Thank you very much for the comprehensive answer, my friend is a specialist in the subject.
... You also have a COM terminal on the input terminal block - this is the input reference voltage terminal (against which Uwe is measured). For PNP they usually need to be connected to 0V, for NPN - to Vcc. In the case of modules for CQM1, e.g. ID212 - it is a universal module. The way of COM connection determines the type of input.
So it is possible to wire several WE ID 212 modules within one CPU in such a way that one module will be PNP and the other NPN? and thus some sensors in the machine are PNP and others are NPN?
For NPN, shorting the input to 0V gives logic "1" in the program. The unconnected input will have a potential close to 24V.
And for PNP, shorting the input to 24V gives a logic "1" in the program. An unconnected input will have a potential close to 0V. According to this:
WE PNP - SINK - 24V signal fed to WE gives a logical 1, COM to 0V
WE NPN - SOURCE - 0V signal fed to WE gives a logical 1, COM to 24V
It seems that the machine I am talking about (produced around 1983) was modernized by two programmers / assemblers - one from Europe and the other from Asia. They both wanted to use the OMRON but they argued when connecting the signals and I have a nice mix of PNP and NPN.
But it was more about the cost of upgrading: some sensors remained old - NPN, and new ones were added to PNP. Hence the difference in the method of wiring EC modules.
From myself I will add that diagnosing a fault in such a control is a pure torment, especially since most sensors have unreadable markings (PNP NPN).