Issue with LTC4267 for Power over Ethernet

[reportingsjr]

Hi,
I designed a board to use power over ethernet (PoE) using the LTC4267-3 (http://www.linear.com/product/LTC4267-3) to do all of the required handshaking plus be a 48V -> 5V isolated dc-dc converter. I almost exactly copied LT design note 361 (http://cds.linear.com/docs/en/design-note/dn361f.pdf) without the battery part.

I built up the board and almost everything is working on it except for the power over ethernet handshaking! I was able to inject 48V in to the power input and got 5V out so the dc-dc convertor and LTC4267-3 are working fine in that respect, but when I connect the board to a PoE switch it refuses to supply 48V.

I went ahead and attached a scope probe across the positive and negative PoE inputs and it looks like the switch tries to start identification and then just stops for some reason. I tried two different PoE sources (one a cisco switch and the other a tp-link PoE injector) and got the same results with both. I also tested both PoE sources with known working devices (arduino PoE shields) and they work fine. Anyone know what is up?

I attached pictures of the scope shot (note that PoE is actually -48V) and part of my schematic.

Thanks!

[Jay_Diddy_B]

Hi,

Look carefully at C5, the negative end of C5 should be connected to Pout (GNDA) on the IC. The capacitor is switched in after the classification.

Regards,

Jay_Diddy_B

[reportingsjr]

Ahhh, I see the difference looking at the LTC4267-3 datasheet and dn361. So should I cut the trace from the negative end of C5 to VPortN and tie it to PGND?

Is this an issue with dn361 then?

BTW Jay_Diddy_B, I met you at dayton hamvention last year. I had that HP electron load that you swapped that 6632A for. Thanks for that! I has really come in handy lately with some power supply stuff I have been doing. I did a few of the standard mods to it (quieter fan, binding posts on the front, cleaned it out, etc) and it has been awesome.

Are you going to be at hamvention this year again? It is a bit early, but I figured I would ask!

[Jay_Diddy_B]

Ahhh, I see the difference looking at the LTC4267-3 datasheet and dn361. So should I cut the trace from the negative end of C5 to VPortN and tie it to PGND?

Is this an issue with dn361 then?

Yes, to the reconnection.

Yes, to there is an issue with dn361. The 0.1uF capacitance is defined in the IEEE-802.3af specification.

Regards,

Jay_Diddy_B

[reportingsjr]

I went ahead and connected the negative pin of C5 to Pout and it works a treat now. Thanks for the help!

[Anand Ram]

Hi All,

We have designed the board with LTC4267 PD device isolated topology to output 5V. We referred to the LT design note (https://www.analog.com/media/en/reference-design-documentation/design-notes/dn361f.pdf). But we are unable to get 5V. The output is stabilizing at 2.8V. I have attached a schematic here for the reference.

Please, anyone, check and let us know if missed anything here.

[Jay_Diddy_B]

Hi All,

We have designed the board with LTC4267 PD device isolated topology to output 5V. We referred to the LT design note (https://www.analog.com/media/en/reference-design-documentation/design-notes/dn361f.pdf). But we are unable to get 5V. The output is stabilizing at 2.8V. I have attached a schematic here for the reference.

Please, anyone, check and let us know if missed anything here.

Anand,

You missed a couple of things:



You are missing the 1uF capacitor C12 in the original design.

The reference on the low voltage versions of the 431 circuit is 1.24V, the divider ratio should be 3:1 not 1:1 like you have.

Regards,
Jay_Diddy_B

[moonsunshoot]

Hi,

I am facing the issue in output, designed using LTC4267-1 IC for 12V output design with transformer PA1135NL. But I am getting only 3.3V out. I have attached the design here, can anyone please help me to get the correct output 12V.

[Leo Bodnar]

Output voltage is defined by R23/R29 divider.