Building a networked device and want a dual power supply option. So, I’m working on a redundant power design based on Silvertel’s app note, using the AG9205 PoE module (+5v_POE on schematic below) and a Murata OKL-T/3-W12P-C 12V to 5V DC/DC converter. PoE should have priority: if both power sources are present, PoE should remain active and inhibit the 12V path via Q1.

The circuit isn't switching over correctly. Should I add a second diode at the output of the Murata module?

Currently, I have one Schottky diode (SS34-E3/57T) at the output of the AG9205 (PoE) per the app note, but no diode on the output of the Murata module. I’ve noticed some strange behavior when both supplies are present, and I’m wondering if "backfeeding" from the Murata into the PoE path might be the cause, even with the AG9205 output diode in place.

Would adding a second SS34 (after Murata VOUT, before merging into the system +5V rail) solve this? Or is something else likely at play?

Any advice appreciated!

Thanks.

 

Yes, I would add a isolation diode on the output of the Murata module for a proper diode OR circuit and adjust the Murata module trim resistor to compensate for the diode voltage drop at the expected load power.
https://www.vishay.com/docs/88751/ss32.pdf

 

Thanks. Something I notice is that the collector of Q1 (pin 3) is always at 12v when this supply is attached. 5v should be switching this off when connected, but its not.

 

Thanks. Something I notice is that the collector of Q1 (pin 3) is always at 12v when this supply is attached. 5v should be switching this off when connected, but its not.

Can't work. When 5v_POE is on, it raises Q1 base to 5v, but the emitter is at +12 so base << emitter => Q1, being PNP, is on. To turn Q1 off you need to bring the base up to less than -0.6v wrt emitter. The selected BJT at Q1 is a poor choice for this, a MOSFET would be better.

 

Got it. I suppose I got too hung up on following that Silvertel app note.

What do you think of my sketch below?

 

No, that won't work either. When POE is +5 the BAT diode is reverse biassed as the other end at the gate is +12, so MOSFET is off, and with POE at zero its still reverse biassed so MOSFET is still off.

You need something like:

 

Really helpful, thank you!

 

In the end it appears the Murata has too long of a turn on time to be useful in this configuration. I'll probably get rid of the gating circuit all-together and just keep the Murata module always powered up if 12v is applied and then simply OR the output of Murata and PoE 5v together at the output. Any issue you can think of running this way?