How does Power over Ethernet classification work?

Question

In this TPS2372 PoE PD there is a table about how to set up classification of the device. I think I understand the concept of classifying PD devices, it's to tell the supply equipment the power demand the PD needs.

But there are a few scenarios that in which I'm not sure what will happen.

1. Why does there seem to be a minimum power for certain class (e.g. PD class 2, 3.84W - 6.49W min to max)? What if the PD device idles down and is using a low 0.5W power? What would happen then? Must I design to ensure that it doesn't go below that?
2. What would happen if the PSE can't deliver the power requirement of the PD device, so the PD device is on a higher classification than what the PSE can deliver. Would power still be given to the PD device but only the highest it can go? Where does the responsibility of protecting both equipment from damage due to lack of power? Is it the job of the PoE PD controller to outright not give power, or its further down the line where the PoE PD controller send a signal to the microcontroller/processor that not enough power is available (designer's responsibility), or simply the PSE will just straight up not give power?

Answer 1

Why does there seem to be a minimum power for certain class (e.g. PD class 2, 3.84W - 6.49W min to max)? What if the PD device idles down and is using a low 0.5W power? What would happen then? Must I design to ensure that it doesn't go below that?

The Maintain Power Signature (MPS) parameter is used by the PSE to detect when then PD has been disconnected, so that the power can be removed from that port. As per DC MPS, for each class there shall be a minimum current draw from PSE at a regular interval. If the PD is not drawing the min current for more than the specified time within each time interval, the PSE will assume that there is no PD connected on the other end, so it will stop supplying power. the DC MPS current, time limit and time for validity all are specified in the IEEE802.3 standard.

What would happen if the PSE can't deliver the power requirement of the PD device, so the PD device is on a higher classification than what the PSE can deliver. Would power still be given to the PD device but only the highest it can go? Where does the responsibility of protecting both equipment from damage due to lack of power? Is it the job of the PoE PD controller to outright not give power, or its further down the line where the PoE PD controller send a signal to the microcontroller/processor that not enough power is available (designer's responsibility), or simply the PSE will just straight up not give power?

Most of the PDs will support feature called demotion, if the PSE is in lower class, the PD will demote itself to lower class and the PSE will provide the maximum power it can provide. If the PDs supports multiple classes, most PDs will have output status pins which will give status about with which class it is working. The designer can use the status pins let know the controller or any power management IC how much power its getting and whether to turn off the device or operate in a low power mode.

Answer 2

1. Why does there seem to be a minimum power for certain class (e.g. PD class 2, 3.84W - 6.49W min to max)? What if the PD device idles down and is using a low 0.5W power? What would happen then? Must I design to ensure that it doesn't go below that?

I'm not sure exactly what the "minimum power" in that table means, but the "maximum power" column looks like it's indicating the maximum power than the PD can expect to receive for a given classification after accounting for cable losses. For example, for class 8 the PSE will deliver up to 90 W but the PD can only expect to receive 71 W after accounting for cable losses.

In any case, section 7.4.8 "Maintain Power Signature" of the linked datasheet describes the minimum load the PD must provide in order to ensure that the PSE does not disconnect. That section describes the requirements of the PoE standard as well as the controls you have for that particular PD chip to program the MPS current pulses that the chip generates for you (AMPS_CTL and MPS_DUTY pins).

2. What would happen if the PSE can't deliver the power requirement of the PD device, so the PD device is on a higher classification than what the PSE can deliver. Would power still be given to the PD device but only the highest it can go? Where does the responsibility of protecting both equipment from damage due to lack of power? Is it the job of the PoE PD controller to outright not give power, or its further down the line where the PoE PD controller send a signal to the microcontroller/processor that not enough power is available (designer's responsibility), or simply the PSE will just straight up not give power?

The PD determines the maximum power capabilities of the PSE by the number of class events the PSE uses (and the duration of the first class event). This particular PD chip then outputs the PSE Type on its TPL, TPH, and BT pins. See section 7.3.5 and in particular Table 3, which deals with power demotion cases in which the PSE cannot supply as much power as the PD is requesting.

The PD's TPL, TPH, and BT pins provide you with the information you need to determine how much power the PD is allocated by the PSE so that you can prevent the power drawn by the PD and its load from exceeding that allocated power. The PD does have a current limit, but this is designed protect against a short circuit -- it allows up to the maximum power that the maximum classification supported by the PD allows (i.e. it isn't based on the classification the PD is requesting nor on the power allocated by the PSE). The PSE also provides protection if too much power is drawn -- e.g. TI's TPS23881 PSE, which is a Type 4 PSE designed to work with the TPS2372 PD you are looking at, has its own current limiting as well as classification-based power limiting ("port power policing").