Hi Yair, I have less concern about the PSE than I do about having a PD PI Runbalance specification.  Testability is the primary concern, but there are other concerns as well.  I've attached some slides which address a new problem I'm looking into concerning a PD PI Runbalance requirement.  The slides discuss how a PD PI Runbalance spec based upon a diode bridge can fail a FET-based bridge which would not violate the overall Iunbalance.  (These are preliminary due time constraints.) Regards, Ken On 5/7/2014 6:51 PM, Darshan, Yair wrote: No this is not what I meant.   The best way in my opinion  to discuss about technical solution is to sketch a complete solution i.e. present a solution that starts from simple high level requirements such: -Definition that is not depend on implementation. -Unbalance requirement (current or resistance or else) is one number, worst case, specified in the UUT operating state/range -Can be described by solid mathematical _expression_ or other logical way -Easy to explain -Easy to test   I'll try to show one example for PSE if I'll have time for Thursday meeting. If not, we will have time on IEEE week.   Meanwhile consider the following EXAMPLE: Use case #1: PD is single load. (we need to check other use cases and generate set of requirements that doesn’t contradict each other. It is possible if we replace ideas with mathematical expressions so we can plug there physical limitations, backward compatibility considerations that can be described by numbers etc.)   1.       PD is brought to ON state 2.       PD operating input voltage is 36 to 57. 3.       PD min- max total load current (Itotal)  is 10mA to TBD 4.       The specification will say the worst case PD PI Pair to Pair DC input resistance is e.g. 15%. (please note that in single load PD, Iunb=P2PRUNB*Itotal. This is the responsibility of the PD vendor to make sure that he doesn’t cross his P2PRUNB no matter where this point is on his P2PRUNB curve so it doesn't matter if it is linear curve or other .                 THE PD PI P2PRUNB is number that is derived from End to End Channel Resistance number that we define what we want it to be to optimize                cost of the entire system component. Same PSE PI P2PRUNB. 5.       Etc. continue until there is clear solution. I guess we all know what we want. We just need to bring complete solution so we can discuss it.   I'll try to follow this concept and see if I can have something solid for discussion.   Yair   From: Ken Bennett [mailto:ken_bennett@xxxxxxxxx] Sent: Wednesday, May 07, 2014 11:42 PM To: Darshan, Yair; STDS-802-3-4PPOE@xxxxxxxxxxxxxxxxx Subject: Re: [802.3_4PPOE] Fwd: Re: [802.3_4PPOE] IEEE802.3bt End to End Channel Pair to Pair Resistance Unbalance - Meeting material for Thursday May 8, 2014 meeting.   Hi Yair, We can discuss the points of disagreement as you indicate.  One initial point I'd like to point out is the following: (From the slide:) PD PI Runbalance testing complexity:       Has a time-varying load, so Sequential measurements won’t work universally unless special operating modes or internal probing is included Yair:  (Disagree. When power on, look for DC behavior ,or specify DC behavior)    Ken:  If the first step of an PD PI Runbalance test is to require a DC load behavior, you have already crossed a boundary forcing PD designers to implement a special operating mode that may not otherwise exist.  If testing worst case, then the DC mode has to be at full power, and full power without a varying load may not even be possible.  Regards, Ken On 5/7/2014 3:45 PM, Darshan, Yair wrote: `Hi Ken,` `Thanks. See my comments up to page 5 due to lake of time.` `We can discuss more at the IEEE meeting.` `Yair` ` ` `-----Original Message-----` `From: Ken Bennett [mailto:ken_bennett@xxxxxxxxx] ` `Sent: Wednesday, May 07, 2014 8:42 PM` `To: STDS-802-3-4PPOE@xxxxxxxxxxxxxxxxx` `Subject: [802.3_4PPOE] Fwd: Re: [802.3_4PPOE] IEEE802.3bt End to End Channel Pair to Pair Resistance Unbalance - Meeting material for Thursday May 8, 2014 meeting.` ` ` `Hi All,` `I've attached some slides relating to the Runbalance PI Specification.` `Comments are welcome.` `Best Regards,` `Ken` ` ` ` `