TPS23754: Using TPS23754 with un-reliable POE sources....

Hi Ted,

This is a very interesting topic and completely agree with you. I have found that in typical PoE applications in the past, these end equipments were deployed in very 'safe' envronments (like in a corporate building, strategically placed, away from foot traffic and in proper bundles). With the Internet of Things, PoE devices are being put into non-traditional environments like warehouses, conveyer belt apparatuses, or entertainment studios (lots of moving parts). There is a current trend for high reliability PoE PDs to have dual PoE inputs for PoE redundancy where if one PoE input is faulted, the other PoE input will provide the power/data. Is this the kind of reliability you're referring to?

Darwin,

Actually, imagine something much worse. Possibly a POE nightmare scenario. Imagine an apartment building with 100's of units in it, where the landlord was providing an Ethernet jack, with free internet for all the tenants. So, dicey CAT5 wiring already all over the place, tons of other unknown devices on the network, and then the landlord decides to "upgrade" and install POE switches, so that every jack has POE. So, now you are sharing this network with 100's of unknown PD's hanging off of it. Now someone comes along and plugs a device that YOU designed into this mess of a network. How does it behave? At some point your device needs to be smart enough that if the infrastructure is super-duper crappy, and can't support your PD, it shouldn't power up at all. It shouldn't sit there and oscillate between powering up, browning out, and powering up again and again. Also, if you are supporting the ability to run off a separate DC jack, as an alternate power method, the unit shouldn't sit there and negotiate the POE, work for awhile, have the connection crap out, and then flap back to the DC supply over and over. The document TPS23753-Support Features.pdf seems to document lots of these connection ORing solutions, and ways to gang things up to be smart, but it really doesn't show the basics of how do I setup the device so it won't power up at all if the POE is good, but not good under load.

Any thoughts on how to set-up a circuit so that the threshold for power-good is higher than normal, so it just doesn't come on at all? The only one I have been able to think of so far is to use a microcontroller to forcibly be able to disable the POE supply if on a DC jack using the stuff shown on page 13 of that document. Another thing is to just not use the onboard MOSFET switch to handle the switch over from DC jack to PD mode. One solution I have thought of here, is to put a separate DC jack supply into the circuit, and use an ideal-diode block to allow for both to either power everything in parallel, or use one or the other. This removes all the in-rush issues from the puzzle completely. It still doesn't totally solve the problem doing all this though, because it still negotiates the POE PD mode if it is present. The only true way to disable it, is to use the Page 13 circuit to shut off the DEN completely if you detect the POE power sucks. I am looking for better suggestions than stuff like this I can easily come up with on my own..... Is there any other references other than this I could be looking at?

Thanks,

-Ted