Maker Pro
Maker Pro

Diode across gate resistor of MOSFET

I've heard many people suggest to put a diode across gate resistor to
make the MOSFET turn off faster, and I know the purpose of gate
resistor is to slow down turn on and damp ringing due to tank circuit
formed by miller capacitance and stray inductance.

So my question is, does the ringing problem only affects the turn on
moment and not during turn off? If we don't limit the speed of
discharge of gate capacitance during turn off, wouldn't that also
causes ringing/EMI since the current is falling too quickly?

And also, does it make any difference of using a schottky diode or
typical kind such as 1N4148?

Thanks for any advice.
 
D

D from BC

Jan 1, 1970
0
I've heard many people suggest to put a diode across gate resistor to
make the MOSFET turn off faster, and I know the purpose of gate
resistor is to slow down turn on and damp ringing due to tank circuit
formed by miller capacitance and stray inductance.

So my question is, does the ringing problem only affects the turn on
moment and not during turn off? If we don't limit the speed of
discharge of gate capacitance during turn off, wouldn't that also
causes ringing/EMI since the current is falling too quickly?

And also, does it make any difference of using a schottky diode or
typical kind such as 1N4148?

Thanks for any advice.

Diode??? On gate??..Then again I'm still learning smps.
I think a bipolar transistor can discharge the gate the fastest.
Look at the internal design of mosfet driver IC's.
D from BC
 
I've heard many people suggest to put a diode across gate resistor to
make the MOSFET turn off faster, and I know the purpose of gate
resistor is to slow down turn on and damp ringing due to tank circuit
formed by miller capacitance and stray inductance.

So my question is, does the ringing problem only affects the turn on
moment and not during turn off? If we don't limit the speed of
discharge of gate capacitance during turn off, wouldn't that also
causes ringing/EMI since the current is falling too quickly?

And also, does it make any difference of using a schottky diode or
typical kind such as 1N4148?

Thanks for any advice.

Comments: It depends
1) Sometimes you want to cut back on crossover shoot-through.
2) Sometimes you want to cut back on the switching losses.
3) Your right, the edges will generate more EMI, but I have found that
good cabling will keep the significant radiation below 25 MHz and out
of sight of the FCC.
4) The reason why turn-off is safer than turn-on is that the gain is
lower when the device is turned off vs. when it is turned on.
5) When I use a diode ( I don't need it in my present design) I use a
Schottky in order not to worry about speed or reverse recovery.
Probably overkill.
6) Since your concerned about stray inductance; remember you have to
dump the inductive stored energy somewhere and sometime. Know where
it's going; at least qualitatively, and quantitatively is you aren't
really (5:1) safe.

Ray
 
L

legg

Jan 1, 1970
0
I've heard many people suggest to put a diode across gate resistor to
make the MOSFET turn off faster, and I know the purpose of gate
resistor is to slow down turn on and damp ringing due to tank circuit
formed by miller capacitance and stray inductance.

The 'damped ringing' on a mosfet drain is much more likely to be
associated with turn-off than turn-on.

At mosfet turn on, 'damped ringing' may occur on other coupled nodes
where other semiconductors are being forced to turn off. The active
mosfet cannot damp these nodes, it can only crudely limit the initial
rate of reverse current and its peak. The total removed charge is
unaffected - damping has to be performed locally, where the
disturbance occurs.
So my question is, does the ringing problem only affects the turn on
moment and not during turn off? If we don't limit the speed of
discharge of gate capacitance during turn off, wouldn't that also
causes ringing/EMI since the current is falling too quickly?

And also, does it make any difference of using a schottky diode or
typical kind such as 1N4148?

Besides other posted advice, you've also got to remember that the
mosfet turn on threshold is fixed and is not likely to be half of the
drive peak voltage - this makes it unlikely that turn on and turn off
speeds are equal - if that was your original intention.

The high forward turn-on overvoltage and higher conductive impedance
of regular diodes will reduce their effectiveness in more demanding
situations. Remember that you are asking the part to perform at high
pulse current for short intervals.

Externally induced dVDS/dt can force a mosfet to turn on, if the gate
resistor or drive impedance is too large.

RL
 
Comments: It depends
1) Sometimes you want to cut back on crossover shoot-through.
2) Sometimes you want to cut back on the switching losses.
3) Your right, the edges will generate more EMI, but I have found that
good cabling will keep the significant radiation below 25 MHz and out
of sight of the FCC.
4) The reason why turn-off is safer than turn-on is that the gain is
lower when the device is turned off vs. when it is turned on.
5) When I use a diode ( I don't need it in my present design) I use a
Schottky in order not to worry about speed or reverse recovery.
Probably overkill.
6) Since your concerned about stray inductance; remember you have to
dump the inductive stored energy somewhere and sometime. Know where
it's going; at least qualitatively, and quantitatively is you aren't
really (5:1) safe.

Ray

Hi Ray, thanks for answering. when you mentioned the gain is lower
when device is off, is it because the gain is proportional to (Vgs -
Vth)^2, and when Vgs is small the d(ID)/d(Vgs) is also smaller?
 
Hi Ray, thanks for answering. when you ntioned the gain is lower
when device is off, is it because the gain is proportional to (Vgs -
Vth)^2, and when Vgs is small the d(ID)/d(Vgs) is also smaller?

For turnon/turnoff ringing you should refer to RL's comments.
I spent about 10 minutes trying to find an excuse for my mistake.
Didn't find a good one. The typicall effect is in fact the turnoff
due to dumping of the L energy into various capacitances including
Cdg , which can turn the fet back on if the gate impedance to ground
isn't low enough. If you end up with turnoff ringing you should
probably prevent overshoot with a catch diode; the diode to the PS
(or in some cases a zener to ground) has to be rated for the full
current. Typically very little power, but high peak current. I have
seen several instances where the catch diode is rated for power but
gets blown away because the current rating is exceeded. In any case
you don't want the transient dv/dt turning the FET back on because a
lot of power can be dumped into the Fet during the turnon event.
Checking the drain voltage in the application is a really good idea;
better to be surprised in the lab than in the field. In my last
application I put a series RC network between the drain and gate to
control the rise and fall times. This was possible because I had
plenty of power margin in the design. You usually don't have this
leeway available.

Ray
 
Hi Ray, thanks for answering. when you mentioned the gain is lower
when device is off, is it because the gain is proportional to (Vgs -
Vth)^2, and when Vgs is small the d(ID)/d(Vgs) is also smaller?

I made a mistake with respect to the turn-off/turn-on; RL is correct.
I spent about 10 minutes trying to think of an excuse for making the
error but couldn't find a decent one.
Don't forget to check the drain output in the real application to make
sure you don't need catch diodes. If you do make sure they are rated
for the full current; not just the wattage.

Ray
 
Top