PoE based Device - Short high voltage spike issue damaging devices

[Infan Neraysh]

We have designed an PoE based device - Getting powered using PoE (Power Over Ethernet). Which is an PD with integrated DC/DC converter with isolated flyback topology to generate 3.3V from 48V(PoE input).

We are facing an fatal issue, When we use some PoE Injectors to power our board, during RJ45 hot plug or AC side hot plug, we observed some of the devices in our board are getting damaged and observing short in 3.3V rail. All those devices are powered using the 3.3V (Primary power supply).

we are observing a short high voltage (16V pk-pk pulse for 200ns) pulse in the 3.3V, we are thinking it might be the cause for the damage. whereas we couldn't suppress this voltage spike, it is seen both in GND and 3.3V. We have tried below things and still not able to suppress the spike,

TVS diode and Schottky Diode to 3.3V and Capacitor coupling to Transformer o/p and i/p
Zener Diode with Schottky to 3.3V
Load with parallel RC circuit in 3.3V
470uF capacitor in 3.3V
470uF capacitor in 52V line side (board was not turned on)
pF range capacitor to 52V line
Inductor 100uH to 52V input line
Inductor 100uH to 52V input line and 10uH inductor to GND
TVS diode to 3.3V and 52V
Common mode choke to Input line
470uF to TP-Link side
Multiple Ethernet cables
UVLO RC circuit
GND_VPORTN (PoE Return) & GND_POWER (PD devide reference ground) capacitive coupling
Soft start capacitor tweaking

Any Expert suggestions are very much appreciated.

Thanks in advance.

 

[Ralston18]

Figure out the cause of the 16V pk-pk pulse for 200ns pulse/spike.

Reverse engineer it. That is pretend you want to do something like that. How would you do it? Then design that circuit via the inputs you have.

Compare that "pulse" circuit to your schematics - learn what is causing the pulse to help determine how to stop the pulse or make it suppressable.

May need some redesign.

Just thinking out loud here......