perhaps (reciprocal?) absolute temperature in Kelvin (or 1/K) is better suited than the one in Celsius (Wikipedia article).

The example indeed looks like to feature a lock-in. I can't identify which of the sites on stackexchange were more suitable to address viscosity (e.g., physics, génie civil/engineering).