lt4294 behavior with passive POE adaptor

Hello,

The LT4294 will operate with "Passive 48V" PoE. I highly recommend using an IEEE compliant PSE for reliability and safety.

Applying 48V onto the ethernet cable without detection or classification looks a Type 1, 13W PSE, since Classification was optional for these early devices. Once the port voltage exceeds VHSON, the LT4294 will turn on and inrush the downstream bulk capacitance. The /T2P pin will indicate the PD has been allocated 13W. If the port voltage drops below VHSOFF, the downstream circuitry will be disconnected from the port.

Best Regards,

Eric

Hi Eric,

Thanks for you replay.

To go further we have 30% of fails, as it seem related to the LT4294 circuit. I'm attaching schematics(hope it visible, otherwise please let me know how to send) and will very appreciate your feedback. Here are some more details:

Till now we mostly used the Ubiquity POE-54-80W passive 48 POE and PHIHONG POE60U-1BT active POE supplies. After few months of successful working boards suddenly started to fail. By fail I mean PWRGOOD output became constantly low and gate level voltage became equal to the input voltage(while at time of correct conditions it should be ~10V higher than the input), therefore not opening the N-CH transistor.

I have disconnected all loads from the output in order to be sure that no current sinking appears during the handshaking process, however it did not help. Additionally, I noticed that LT4294 became very hot when the input voltage falls below ~30-35V. For example if I apply power with lab power supply (in this case it acts as passive POE) with 48V and starting to decrease voltage till it below ~30-35V the consuming current starts to be constant around 40mA while LT4294 starts to be hot.

Explained above scenario repeats on all boards.

Please help to solve our problem.

AndreyPDF

Hi Eric,

I would appreciate your opinion in relation to the presented problem. If any clarification required I would be glad to provide them.

Thanks,

Andrey

Hello Andrey,

I don't see anything with your application that would cause this issue that you are seeing. Is there anything in your circuit which could be overstressing the hotswap FET? For these boards that are having difficulties, can you try replacing Q6 and seeing if the solves the issue? 

Can you confirm the part number for Q6? Was its Safe Operating Area (SOA) tested? Is PWRGD being used to enable the downstream circuitry so it is not drawing current while the PD is inrushing the bulk capacitance? How much capacitance is connected downstream? I only see C106 at 100uF on your schematic.

The LT4294 becomes hot during your test because you lab supply is not acting like a normal PoE injector. The PD is drawing a Classification Signature current, which can be as high as 40mA. An IEEE 802.3-compliant Power Sourcing Equipment (PSE) will Classify the PD to determine the requested power. Classification usually takes a few hundred milliseconds, so not much heat is generated. The PSE will then provide between 37V and 57V, this turns off the Class Signature current and the PD will then inrush its downstream capacitance. When inrush is completed, PWRGD will no longer be pulled to GND. Also, the AUX pin will disable classification if it exceeds VAUXT or approximately 6.3V.

Best Regards,

Eric

Hi Eric,

Thanks for your comments.

The hot swap FET Q6(is actually Q5) is FDMC86102LZ.

I sent you initially simplified schematics. Here is the schematic with complete power stage of the board. The downstream capacitance that LT4294 sees is actually pretty small and about 122uF(mostly C106 and C100). PWRGOOD pin is used to enable downstream power supplies, except the one(U25) which consumes constantly up to 10mA, much less then the 350mA that allowed during classification.

In order to verify the problem source I have removed PWRFIL1 from the boards were problem exist. This is a reason I sent you simplified schematics first.

I will replace the hot swap FET as you suggested and will let you know the result.

Thank you,PDF

Andrey

Hi Andrey,

122uF should be OK for this application. The LT4294 will inrush this capacitance with approximately 60mA for about 120ms. I have not tested FDMC86102LZ's SOA performance, but the FDMC86102 performs quite well and may be a better alternative for your application. The SOA curve of the FDMC86102LZ's datasheet indicates it may handle this inrush current, but have you tested it in lab? I've also noticed your gate resistor is lower than recommended, R115 should be 3.3k and not 2.2k.

Best Regards,

Eric

Hi Eric,

Thanks a lot for your feedback!

Replacing FET did help. Problem was eliminated on few boards were the problem was initially found, however the reason that lead to the fall of the FET is not understandable yet. Do you have any suggestion regarding what are critical parameters in this particular case. As we see FET characteristic are pretty good, however somewhat cause it blow?

Thanks,

Andrey

Hi Eric,

Thanks a lot for your feedback!

Replacing FET did help. Problem was eliminated on few boards were the problem was initially found, however the reason that lead to the fall of the FET is not understandable yet. Do you have any suggestion regarding what are critical parameters in this particular case. As we see FET characteristic are pretty good, however somewhat cause it blow?

Thanks,

Andrey

Hi Andrey,

It sounds like that FET failed for some reason, and it could be related to SOA. ADI recommends using a FET that has been lab tested for high SOA; that includes the part numbers in our documentation material and the FDMC86102.

Best Regards,

Eric 

Thanks for you help!