I am currently in the fourth year of my bachelor's degree and my knowledge on this is really really really very limited, but i got a simulation circuit of an isolated dc-dc converter from my senior and im supposed to identify the losses of the circuit, mainly at the inverter, the transformer and the output rectifier.

From what i found out, synchronous rectification can reduce the losses at the output but switching the switch on at the same time the current flows through the diode, hence reducing power losses and improving efficiency. Looking at the secondary side of the transformer, so basically the gate and source of the MOSFET is shorted so that only the body diode is used, to replicate a normal output rectifier. I was told that this was done because synchronous rectification will be applied later on in the process(that's why MOSFETs are used instead of diodes), but because the controller for sync rect is not done yet, and this was done to replicate a normal diode output rectifier.

But I heard that doing this is not good, and shouldn't be used in an actual circuit, but what exactly about shorting the gate and source is not good in an NMOS? (As far as i know, the thing that destroys an nmos is only if you short the drain to source. I might be wrong :x)

Also about the recovery time etc. of the body diode being worse than a normal diode and stuff, too many things that im not too sure about.

enter image description here

EDIT: Additional question, one more question is that if you look at the MOSFET on the primary side it has a diode connected parallel to it(i was told that this is the body diode and not an external diode), but on the secondary side the ideal NMOS doesnt have the body diode shown. Why is that?

  • \$\begingroup\$ I would say that the bad thing here is to use the body diode, which is usually just a parasitic property and not up to any real task that you would use a normal diode for. \$\endgroup\$
    – PlasmaHH
    Oct 25, 2018 at 12:42
  • \$\begingroup\$ In most applications the body diode is fine to use as a normal diode. But in switching regulators it can be problematic due to the high \$t_{rr}\$ (recovery time) of the body diode. \$\endgroup\$ Oct 25, 2018 at 13:07
  • \$\begingroup\$ So basically if synchronous rectification is not applied here, a normal diode should be used instead of this setup, right? And one more question, what exactly does "shorting the gate to source" do to the MOSFET, does it mean that by shorting the gate to source, whenever the secondary side of the transformer outputs a current, Vgs=0, and thus the switch wont activate, and all current flows through the body diode, is that right? Btw if im not mistaken this system runs at like 100kHz or so \$\endgroup\$ Oct 25, 2018 at 13:09
  • \$\begingroup\$ @EzmirIzammel: yes, shorting the gate to the source just turns off the FET, and you end up using the body diode. If this project has a physical board associated with it, I'd just hack real diodes onto it in place of the FETs, and I'd simulate it with the diodes. But you're not in charge of that part, I think. \$\endgroup\$
    – TimWescott
    Oct 25, 2018 at 14:37
  • \$\begingroup\$ Connecting gate to source is just ensuring that the FET remains off. The body diode, though it's a parasitic feature of the FET construction is a decent size, and can generally carry the same current as the Id of the FET, thermal considerations allowing, but that size gives it a large trr, so wouldn't be a good choice for a high frequency rectifier. If the board's thermal layout has been determined for the dissipation of the FETs in active mode, you may find issues with heat when relying on the diodes. If the secondary voltage is low enough, you might find Schottky diodes in the same package. \$\endgroup\$
    – Phil G
    Oct 25, 2018 at 14:51


Browse other questions tagged or ask your own question.