Hello there!
I am designing with the MP6004 for a PoE+ application. I use the Pulse 3855.002NLT transformer, which has two outputs, with turn ratios PRI:SEC:SEC:AUX = 5.33:1:1:2.
I want to stack the secondar windings to get 12V and 5.6V outputs (both with a 0.7V drop schottky diode). How will the feedback react? The load is variable on both outputs, so there could be a high load on 5.6V and no/light load on 12V, and vice-versa. Ideally, I’d like the 5.6V to be the master, if that’s possible.
I have no idea how equation 6 of the MP6004’s datasheet would change with two stacked outputs, as the formula specifically references the output voltage, not the current.
How would cross-regulation affect the outputs?
EDIT: I have found an interesting explanation here: [SOLVED] - Auxiliary Flyback Supply | Forum for Electronics. According to this, we already are in the case where the regulated output is the auxiliary, and the other outputs are the secondary windings. So not much would change regarding the stability of the outputs. Is that correct?
This part really regulates the voltage at the feedback winding. So the regulation you achieve on the secondary windings depends on how well coupled they are to the feedback winding. It should be possible to make it work. I would be prepared to have some preload on each output. At low load the leakage inductance kick can peak charge the output cap to a voltage that is higher than the normal ( turns ratio determined ) flyback voltage. Plan for 5% load on each output and back it down from there.
I’m happy to read this as it is similar to my own reasoning. I’ll be using caps with 1.6x the output voltage and a snubber on the output to minimize transients (TI’s Power Stage Designer Tool is quite useful for this, I don’t know if monolithic has something similar). I’ll also move the PoE’s Maintain Power Signature circuit (yay for parts shortage!) to the 12V rail from the 5.6V rail, with the logic on the 5.6V rail (with a 3.3V LDO). Hopefully it will be enough to keep things stable.
Right now everything is on the 5V rail with no 12V rail, and there are no stability issues with 50 mA output current. I followed the datasheet and completed the missing bits with TI’s tool. Output voltage is within 1% of expected value, and also reliably delivering 3A of current for hours, although the output diode and the MP6004 do get quite hot to the touch. The only issues are with transient spikes (as with any DCM flyback). Hopefully, following TI’s SLUA469 and improving my copper planes will help.
The part can run at 100C so hot to the touch isn’t too hot as far as the part is concerned. The diode can be replaced with a bigger one, or you might investigate the fast ideal diode controllers that replace the diode with a synchronous FET. I have used them with success. That bumps the efficiency of the circuit by almost 10% or on a limited PoE budget gives you extra power to the output. The output voltage will be different because the 500mV diode gets replaced by a 50mV “diode” I would guess 5 is what you are trying to keep about right and 12 can be a little bit higher or lower with out disaster striking.
Thanks for your helpful insight. I looked into the synchronous rectification as well, and the performance boost would be nice, but it also has significant disadvantages in my case:
Higher EMI;
Longer BOM (by one part, but still.) and in the end same cost because I need “high” voltage/current NMOS;
Only ~4% higher efficiency on the output (12.2V instead of 12.7V at the transformer’s terminals), and it becomes less than that overall.