Parallel DC DC switching boards for higher current output

[Daniele]

I'd like to use cheap XL4015 boards to have a maximum constant output of at least 5A. Using a single XL4015 board at 5A is not feasible since it overheats (inductor can go over 200°C!) and XL4015 must be well heat-sinked and also XL4015 efficiency drops a lot at high currents. So the choice is to use 2 x XL4015 boards in parallel. Knowing the related problems of parallel operation, I added a load sharing system using op amps, so that the output currents can be evenly shared by the two supply boards. The op amp outputs act on the feedback of each XL4015 board. Each board has an output diode to avoid negative output current.
First op amp:
--------------
Of course the diode causes a slight voltage drop current-dependant. This is compensated by the first op amp, which operates to minimize the difference between a voltage reference and the voltage output measured after the diodes (where the actual load is connected).
Second op amp:
--------------
The second op amp minimizes the voltage difference bewtween the two XL4015 boards, measuring their voltages before each diode. This should ensure equal load sharing. Differences in diodes can be neglected, since even a slight load unbalance around 2.5A per board will not be a big deal.

Testing enviroment:
-------------------
For testing my idea I added one current sensor after each diode, so I can monitor the current output of each board. Corrent sensor is ACS712 hall effect sensor.

Picture of prototype circuit:
https://ibb.co/fmJJgG
Power ON:
https://ibb.co/ceoJgG
Load step response 0A to 5A:
https://ibb.co/jCVW1G
Output voltage ripple around 5V is below 30mV:
https://ibb.co/b814MG
Schematic (capacitors are missing):
https://ibb.co/iaaW1G



Of course the current output could be easily extended with more XL4015 boards in parallel in a similar way.
Any feedback is appreciated!

[Gribo]

What happens if one of them fails?

[Daniele]

it depends on the type of failure, I see 2 possible cases:
1) fails with switcher open: output voltage = 0, thus the whole load will be picked up by the other DC DC and eventually the fuse PTC will cut the current.
2) fails with switcher closed: output voltage = MAX, thus the failed converter will pick up all the load. This will expose to the risk of overvoltage, thus I foresee an overvoltage protection which will shut down the whole thing by blowing the main fuse via SCR (not shown in schematic).