
Application details
Special care has to be taken here in the selection of M3 and M4 mosfets rdson and Qg, because of the current
shape. Being the current sinusoidal, a certain delay is expected from the time the M3 (M4) body diode will start
conducting to the time the M3 (M4) channel will be closed. This depends on the di/dt and on the forward recovery
time of the body diodes. In any case, it is necessary that diode forward voltage rises (in absolute value) above
Vth2 before the gate drive is activated.
Because the forward recovery time of the state of the art mosfets is very short (few tens nsec) and forward
recovery voltage may be several volts, this delay may be neglected, except in case of MHz operation.
So, in absence of any filter on pin Vd, the only significant delay at turn-on is due to the mosfet Qg, charged by the
peak current of the IC gate driver output (3A typ).
On the other side, as deeply discussed in AN1205, some filter is needed on pin Vd, which delays the fall time at
the IC Vd input; for that reason, AN1205 suggests the filter capacitor is quickly discharged trough a diode, whose
anode is connected to pin Vd and whose cathode is connected to the mosfet drain terminal, as shown in fig.12:
Fig.12: Vds filtering of noise generated by layout parasitics
At turn-off, because Vth1 is very low (few mV), very little delay is expected.
As said before, the turn-on delay introduced by the filter on Vd-Vs is deeply discussed in AN1205 and only few
guidelines how to calculate the filter and the relevant delay are given here.
Moreover, an example of estimation of the delay introduced by M3 (M4) rdson and Qg is given.
To such delays, the delay introduced by the Vd-Vs filter has, of course, to be added.
Vds filtering
In resonant and high power applications the synch. rect. switch signal Vds may be not so clean, because of
secondary stray inductances (see an example in figure 13 where dark blue is the voltage across the secondary
switch, called S1; green is the total secondary current, flowing into the two alternate branches of the rectifier, S1
and S2; purple is the primary current and light blue the gate signal of the AUIRS1170S which drives S1).
Except when working at exactly the resonant frequency, there will always be a phase shift between the half (or
full) bridge center tap voltage and the current. This is clearly visible in figure 13, where some extra Vds ringing
appears in the middle of the Vds pulse, caused by the commutation of the primary switches.
This extra ringing on Vds, if not properly filtered, may induce false or premature commutations of the
AUIRS1170S; to be more specific it may happen that at the primary switching transition the induced noise from
primary to secondary forces the turn-off of the active synch. rect Fet. This effect is more evident at low output
current when the signal across the synch. rect Fes is low.