One of the biggest challenges that the IEEE802.3bt Task Force faces is how to define the behavior for new 4-pair devices when they are attached to legacy 2-pair devices.  In fact, it is such a challenge that I wrote my very first blog post about it. In this post, I want to dive a little deeper into the issues that we face.

At first glance, it might seem fairly simple to define the behavior for a 4-pair PSE that is connected to legacy 2-pair PD.  Detection and classification protocols are already defined and do not need to be changed for the 2-pair PD (with one small exception that I will talk about later).  Unfortunately, that is just about the only thing that is obvious.  Once the PSE has detected and classified the 2-pair PD, it will need defined behaviors for a variety of situations.  Here are some of the questions that I believe we still need to answer:

  1.     Do we power legacy 2-pair devices over 4 pairs?
  2.     How does a 4-pair PSE handle inrush when connected to a 2-pair PD?
  3.     How does a 4-pair PSE handle current limiting when connected to a 2-pair PD?
  4.     What does a 4-pair PSE do when a fault is sensed on one pair set?

Clearly, questions 2-4 depend greatly on the answer to question 1.  Even though legacy 2-pair PDs only receive power over 2 pairs at a given time, they are required to be able to accept power over either pair set (this is normally implemented by using a diode bridge on the input to the PD).  This means that applying 4-pair power to legacy 2-pair PDs should be possible.

 IEEE802.3bt However, accepting 4-pair power is an undefined state for a 2-pair PD and thus we must consider the implications of this carefully.

If we decide that we should power legacy 2-pair devices over 4 pairs, we now need answers for questions 2-4.  These questions encompass a lot of details—Do we current limit each pair set individually or use a total current?  Do we control inrush per pair set?  Do we only use one pair set for inrush and then turn the other pair set on?  Do we shut off both pair sets if we see a fault on one?  Do we shut off only the pair set with the fault?

*Note that many of these questions also apply to a 4-pair system in general and will need to be answered for that case no matter the answer to question 1 above.

Now, how do we define the behavior for a new 4-pair PD when it is connected to a legacy 2-pair PD?

It turns out that this is actually a little simpler than the previous case because the 2-pair PSE’s behavior is already defined by the 802.3at specification.  This means that once a 4-pair PD knows it is connected to a 2-pair PSE (which results from the PD seeing one or two fingers during classification), the 4-pair PD must behave the same way as a 2-pair PD.  The only question that remains is how to extend the mutual identification process (classification) so that 4-pair PDs can ask for more than 25.5W when connected to 4-pair PSEs, but receive the maximum possible power when connected to 2-pair PSEs.

This post originally published on TI’s Power House blog.