Thanks for the reply.
The problem is that in each graph I have three data series, and I have 4 graphs. In total I have 12 data to graph. Each group of three forms a graph, the data are tension, intensity and power. The power is calculated from the product of the tension by the intensity and by a factor that is not worth explaining. What happens is that the voltage sometimes reaches values of 1000V while the intensity reaches values of 0.1A and the power then is 100.W if the factor is 1, which it never is.
As you can see, the values are quite different at that time and you may be interested in seeing the power at that time but with a higher resolution of the graph. The idea was to allow the deactivation of the graph for the series that are not temporarily of interest to the operator's selection.
Another option is to have a logarithmic scale but that does not suit me either because at other times all the values are quite close and the logarithmic scale is not adequate either.
I get the data from a communication with an ESP32 that is reading 8 analog inputs at high speed in pairs, let me explain, the ESP32 reads the 300 voltage and intensity measurements of a converter in 0.1s, processes them and performs the pertinent calculations and then send them for the B4A application to store them in files and view them in the graph, and thus each one of the 4 double channels it has.
The storage is carried out in temporary files, if necessary or an anomaly is detected, it is sent to the server for further processing.
I think I explained it quickly and not very precisely but the process is complicated, at least for me.
Any idea how to solve the problem.
Thanks