vport ad hoc current limits may 2007

20
1 Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva Vport ad hoc Current Limits May 2007 Fred Schindler Cisco Systems Four ad hocs with an average attendance of 16 people since the last IEEE meeting. People that attended since the last IEEE meeting are shown in bold. Andrew Smith Power Integration Anoop Vetteth Cisco Systems Bill Delveaux Cisco Systems Brian Buckmeir Bel Chad Jones Cisco Systems Christen Beia ST Microelectronics Clay Stanford Linear Technology Dan Dove HP Procurve Networking Daniel Feldman Microsemi David Law 3COM David Lucia SIFOS Derick Koonce Independent Ferdinando Lari ST Microelectronics Frank Yung SystemX Fred Schindler Cisco Systems Geoff Thompson Nortel Networks Gianluca Sanità Nokia Siemens Network Helen Kastner Cisco Systems Hugh Barrass Cisco Systems James Blanc Indep. VC investigation Jean Picard Texas Instruments Jeff Heath Linear Technology John Jetzt Avaya Keith Hopwood Phihong Ken Bennett SIFOS Martin Patoka Texas Instruments Matthew Landry Silicon Labs Michael Altmann Akros Silicon Pavlick Rimboim Microsemi Ramesh Sastry Cisco Systems Raul Lozano Pulse Riccardo Russo ST Microelectronics Sajol Ghoshal Akros Silicon Taufique Ahmed Akros Silicon Thong Nguyen Maxim Thuyen Dinh Pulse Tim Parker Nortel Networks Val Rybinski Siemon Wael Diab Broadcom Yair Darshan Microsemi Youhoa Xi National Semiconductor

Upload: others

Post on 05-Feb-2022

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Vport ad hoc Current Limits May 2007

1Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

Vport ad hoc Current LimitsMay 2007

Fred SchindlerCisco Systems

Four ad hocs with an average attendance of 16 people since the last IEEE meeting. People that attended since the last IEEE meeting are shown in bold.

Andrew Smith Power IntegrationAnoop Vetteth Cisco SystemsBill Delveaux Cisco SystemsBrian Buckmeir BelChad Jones Cisco SystemsChristen Beia ST MicroelectronicsClay Stanford Linear TechnologyDan Dove HP Procurve NetworkingDaniel Feldman MicrosemiDavid Law 3COMDavid Lucia SIFOSDerick Koonce IndependentFerdinando Lari ST MicroelectronicsFrank Yung SystemXFred Schindler Cisco SystemsGeoff Thompson Nortel NetworksGianluca Sanità Nokia Siemens NetworkHelen Kastner Cisco SystemsHugh Barrass Cisco SystemsJames Blanc Indep. VC investigationJean Picard Texas Instruments

Jeff Heath Linear TechnologyJohn Jetzt AvayaKeith Hopwood PhihongKen Bennett SIFOSMartin Patoka Texas InstrumentsMatthew Landry Silicon LabsMichael Altmann Akros SiliconPavlick Rimboim MicrosemiRamesh Sastry Cisco SystemsRaul Lozano PulseRiccardo Russo ST MicroelectronicsSajol Ghoshal Akros SiliconTaufique Ahmed Akros SiliconThong Nguyen MaximThuyen Dinh PulseTim Parker Nortel NetworksVal Rybinski SiemonWael Diab BroadcomYair Darshan MicrosemiYouhoa Xi National Semiconductor

Page 2: Vport ad hoc Current Limits May 2007

2Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

Agenda

• SELV correction.

• Current limits.

• Next step.

Page 3: Vport ad hoc Current Limits May 2007

3Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

SELV Correction

• IEC 60950, 3rd edition, 1999-041.2.13.4 DC VOLTAGE: The average value of a voltage (as measured by a moving coil meter) having a peak-to-peak ripple not exceeding 10 % of the average value.

• IEEE 802.3-2005, Table 33-6 PSE PI parametersV_open; Vpp; Min: 1.9 V;Max: 10% of the average value of VPort, 44V < VPort < 60V.

• 57 V x 10% = 5.7 Vpp, 57 V + 2.85sin(ωt)57 + 2.85 = 59.85 V, ~ 60 Vpeak_value

Page 4: Vport ad hoc Current Limits May 2007

4Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

Proposal corrected for SELV

• Propose a PSE PI voltage limit, for transients present more than 30 µs, of 7.6% below the PSE Vport_min level for less than a period of 250 µs and 10% above the Vport_max level.

46.2

60

50

802.3at

Duration ( s

57

30 2500

Normal Operating

Range

Transient

SELVTransient

Remove this text and continue using existing SELV guidance.

Page 5: Vport ad hoc Current Limits May 2007

5Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

Motion

Move that:

The IEEE 802.3at Task Force adopt presentation schindler_5_28_07.pdf slides 4 to be incorporated in the next P802.3at draft.

M: Fred Schindler

S: Yair Darshan

All Present 802.3 Voters

For: For:

Against: Against:

Abstain: Abstain:

Page 6: Vport ad hoc Current Limits May 2007

6Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

IEEE 802.3 vs IEEE 802.3at

• The legacy standard reused the in-rush current limits to address the power-on current limits.

• The proposed IEEE 802.3at standard:Reuses the legacy in-rush current limits.

Raises power-on currents.

Limits PD di/dt rates during power-on.

Opens up the design space to allow:

Scaled legacy current thresholds

Aggressive fold back

An energy based limit

Page 7: Vport ad hoc Current Limits May 2007

7Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

System In-rush and Power-on

Parameter 802.3 802.3at max.

PortVPortP

350400

400450

ILIM min. max. PortV

PortP350400

New

ICUT min. PortVPortP

VPSE

time

PPort

PD limits di/dt

Power can be classified using

layer-2.

IEEE 802.3 IEEE 802.3atPass Element Off/On/, I-limit Off/On or I-limit

VPSE

time

<= 15.4 W PPort

PSE limits to IInrush

PD limits di/dt

IEEE 802.3 IEEE 802.3atPass Element Off/On, I-limit Off/On or I-limit

Layer-1 classification Layer-2 classification

<= 15.4 W

Mutual Identification Achieved

In-rush Power-on PSE limits to IInrush

In-rush Power-on

Page 8: Vport ad hoc Current Limits May 2007

8Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

System Concerns being addressed by ILIM

• Situations that lead to a PSE dv/dt rate that causes excess PD current demand.

Multiple PDs reducing their load.

A PSE supply voltage change. ex/ Switching in a new power supply to deal with a power supply failure.

Page 9: Vport ad hoc Current Limits May 2007

9Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

Collected Worst-case Current Limits

48 ports, Rdistribution = 80 mΩ, Rport = 0.9 Ω, Rchannel = 0.8 Ω (short),CPD_ON = 5 µF (no ESR), SOA for 805 size resistor 1 Ω in parallel with 1 Ω,1 A fuse (1.98 A2s), Generic NCH MOSFET, VDS = 10 V @ ID = 14 A

Load drop: 47/48 ports, short channel 5 µF, 29.5 W PD.PSE voltage ramp: short channel 180 µF, 29.5 W PD.

IEEE 802.3at Current Limit

0

2,000

4,000

6,000

8,000

10,000

12,000

0 0.2 0.4 0.6 0.8 1 1.2 1.4

time (ms)

Ipd

(mA

) Load dropVoltage RampSOAFig. 33C.4

IEEE 802.3at Current Limit

0

100

200

300

400

500

600

700

800

900

1,000

1,100

1,200

1.5 2 2.5 3 3.5 4 4.5 5

time (ms)Ip

d (m

A)

Fig. 33C.4 undefined, 1 ms < t < 2 ms

Page 10: Vport ad hoc Current Limits May 2007

10Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

Iport at startup, short, and transient

0.0

5.0

10.0

15.0

20.0

25.0

30.0

0.0000 0.2000 0.4000 0.6000 0.8000 1.0000 1.2000

time (ms)

Ipor

t (A

)

SOAFIG 33C.4

The SOA Curve is for a PD

• Assumes a straw man PD.This PD does not exist but could be created.Using two 1 Ω resistors in parallel to sense current.

• Most PD silicon vendors limit current at the PD.FreescaleLinearMaximNationalPower IntegrationSilicon LabsSTTIOthers?

Copper trace fuses open

PD Rsense fusesNo damage

Page 11: Vport ad hoc Current Limits May 2007

11Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

What Current Limit Should be used?

• Figure 33C.4?+ Shorter peak current time.+ A larger range of values supports the standard.- Could damage a PD we are not aware of.

• PD Rsense SOA?+ Interoperability with out damage.- Longer time to transition the PD voltage.- Tighter tolerances.

• 33.2.8.8 Output CurrentMeasurement after 1 ms ….See Figure 33C.4

• Ad hoc 100% ok withFigure 33C.4 limit.19 people

Iport at startup, short, and transient

0.0

5.0

10.0

15.0

20.0

25.0

30.0

0.0000 0.2000 0.4000 0.6000 0.8000 1.0000 1.2000

time (ms)

Ipor

t (A

)

SOAFIG 33C.4

Copper trace fuses open

PD Rsense fusesNo damage

Page 12: Vport ad hoc Current Limits May 2007

12Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

Power-on proposed compliance curve

On Curve-2, 1.6 A @ t = 10.4 mS; 920 mA @ t = 29.5 ms.

60 s

Iport

time

20 A

62.5 s TLIM_MAX= 75 ms

1.75 A

1

2

4

tportI 025.0=

3 mS

ICUT_MAX = ILIM_MIN

ICUT_MIN

SOA

Interoperable

2 A

Limited Power SourceIEC 60950

TLIM_MIN

3

0 0

Page 13: Vport ad hoc Current Limits May 2007

13Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

60 s

Iport

time

Power Removed

20 A

62.5 s TLIM_MAX= 75 ms

1.75 A

1

2

4

tportI 025.0=

3 mS

ICUT_MAX = ILIM_MIN

ICUT_MIN

SOA

Interoperable

2 A

Limited Power SourceIEC 60950

TLIM_MIN

3

Power-on0

0

Understanding the Current Limit Curve

PSEs shall turn-off power before this region is entered.

PSEs can turn-off power in this region.

PDs shall operate in this region with a static PSE voltage.

PSEs may supply a constant current of ILIM to this point to ensure interoperability.PSEs supplying a constant current of ILIM shall operate to at least TLIM_MIN.

PDs may operate in this region.

PSEs using energy based power transfermethods shall operate within this region.

Same as 802.3

The light green boundary has 45% average margin above the simulated system needs.

Page 14: Vport ad hoc Current Limits May 2007

14Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

Simplified System in Current Limit

Cdtdv

PD IC <

CIdv

PDCdt >

Short channelRch = 0, Constant IPDIEEE 802.3 TLIM = 47 msIEEE 802.3at, TLIM = 12.6 ms

Long channelRch = 20 or 12.5 Ω, Constant IPDIEEE 802.3 TLIM = 43 msIEEE 802.3at, TLIM = 10.4 ms

MINLIMchRPDVPSEV I _=−

C

LIM_MIN

PSEcurrent limiting

PSE< ILIM_MIN

This simplified case assumes the PD always draws ICUT_MIN.

Page 15: Vport ad hoc Current Limits May 2007

15Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

Simplified System in Current Limit

CIdv

PDLIM CT >

3504004457180 −

−> FTLIM µ

msTLIM 47>

Short channelRch = 0, Constant IPDIEEE 802.3 TLIM = 47 msIEEE 802.3at, TLIM = 12.6 ms 720820

5057180 −−> FTLIM µ

msTLIM 6.12>

IEEE 802.3 IEEE 802.3at

Long channelRch = 20 or 12.5 Ω, Constant IPDIEEE 802.3 TLIM = 43 msIEEE 802.3at, TLIM = 10.4 ms

LIMchMIN IRdv >

mAdvMIN 40020 ×Ω>

VdvMIN 8>

35040084449180 −−−> FTLIM µ

msTLIM 43>

492035057,57 =Ω×−== VmAVVV PDPSE

mAdvMIN 8205.12 ×Ω>

VdvMIN 3.10>

7208203.104157180 −

−−> FTLIM µ

msTLIM 4.10>

mAVVVV PDPSE 720@41,50 ==

mAVVVV PDPSE 720@48,57 ==

93% of ad hoc okay this model (13/14).

Page 16: Vport ad hoc Current Limits May 2007

16Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

IEEE 802.3 Concern

• A PD can legally draw 400 mA for 50 ms.Table 33-12, Item 4.

• A system permitting this will not ensure interoperability when the PD is drawing its maximum allowable current and a PSE voltage transient occurs.

• Proposed Solution:Recommend that no more than ICUT_MIN be drawn by a PD with a static port voltage.

IEEE 802.3at will use the correct value to begin with.

The previous IEEE Task Force may have allowed ICUT_MAX (PSE) to ensure circuits could accommodated this current value, but expected no more than ICUT_MIN to be drawn normally by the PD.

Page 17: Vport ad hoc Current Limits May 2007

17Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

PSE SOA

In-rush & Power-onIEEE 802.3

Power-onIEEE 802.3at

60 s

Iport

time

Power Removed

20 A

62.5 s TLIM_MAX= 75 ms

1.75 A

1

2

4

tportI 025.0=

3 mS

ICUT_MAX = ILIM_MIN

ICUT_MIN

SOA

Interoperable

2 A

Limited Power SourceIEC 60950

TLIM_MIN

3

Power-on0

0

Upper limit changes

A maximum change of 7V for a durationof > 250 µs is possible for aninteroperable system.Change the focus from VPSE to the voltage drop across the PSE pass element.

Page 18: Vport ad hoc Current Limits May 2007

18Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

Next Step

• The task force should review this proposal.

• Final ad hoc clean up.

• Motion to set this baseline will be made at the July Plenary.

Based on V14 Spreadsheet.

Page 19: Vport ad hoc Current Limits May 2007

19Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

2 43

2 43

PDChannel

PSE

Power Distribution

Port Control

1

Use Case: PSE load drop1. All PDs draw maximum power.

2. All but one PD reduces power demand to zero.

3. The PSE distribution voltage ramps up.

4. The remaining on PD has current injected into it.

60 – 80 mΩ0.9 – 3.2Ω

0.8 – 12.5Ω (49.3)2 Ω

0.8 – 12.5Ωn

29.5

(49.3)2 Ω29.5n

A PSE providing 50 V at 600 mA with PDs consuming 29.5 W.

(5 – 180 µF)

n

This is a don’t care when the load is dropped.

Worst-case: 5 µF (no ESR), load dropped.

Page 20: Vport ad hoc Current Limits May 2007

20Vport ad hoc IEEE 802.3at Task Force, May 2007, Geneva

Use Case: PSE power supply backup

VPSE

time

5.6 V in 2.4 ms

IPD

time

0.9 A, 5 ms, 180 uF

650 mA720 mA

57 V

51.4 V

0.7 A, na, 2.2 uF

Current, time constant Worst-case: 180 µF.