LM5116: My DC/DC does not want to go beyond 60% duty cycle

Hi, mr Vincenzo,

It took too much time for me tampering with this circuit. I did used three chips; all of them failed differently - first one did not raise to higher then half voltage without doing something which look a little like parasitic subharmonic oscillations and all sorts of other; after altering the DEMB system which i did not implemented first time and resoldered the chip, it did no longer work, except very little output, like it was all the time in protection - except it wasn't.

The second chip i soldered did worked, however with the same kind of oscillation; i did some improvements - not exactly to the highest current and voltage, but approaching; and suddenly fails the HI output. Shorted (HO to SW). No reason.

The third chip started from the very first moment with very high current consumption, 200mA; heated uo immediately; Again, no reason, i did disconnected the output, checked them with ohmmeter; no "visible" fault;  the oscilloscope shown normal outputs HO and LO; the bootstrap was ok; nothing except internal fault could be..

I quitted using this version and started a new approach, designing myself a combination, using 555 as PWM oscillator, a IR21094 as half-bridge driver, TC4422 as final drivers and some 393 comparators for voltage feedback, current regulation, (slow) protections; the fast shutdown protection i am just implemented them today (a small paralleled MOS (attacked in gate through a capacitive divider from the main MOS gate signal) read the current in the top-MOS, a voltage amplifier in common-base configuration amplifies that and send it to a repeated and after that to a totem-pole and a small MOS to shut-down the gate of the main Hi-MOS; also the signal is sent to the slow-shutdown entrances at inputs of the 4422's and 21094. And everything works like a charm.

Still have to implement one last thing: a positive current sensing through the LO-MOS, to shut-down this MOS if the output coil is depleted before the end of the return cycle (when starting into biased load or charged capacitors), somehow like the 5116 "diode emulation" function. It should work, in a few days i'm done.