Some explanation about POE classes

View the concept of classes in the light of multiport PoE sources.   The ability to alocate power based on declared class offers the benefit of more ports served for a fixed power source (i.e. cost per port).  The assumption is that an oversubscription results in an overload that results in all PDs loosing power (at least temporarily).

Class 0 was implemented for those PDs that did not want to waste operating power, but chose to not implement a switched detection signature (cheap, maybe used as one per installation).  The range of classes is a roughly 2:1 set, and class 3 completes the set.  It was seen as a competitive advantage for PDs to use lower power (lower class) and to declare it, as the end implementer could extend the number of devices per multichannel PSE.  So for a 250W capacity, strictly speaking, a PSE could support 250/15.4 Class 0 PDs  or 250/4 Class 1 devices. 

I believe that the spec states that the declared hardware class is the maximum the power the PD will draw from the PSE.

With the advent of DLL class in .at, a PD may reserve and release differing power over time.  This will not be true for all PSEs.

The presence of class 3 means that there is a class 2 and then class 1 ... This is a "binning" algorithm.

Thank you very match Mr. Patoka

In you mail you probably thought 25 W (not 250W) ?

As I understand you well, you agreed that  POE class4  (25 W) should

suit for my case (nominal power will be 8-10 W and maximal 16-18W, very seldom) .

Expecting your final confirmation, best regards.

See 802.3at or 802.3 base standard (if/after update to include .at).  Sections 33.3.2, 33.3.5, 33.3.6, 33.3.7 (esp table 33-18 and 33.3.7.2), and 33.6.

It is always best to understand the base specification.  

In case you don't have a copy:  https://standards.ieee.org/about/get/802/802.3.html