I'm using Schottky diodes and rectifier diodes to protect the Motor, relay or some other Inductive products, I'm not using TVS to protect MSOFET, maybe someday I will do that.

About three weeks ago, one of my friend asked me to help him to debug for a interference of transformerless circuit, I saw that he using TVS to protect the DC circuit for surge voltage.

Using TVS to protect MOSFET.

Ok, finally got it solved, thanks to everyone here! Q2, the low power P-channel MOSFET needed a Transient Voltage Supressor, NTE4926, or similar, to protect it from voltage spikes caused by the pump motor. I connected it between Unswitched Power and Ground.

I also added the recommended capacitor to decouple, but removed it after determining it wasn't needed.

Thanks again,

Marty

 

Where is the control signal for the upper left MOSFET from?

I'm guessing the upside down ground on the bottom circuit?

Maybe use matched arrows (different from the ground symbol) so the paths are more clear.

This looks pretty close to the soft on/off switch from EEVBlog, is that where you got the base circuit?

 

Where is the control signal for the upper left MOSFET from?

I'm guessing the upside down ground on the bottom circuit?

Maybe use matched arrows (different from the ground symbol) so the paths are more clear.

This looks pretty close to the soft on/off switch from EEVBlog, is that where you got the base circuit?

The control signal comes from a microprocessor, 0-5vdc. The MOSFET is a logic-level MOSFET.

Got the basic idea from here.

Works well now that I added a TVS across the battery.

Marty

 

Putting TVS devices on everything to make stuff work is a sign of a bad design.

If you had the right decoupling caps on the power lines to separate the noisy power from the good power and the right attention to current paths of the noisy power and good power you would have been fine.

What you have done is leave things bad and making nasty noise in the wrong places, but try to soak up the noise with a TVS.

Myself and others asked for a photo of your construction before, which would have helped a lot finding out what is wrong with your layout.

 

Thanks THE_RB,

Good point. I suppose I can experiment with decoupling without the TVS and see what the 12vdc line looks like on the scope and see if it still blows up the mosfet. Unable to post a photo of this circuit

Marty

 

When the motor switches off you will get a huge spike on the 12 volt rail.
You need to decouple the 12 volts using a small resistor and a capacitor.

 

When the motor switches off you will get a huge spike on the 12 volt rail.
You need to decouple the 12 volts using a small resistor and a capacitor.

Hi Nigel,

I am worried about adding the small resistor as I will be using the switched 12vdc to power an audio amplifier, and the higher the voltage, the louder the sound I can make, and it's very important for the sound to be LOUD.

Can I just use a cap across the power rail and leave it at that?

Marty

 

Are going to power an audio amplifier from the same supply this circuit which blows MOSFETs is on?

If so, there will be issues with noise or damaging the power supply in the amp unless the design is corrected.

 

it's very important for the sound to be LOUD.

audio amp powered by 12V and drawing 50mA is not going to be very loud.

 

audio amp powered by 12V and drawing 50mA is not going to be very loud.

In one application, that circuit does only draw 50ma, but it draws 2 amps when connected to the audio amp. You're right, I wasn't very clear about that, just trying to show that the current thru that part of the circuit wasn't what was blowing up the mosfet.

Any ideas how large a capacitor to add across the power rail to decouple the spike caused my the motor's inductance kick?

Marty

 

If you are not using PWM on the motor, just on/off switching, you can put a cap from gate-source on the big FET that switches the motor. That will ramp motor current up/down over a fraction of a second, and you won't get a "big spike" from switching the motor on or off. Somewhere from 0.1uF to a couple of uF.

Also a decent cap across the motor terminals will help, maybe 0.47uF 60v greencap poly.

The the motor current path (as a circular path) should be wired directly from battery + to - terminals. Then the current path for the other FET and audio amps should also directly be from battery + to - terminals again as a separate circular path. The big filter cap for the audio FET and amp should be local to it, on it's board, far away from the other FET in the other circuit and other wiring that switches the motor power.

 

If you are not using PWM on the motor, just on/off switching, you can put a cap from gate-source on the big FET that switches the motor. That will ramp motor current up/down over a fraction of a second, and you won't get a "big spike" from switching the motor on or off. Somewhere from 0.1uF to a couple of uF.

Also a decent cap across the motor terminals will help, maybe 0.47uF 60v greencap poly.

The the motor current path (as a circular path) should be wired directly from battery + to - terminals. Then the current path for the other FET and audio amps should also directly be from battery + to - terminals again as a separate circular path. The big filter cap for the audio FET and amp should be local to it, on it's board, far away from the other FET in the other circuit and other wiring that switches the motor power.

Hello The_RB,

All great ideas. The motor is not being PWM'd, just ON/OFF, sometimes once per second. I hadn't thought about reducing the inductance kick by reducing di/dt. Only downside might be extra heat to Q1 during the partially open times. Would be worth trying to see if it helps. The gate drive is from a uC.

Although it all happens on one sircuit board, each function has its own space on the board, so I should be able to locate the decoupling caps near their desired functions.

I am trying to use minimum parts count here, so I do have to balance idealist with cost (time, space on the board, cost of parts, etc.).

Maybe one cap across the G-S of Q1 and one large one near the unswitched power part of Q2/Q3.

Thanks for your excellent ideas.

Marty