[SOLVED] MOSFET ringing due to noise-filter capacitor.

[mrinalmani]

I simulated a simple low-side gate driver made of BJT totem-pole. The simulation was clean.
Now I added a 10nH inductor in series with the power supply of the driver, to get an exaggerated view of the parasitic track inductance. The simulation was still clean.
Now, I added a 100nF noise filtering capacitor across the totem-pole. The gate began to ring. Soon I found that the gate was ringing because Vcc itself was oscillating! Now I added a small resistor (100 mohm) in series with the capacitor. Transients died extremely fast and the results were clean again. The reason appears to be the "initial charging" of the filter capacitor through the inductor during power-ON.
Simulations with and without the series resistor are attached below.

Now, my question is...
Should it be a good practice to place a small series resistor with all noise filter capacitors, to avoid transients during power ON? (Considering that ceramic capacitors may have an ESR as low as 10 mOhm)
Or is this situation unlikely to occur in the real world?
Thanks...

 

[KlausST]

Hi,

i bet ringing is a 3.56 MHz.


The L with the two paralleled Cs give a wonderful resonant.

F = 1 / ( 2 * PI * sqr(L * C))

adding a series resistor dampens the resonant, but also decreases the filtering and step response.

Instead of the low ohmic series resistor connect an RC in parallel to the Cs. C = 100n, R= 100

Additionally put a 10u tanatalum in parallel to the supply Cs for better response on higher load.

Or is this situation unlikely to occur in the real world?

Indeed it happens too often in reality.
This is one point why electronic equipment react on electromagnetic fields.
A inductivity (wanted or unwanted) combinied with a capacitor is problematic when no load ias connected.
A load - and this might be a LED for example - helps to reduce the resonant.

With the RC we built a load that consumes no power during normal operation...


Hope this helps
Klaus

 

[mrinalmani]

Thanks for the reply...
But I dont quite get the reason of adding an RC in parallel with the Cs. Though I tried, it didn't help. (see attachment)
I intentionally did not add a 10u capacitor, because I am assuming that the 10nH parasite lies "after" the tank capacitor and before the driver power pin.

 

[KlausST]

Hi,

yes, in your scope you see the ringing when power is switched ON.

But the same happens when load is switched, but with lower ringing voltage.

For faster decay you need a lower resistor.

It seems like it tends to your solution, but a pulse caused by the load is best handled with lowest ESR.


Klaus

 

[FvM]

But I dont quite get the reason of adding an RC in parallel with the Cs. Though I tried, it didn't help.

Yes, because the snubber becomes useless due to too large resistor value. You should try a value in the range of LC characteristic impedance √L/C, e.g. 0.3 ohm. You should play around with R and C values to get an idea.

 

[mrinalmani]

Thanks everyone for the reply.
I'll be using a 250mOhm resistor in series with all ceramic type filter capacitors.