Thread Links | Date Links | ||||
---|---|---|---|---|---|
Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
Hi all,
After the last ad hoc meeting I felt that the views of some of the guys is not clear to me when I try to confront them with the two models discussed.
So to start a discussion in order to get consensus on next Wednesday I have summarized the points made by some of the guys.
Please let me know if I capture those view correctly so we can continue the discussion afterwards.
1. We would like that the standard will support system scenario in which the PSE generate transient voltage that cause ILIM and at the same time the PD exercise its right to use peak current of up to 400mA for 50msec.
1.1 In this case we would like to guarantee that the port will stay ON for TLIM duration what ever this number is.
1.2 We agree that this is a corner case which its probability to happen is very low however we still want to support it.
1.3 We also agree and understand that regardless of the PD model (whether if it is constant current or constant power)
We can not achieve goal 1.1 due to the physical fact that if the PD takes 400mA peak and at the same time PSE generating transient voltage which cause ILIM=400mA then the net charging current in the constant current model is zero so there is no capacitance charging so TLIM will be >>50ms and port will turn OFF.
Proof: TLIM=C*dv/(ILIM-Ipd) if ILIM=Ipd=400mA TLIM= Infinit! >75msec
1.4 It was noted that we can't change the 802.3af spec so we will add some informative note to recommend the PD vendor to not work with current above 350mA.
1.5 Other view said regarding (1.4) that we can recommend what to do but we are missing the point why we allowed 400mA peak at 50ms, 5% duty is to allow
utilizing the full 12.95W at the PD, Otherwise if we use max 350mA peak power will be <<12.95W pending the current ripple peak to peak.
1.6 Other view regarding (1.4) and (1.5) and (1.2) said that we don’t have to support corner case and let's leave it to implementations to handle by encouraging some kind of innovation..(we have infinite corner cases and we would like not to pay for solving them all.)
1.7 Other view said, let's use PD constant power model from the following reasons:
Proof: TLIM=C*Dv/(ILIM-Ppd/Vpd). Icharge=ILIM-Ppd/Vpd
According to this model if scenario (1) happens then:
-ILIM is 400mA constant at the PSE as in all cases.
- Ipd is 400mA only if Vpd=36V but if Vpd>36V then Ipd<400mA due to the fact that in this model Ipd=Ppd/Vpd!
- In this case it 100% sure that after PD load transient ended, Icharge=ILIM-Ppd/Vpd >0 SO TLIM is not infinite.
- There is only one corner of corner of scenario that in both models Icharge=0 when the load transient in the PD takes the full 50msec.
So according to (1.7) the constant power model is the best.
1.8) Additional view: We have to divide the big question of PD model and what is TLIM to two orthogonal parts:
First question is what is the right PD model.
2nd question: what is the TLIM that increases the chances of the system to operate in most scenarios.
For example: We can say, in 802.3af the minimum TLIM is 50ms in order to increase system operation probability.
And the PD model is constant power because it is really the physical (and not! typical as some said) model and 802.3af
has already use it as the only model.
-------------------------------
Please read the above and let me know if I captured correctly all views.
After wards let's start to discuss each one of them.
Thanks
Yair
Darshan Yair
Chief R&D Engineer
Analog Mixed Signal Group
Microsemi Corporation
1 Hanagar St., P.O. Box 7220
Neve Ne'eman Industrial Zone
Hod Hasharon 45421, Israel
Tel: +972-9-775-5100,
Cell: +972-54-4893019
Fax: +972-9-775-5111
E-mail: <mailto:ydarshan@microsemi.com>.