Add Zener and C to MOSFET

[ervinl]

Hi,

I would like to add zener and small pf C in front of power amp/final MOSFET to stabilize better (original schematic does not have Zd, C, stopper, and I experience somehow the MOSFET is hot and gone by accident).

If I also put stopper resistor in front of gate, and limitter resistor at source, where should I connect this Zd and C, before stopper, or after stopper in the gate pin?

And what about the other pin, direct to source, or after limitter resistor?

What value of C will prevent high freq oscilation better, but still larga audio bandwidth, will 47pf enough (less or more)?

Thanks,

Ervin L

[CBS240]

Hi

Place two Zener diodes between gate and source pins for Vgs protection; in series, opposing polarities so the gate source barrier is protected in both directions. Use a C of ~30-40pf and make a Zobel filter from gate to GND as close to the device package as possible with the zero set at ~30-40MHz. Mosfets can oscillate very fast if not dampened properly. The optimum value of the gate stopper resistor will depend on the specific mosfet and may not be the same for both P-ch and N-ch of relative equal Gm. The layout of the GND path back to the star may affect the effectiveness of the gate LP filter.

[CBS240]

....also try to keep the leads as short as possible, no wires trailing from the PCB, and place your PS decoupling cap as close to the drain pin as you can. You do use de-coupling, yes? This is where 2 sided PCB and SMDs come in handy, it allows you to better optimize layout. This is important in getting the most of the very high effective Ft from the mosfet and not have to locally compensate all it's BW away.

[nigelwright7557]

I dont think you need a zobel network on the input to a MOSFET.
The MOSFET already has considerable gate capacitance or certainly the ones I use.

I usually use a 390R gate stopper resistor.


If your getting oscilation then look at the VAS miller capacitor again it could be too small.

If its a quasi amp then the lower driver needs a miller capacitor.

[CBS240]

For a source follower the Zobel has little to do with CISS as it connects to GND, since CISS is Cgs+Cgd. It allows you to use a smaller gate stopper and form a steeper filter response. For a mosfet as a source follower with ~1500pf CISS, I use a 75R gate stopper and 47pf in series with 82R from gate to GND, zero ~40MHZ. Allowing the use of the higher BW of the mosfet and still having stability will help to reduce crossover distortion as well. Proper layout is crucial.

Here is a good reference paper pertaining to this subject.

[ervinl]

Originally Posted by CBS240

For a source follower the Zobel has little to do with CISS as it connects to GND, since CISS is Cgs+Cgd. It allows you to use a smaller gate stopper and form a steeper filter response. For a mosfet as a source follower with ~1500pf CISS, I use a 75R gate stopper and 47pf in series with 82R from gate to GND, zero ~40MHZ. Allowing the use of the higher BW of the mosfet and still having stability will help to reduce crossover distortion as well. Proper layout is crucial.

Here is a good reference paper pertaining to this subject.

Why must use zobel (75-82 ohm resistor in series), instead of 47pf directly paralel from G to S (or ground?) Currently the circuit I try is power follower (source = output to speaker).

Unfortunately, there is a cable jumper in the PCB to connect from one MOSFET to another, as it is separated in order to maximize the HS (separate in such a distance). I also suspect this cable may pick up alien signal if the MOSFET is not secured from anti oscilation.

Thanks,

Ervin L

[CBS240]

47pf from G to S will do nothing. If it is a source follower, it will connect from gate to GND. It is advantageous to include a zero (as in poles and zeros) around the frequency where the mosfet may oscillate. Read section 1.6 on page 7 of the artical I referenced in post 5 relating to the types of oscillators created by mosfet internal parisitic components. Also the last paragraph just before section 3 on page 12. Relying on just a gate stopper and increasing the value until the 'oscillators' are dampened is a cheap fix and reduces the gain and increases the phase shift at higher frequencies and is far from optimal. Better to go for a sharper 'knee' and slope in the Bode plot and less phase shift.

Unfortunately, there is a cable jumper in the PCB to connect from one MOSFET to another, as it is separated in order to maximize the HS (separate in such a distance).

If it is possible, a de-coupling cap, ~1000uf || 100nf film cap, at the drain pin to GND (with a seperate path to the power 'star' GND junction) of each transistor may help maintain stability with the cable jumper connection.

[Workhorse]

I prefer a ferrite ring[bead] in the gate leg of mosfets to suppress oscillation in MHZ region

[ervinl]

Originally Posted by CBS240

Hi

Place two Zener diodes between gate and source pins for Vgs protection; in series, opposing polarities so the gate source barrier is protected in both directions. Use a C of ~30-40pf and make a Zobel filter from gate to GND as close to the device package as possible with the zero set at ~30-40MHz. Mosfets can oscillate very fast if not dampened properly. The optimum value of the gate stopper resistor will depend on the specific mosfet and may not be the same for both P-ch and N-ch of relative equal Gm. The layout of the GND path back to the star may affect the effectiveness of the gate LP filter.

About the Zd, what if I only user 1, only protect Vg-s to always < max voltage, and the minimum is >= 0.7 (Zd reverse voltage ~ dioda reverse voltage). I think it will also protect the MOSFET.

Will it hurt the sound just by connecting like that, as I know Vg will never < Vs normally (CMIIW), and if it is, reverse dioda will keep itu < 0.7 voltage difference.

Thx,

Ervin L