I’m creating a PD which has two Ethernet ports.
Device will be able to receive PoE either by any of the two ports or by the two simultaneously (in that case load sharing would be expected).
Under normal condition device should consume around 0.8W. Short term (few seconds) spikes up to 4W are however possible and normal.
My initial idea was to implement this using 2 x TPS2378 and tying their RTN pins together to GND.
However by reading the two docs below, I understand this would probably not be the right solution (because of the back-bias) :
- App report : www.ti.com/.../slva625a.pdf
- E2E post : http://www.ti.com/lit/an/slva625a/slva625a.pdfhttps://e2e.ti.com/support/power_management/power_interface/f/204/p/271886/949905#949905
Here are my questions :
Q1/ Is inserting a diode between GND and RTN enough to achieve my requirement ? What are the required characteristics of these diodes ?
Q2/ From the app report slva625 , I understand an additional mosfet could be an alternative. This would increase complexity but maybe lead to a better efficiency. However compared to the app note, I don’t need in my case to always have both ports providing power, since one in enough. What is the correct way to wire the gate from the mosfet in that case ?
Q3/ In the case where both ports would be providing power, it seems 0.8W consumption could trigger the MPS disconnect. Is there a simple/efficient way to fix this without implementing the complete feedback loop with an optocoupler ?
Thanks !
Matthieu