Suggested Remedy for comment #183, 184

Ydarshan

Description

IEEE802.3at Task ForceVport ad hocSuggested Remedy for comment #183, 184October 2007Yair DarshanMicrosemi CorporationSuggested Remedy for comment #183, 184, Yair Darshan, October 2007 Page 1www.microsemi.comBackground for comments 183 and 184 Based on Vport ad-hoc meeting on October 11,2007 the group agree on the following baseline concept to address those comments:– (1) PD Model is constant power model Iport=Ppd/Vport– (2) Iport_avg and Iport_peak are function of Vport at PD input/PSE output– (3) Ipeak/Iavg = Ki_CLASS determined by the specifications (design constrains).– (4) Specification shall support legacy PD peak current numbers as presented in table 33-12 item 4. As a result of the above baseline, the following text corrections are proposed to meet (1) to (4) objectives.Suggested Remedy for comment #183, 184, Yair Darshan, October 2007 Page 2www.microsemi.comEq-2 is just IEEE802.3 Draft D0.9 StatusInformative..It is better to allow higher Idc for Ki ⋅(Vpse−Ki ⋅Iport_avg⋅Rc)class class(Eq−2) Kp =class 1,2 and to use the same class(Vpse−Iport_avg⋅Rc)0.4/0.35 ratio for all classes(Eq−3) Iport_ peak = Ki ⋅Iport_avgclass2Vpse−(Vpse −4⋅Ppd _avg⋅Rc)(Eq−1) Iport_avg =2⋅RcShown in the specification or derived directly from itCalculated values based on PD constant power equationsSuggested Remedy for comment #183, 184, Yair Darshan, October 2007 Page 3www.microsemi.comKey inputs Ki_class=Ipeak ...

Informations

• Publié par: Opsue
• Nombre de lectures: 24
• Langue: English

Extrait

IEEE802.3at Task Force

Vport ad hoc Suggested Remedy for comment #183, 184

October 2007

Yair Darshan Microsemi Corporation

Suggested Remedy for comment #183, 184, Yair Darshan, October 2007 Page1 www.microsemi.com

Background for comments 183 and 184

Based on Vport adhoc meeting on October 11,2007 the group agree on the following baseline concept to address those comments: – (1) PD Model is constant power modelIport=Ppd/Vport – (2) Iport_avg and Iport_peak are function of Vport at PD input/PSE output – (3) Ipeak/Iavg = Ki_ CLASS determined by the specifications (design constrains). – (4) Specification shall support legacy PD peak current numbers as presented in table 3312 item 4.

As a result of the above baseline, the following text corrections are proposed to meet (1) to (4) objectives.

Suggested Remedy for comment #183, 184, Yair Darshan, October 2007 Page2 www.microsemi.com

Eq2 is just IEEE802.3 Draft D0.9 Status Informative.. It is better to allow higher Idc for Ki(Vpse Ki Iport_avg Rc) class class (Eq−2)Kp= class 1,2 and to use the same class pse−Iport_avg⋅Rc) 0.4/0.35 ratio for all classes

(Eq

3)

(Eq−1)

Iport_

Iport_avg=

Shown in the specification or derived directly from it Calculated values based on PD constant power equations

_

2 pse−4⋅Ppd_avg⋅Rc) 2⋅Rc

Suggested Remedy for comment #183, 184, Yair Darshan, October 2007 Page3 www.microsemi.com







Key inputs Ki_class=Ipeak/Iavg.Ki_class determines Ipeak, Ppeak. Kp_class is a function of Ki_class, Vpse and Iport_avg. It is not a constant actually it is a complex variable. – Kp transforms current to power in the PD constant power equations. Using only Ki as proposed in the Table below allows meeting all our objectives

Table TBD1

www.microsemi.com

Ki Kp

1.350 1.300 1.250 1.200 1.150 1.100 1.050 1.000

Ppd[Watts]

Suggested Remedy for comment #183, 184, Yair Darshan, October 2007 Page4

Suggested Text changes – Comment #183,184

Suggested Remedy principles:

For maintaining consistency with Table 3312 item 4 and 33.2.8.4: 1. Table 3312 item 4: Change symbol from “Iport” to “Iport_peak” 2. Replace the equation “Pport_max/Vport” with “Iport_peak as defined by Table 3312 item 4”. 2. Add the text “Iport_peak is approximated by Eq3” 3. Add equation Eq3, Eq1 and Ki_class constants from Table TBD1 to table 3310 in separate column. 4. Update accordingly 33.2.8.4 and 33.2.8.4a by removing “17.6W/Vport” and referring to Eq1 and Eq3. 5. Update 33.2.8.6 to reflect the above changes. 6. See attached word document for detailed remedy

Suggested Remedy for comment #183, 184, Yair Darshan, October 2007 Page5 www.microsemi.com

Suggested Remedy Text for comment #183, 184

See attached word document for full description of text changes.

Suggested Remedy for comment #183, 184, Yair Darshan, October 2007 Page6 www.microsemi.com

www.microsemi.com

Additional Information

Suggested Remedy for comment #183, 184, Yair Darshan, October 2007 Page7

Issues to be solved with the original suggested remedy

Ipeak used in Table 3312 item 4 contains margins for class 1 and 2 i.e. Ki>0.4/0.35.Hence the equation as proposed tobe used needs some modifications to present Ipeak as function of Ki_class. – The actual PD constant power model is not pure constant power model since efficiency is not 100%. This is the reason for the margin taken for Class 1,2 at Iport_eak per table 3312 item 4.

Ipeak numbers per table 3312 item 4 can not be changed due to the above rational and to keep legacy PDs compliant.

Suggested Remedy for comment #183, 184, Yair Darshan, October 2007 Page8 www.microsemi.com

1 Simplified PD constant Power Model Assuming efficiency of PD is100% Vpse−Vpd Ppd = Iport Rc Vpd Rc 2 0.5 Vpse+(Vpse−4⋅Ppd⋅Rc) Vpd= Vpse 2 Ppd=Constant Vpd I=P/Vpd 2 0.5 Vpse−Vpd Vpse−(Vpse−4⋅Ppd⋅Rc) Iport= = Rc2⋅Rc Iport_peak Ki= Iport_avg Iport_peak=Ki⋅Iport_avg Vpd_peak=(Vpse−Iport_peak⋅Rc) Vpd_avg=(Vpse−Iport_avg⋅Rc) Ppd_peak Vpd_peak⋅Iport_peak Kp= = = Ppd_avg Vpad_avg⋅Iport_avg 2 0.5 Vpse−(Vpse−4⋅Ppd_avg⋅Kp⋅Rc) class Iport_peak= ⋅(− ⋅_⋅) 2 Ki Vpse Ki Iport avg Rc ⋅Rc Kp= or (Vpse−Iport_avg⋅Rc) 2 0.5  Vpse−(Vpse−4⋅Ppd_avg⋅)  Rc Iport_peak=Ki class   2⋅Rc  

www.microsemi.com

Suggested Remedy for comment #183, 184, Yair Darshan, October 2007 Page9

Actual PD constant power model Assuming efficiency of PD is avariable,eff: Pout Pout eff= = Ppd Pout+Ploss Pout⋅(1−eff) Ploss= eff Pout⋅(1−eff) Reff= 2 eff⋅Iport Vpse−Ipd⋅Rc−Ipd⋅Reff=V P = Ipd 2 0.5 Vpse−(Vpse−4⋅P⋅Kp_class⋅(Rc+Reff)) V Ipeak= 2 2⋅(Rc+Reff) V−V⋅Vpse+P(Rc+Reff)=0 or 2 0.5 Vpse−(Vpse−4⋅P⋅(Rc+Reff)) 2 0.5 V=  Vpse−(Vpse−4⋅P⋅(Rc+Reff))  Ipeak=Ki 2 class   2⋅(Rc+Reff)   Vpse−V Ipd= Rc+Reff For each Ppd we need to find its Reff, P numbers on order to get Vpd and Ppd. Next we have to use Ki, Kp techniques as in the simplified model Vpd=V+Reff⋅Ipd

www.microsemi.com

Suggested Remedy for comment #183, 184, Yair Darshan, October 2007 Page10

Efficiency v.s. Power

In conventional DC/DC converters efficiency is optimized near max. power. If output power is decreased, efficiency is decreased due to the growing effect of constant losses that are not function of the load power (housekeeping etc.) As a result, the simplified model is good enough at max. power range. At lower power range additional design margin is required. Hence Table 3312 item 4 is using higher peak current than Ki=0.4/0.35 for class 1 and 2. eff

Power

Suggested Remedy for comment #183, 184, Yair Darshan, October 2007 Page11 www.microsemi.com