Good morning,
we are testing a prototype of a new 500W isolated DC / DC converter based on the controller LM5046 with reference to EVAL, but we have a problem in particular operating on minimum duty cycle conditions;
the problem occurs when you reach the minimum duty cycle on the transformer (high voltage input or output with no load in minimum output voltage), in which the controller generates the control pulses erratically creating small cross conduction on the primary full bridge MOSFET: in this condition, the controller interrupts some pulses and the duration of others pulse becomes 3 or 4 times greater than the pulse duration in normal conditions; seems as if the controller performs continuous restart random, but the SS pin and pin REF remain always active. the problem is resolved by applying a minimum load of at least 7/8W (47ohm)
We had thought it was a problem of noise from our layout, so we tried to reproduce the problem with the controller powered only by disconnecting the driver output and disconnecting the primary mosfet, so without "creating noises" the ground plane, to simulate the feedback we have connected a trimmer between the REF and GND, and connected COMP to the central point of trimmer; reached the point of adjustment toward the minimum duty cycle control signals behave abnormally as already described above; with driver and power mosfet connected, this abnormal effect produces breakage of primary MOSFET.
to provide further guidance, the basic pattern of the circuit is taken from EVAL LM5046, but the output rectifier is a diode, and the mosfets are driven by primary LM5112 connected in output of the LM5046 with the same configuration of bootstrap; the voltage range input of our product is 24Vdc / 50Vdc while the output voltage is adjustable between 20Vdc and 30Vdc.
We stopped the delivery of prototypes for these reasons, then waiting for your urgent help.
Thanks
Mirko
