PoE unstable power-up #33

New Issue

Open

opened 2024-02-09 13:26:05 +08:00 by sb10q · 4 comments

sb10q commented 2024-02-09 13:26:05 +08:00

Owner

Copy Link

With some PoE switches, the power drops (visible on 12V rail) when downstream power supplies are starting. Sometimes the board does not recover and the 12V power oscillates.

With some PoE switches, the power drops (visible on 12V rail) when downstream power supplies are starting. Sometimes the board does not recover and the 12V power oscillates.

sb10q referenced this issue 2024-02-10 19:09:17 +08:00

Ethernet link drops temporarily when +8V comes on #34

linuswck commented 2024-07-05 17:34:28 +08:00

Member

Copy Link

Description

PM1202 supports the following PoE Standard with the following inrush Current limit
802.3af: 400mA over 50ms
802.3at: 400mA (Type 1 Device) 800mA(Type 2 Device) over 50ms

The office's main switch support both 802.3af and 802.3at. This explains why PoE works for any Kirdy connected onto that switch but not the smaller one.

(Edit: The smaller switch only supports 802.3af Standard)

Inrush Current Measurement

Inrush current is measured across this inductor.

Measured peak current is 3.433A@12V -> 858.25mA @48V (assuming no efficiency loss)

This clearly violates BOTH PoE specs.

Hot Fix for the Revision v0.3 Hardware

We can add soft start to LT3081(+9V Net) LDO, which receives PoE 12V Input and reduce the severity of the inrush current. Turns out the datasheet mentions such soft start circuitry by adding a capacitor at set pin and a diode.

Capacitance: 2.2uF x 3 + 0.1uF -> 6.7uF
Diode: (an unknown through hole diode)

The first current spike is reduced to ~2.5A(Org: 3.433A). Adding more capacitance seems to have diminishing return. The Kirdy board with faulty PoE can startup with this modification but the recorded current spikes still violate both PoE specs. 2.5A@12V -> 633.325mA@48V (assuming no efficiency loss)
(Note: I also tried clamping the output current with the ILIM pin of LT3081 but does not work with the current spikes.)

Solution for Next Revision

First implement the above hot fix on next rev.

1. Relocate the inrush current limiting circuitry for 12V power jack (AOSP21321 HKD3.41 per 10pcs)

• Not recommended
• Also, change C175 to 3.3uF and then R99 to 1MEG Ohm
• This circuit fails to work If you power cycle it too fast, as the capacitor have not been discharged and PMos remains ON
  

2. Use a LM317 to limit the inrush current (LM317MDTG HKD4.69 per 10pcs)
3. Use a dedicated E-Fuse IC (TPS25961: HKD5.48 per 10pcs)
4. Implement Power Sequencing with MCU

• There are enough unused pins to implement this feature

# Description PM1202 supports the following PoE Standard with the following inrush Current limit 802.3af: 400mA over 50ms 802.3at: 400mA (Type 1 Device) 800mA(Type 2 Device) over 50ms The office's main switch support both 802.3af and 802.3at. This explains why PoE works for any Kirdy connected onto that switch but not the smaller one. (Edit: The smaller switch only supports 802.3af Standard) # Inrush Current Measurement Inrush current is measured across this inductor.   Measured peak current is 3.433A@12V -> 858.25mA @48V (assuming no efficiency loss) This clearly violates BOTH PoE specs. # Hot Fix for the Revision v0.3 Hardware We can add soft start to LT3081(+9V Net) LDO, which receives PoE 12V Input and reduce the severity of the inrush current. Turns out the datasheet mentions such soft start circuitry by adding a capacitor at set pin and a diode.   Capacitance: 2.2uF x 3 + 0.1uF -> 6.7uF Diode: (an unknown through hole diode) The first current spike is reduced to ~2.5A(Org: 3.433A). Adding more capacitance seems to have diminishing return. The Kirdy board with faulty PoE can startup with this modification but the recorded current spikes still violate both PoE specs. 2.5A@12V -> 633.325mA@48V (assuming no efficiency loss) (Note: I also tried clamping the output current with the ILIM pin of LT3081 but does not work with the current spikes.) # Solution for Next Revision First implement the above hot fix on next rev. 1. Relocate the inrush current limiting circuitry for 12V power jack (AOSP21321 HKD3.41 per 10pcs) - Not recommended - Also, change C175 to 3.3uF and then R99 to 1MEG Ohm - This circuit fails to work If you power cycle it too fast, as the capacitor have not been discharged and PMos remains ON  2. Use a LM317 to limit the inrush current (LM317MDTG HKD4.69 per 10pcs) 3. Use a dedicated E-Fuse IC (TPS25961: HKD5.48 per 10pcs) 4. Implement Power Sequencing with MCU - There are enough unused pins to implement this feature

SDS2504X Plus_JPG_3.jpg

115 KiB

image.png

9.8 KiB

gkasprow commented 2024-07-05 17:42:25 +08:00

Owner

Copy Link

We already have inrush protection (R62)
Maybe increase this resistor to 22Ohm

We already have inrush protection (R62) Maybe increase this resistor to 22Ohm

linuswck commented 2024-07-05 17:54:08 +08:00

Member

Copy Link

We removed the protection circuitry on rev0_3(#24) to use the PM1202 PoE Power Module instead of AG5300 as PM1202 claims to have inrush current limit already.

We removed the protection circuitry on rev0_3(#24) to use the PM1202 PoE Power Module instead of AG5300 as PM1202 claims to have inrush current limit already.

gkasprow commented 2024-07-05 17:58:47 +08:00

Owner

Copy Link

AG5300 also have inrush protection :)

AG5300 also have inrush protection :)

Sign in to join this conversation.

No Branch/Tag Specified

Branches Tags

master

Labels

Clear labels

No items

No Label

Milestone

Clear milestone

No items

No Milestone

Projects

Clear projects

No project

Assignees

Clear assignees

No Assignees

3 Participants

Notifications

Subscribe

Due Date

The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Dependencies

No dependencies set.

Reference: sinara-hw/kirdy#33

No description provided.