@crutschow, I have learned a lot from your comments on other Zener snubber posts and I kindly request your input regarding the selection of suitable diodes in the application below.


I thought I had finally found a good Zener and diode that could withstand the high current of this inductor kickback, but I soldered these SMD components onto a protoboard and the solenoid fired nicely 5-6 times and then I fried the MOSFET (magic smoke). Modeling in LTspice, I think maybe the voltage on the drain spiked high enough to fail the MOSFET. The voltage on the MOSFET drain is exceeding 80V and the Zener is only conducting 90mA of current. Is this because the kickback is fast and the Zener can't break down fast enough? I thought the voltage would clamp at 30V (Zener breakdown voltage for SZ1SMB5936BT3G) + 19V (Vcc) or approximately 50V. The MOSFET is rated 70V. Here is blue trace current through the coil, green trace voltage at the mosfet drain, and red trace current through zener.



Theoretically, I would want to see this (modeled with default components):


Could a TVS diode replace the Zener in this application and begin dissipating the kickback faster than the Zener as the voltage spikes up? Or do you think the Zener I picked breaks down fast enough and the MOSFET failed for other reasons like bad heat dissipation?

The inductor is solenoid B11HD-255-B-3. It is rated 19V, 90W, 4.75mH, 4.3 Ohms, 10% duty cycle

Are these calculations correct?:
Inductor energy storage E = 1/2 * L * I² is approx. 0.048 J. I would like to turn off the solenoid in 0.5 ms. 0.048J/0.0005s = 96W. So, I want to dissipate 96W over 0.5ms.

 

Please attach your SPICE file.

 

@crutschow, I have learned a lot from your comments on other Zener snubber posts and I kindly request your input regarding the selection of suitable diodes in the application below.
View attachment 262358

I thought I had finally found a good Zener and diode that could withstand the high current of this inductor kickback, but I soldered these SMD components onto a protoboard and the solenoid fired nicely 5-6 times and then I fried the MOSFET (magic smoke). Modeling in LTspice, I think maybe the voltage on the drain spiked high enough to fail the MOSFET. The voltage on the MOSFET drain is exceeding 80V and the Zener is only conducting 90mA of current. Is this because the kickback is fast and the Zener can't break down fast enough? I thought the voltage would clamp at 30V (Zener breakdown voltage for SZ1SMB5936BT3G) + 19V (Vcc) or approximately 50V. The MOSFET is rated 70V. Here is blue trace current through the coil, green trace voltage at the mosfet drain, and red trace current through zener.

View attachment 262356

Theoretically, I would want to see this (modeled with default components):
View attachment 262357

Could a TVS diode replace the Zener in this application and begin dissipating the kickback faster than the Zener as the voltage spikes up? Or do you think the Zener I picked breaks down fast enough and the MOSFET failed for other reasons like bad heat dissipation?

The inductor is solenoid B11HD-255-B-3. It is rated 19V, 90W, 4.75mH, 4.3 Ohms, 10% duty cycle

Are these calculations correct?:
Inductor energy storage E = 1/2 * L * I² is approx. 0.048 J. I would like to turn off the solenoid in 0.5 ms. 0.048J/0.0005s = 96W. So, I want to dissipate 96W over 0.5ms.

Your inductor is ringing at a frequency determined by a parasitic capacitance on the MOSFET or a part of the Zener diode. If you look at the frequency you can determine the value of that capacitance.

The first thing I would do is refine your circuit to get rid of the zener.
The next thing I would try is charging a capacitor connected between the +19V source and some limited higher voltage. It will take me a minute or two to come up with an example. I can eventually adapt it to your parts if you include everything zipped up in a single file.

 

Try this. I know that the MOSFET and the Schottky diode are beasts, but I have the models and the simulation works repetitively. If you want the turnoff to be faster, you're going to need an active driver.

 

As requested, post you LTspice files.

Noticed a couple of concerns:

Why are you using a small 1N4148 diode to carry the 4.5A inductor current during the turn-off transient.
Its Absolute Maximum surge current rating is only 2A.

The Zener diode you are using has an impedance at its test current of 26Ω.
That could be causing the voltage peak you are seeing.

The Zener turn-on time is not a factor here.

What is the pulse frequency?

 

If you want the turnoff to be faster, you're going to need an active driver.

That's what the Zener does, which you left out of your circuit for some reason(?).

 

My simulation with a 33V Zener I had a model for:

R2 damps the observed oscillation.

The turn-off time is slightly less than 0.5ms, as the TS desired.

A Schottky diode is not need for D2 since the turn-on time of standard diodes is quite fast (which is what is needed here).
The diode turn-off is slow but that has no effect on the inductive spike suppression.

 

Please attach your SPICE file.

Sorry about that. Please see attached. Answering comments in order. Thank you for your input.

 

Your inductor is ringing at a frequency determined by a parasitic capacitance on the MOSFET or a part of the Zener diode. If you look at the frequency you can determine the value of that capacitance.

The first thing I would do is refine your circuit to get rid of the Zener.
The next thing I would try is charging a capacitor connected between the +19V source and some limited higher voltage. It will take me a minute or two to come up with an example. I can eventually adapt it to your parts if you include everything zipped up in a single file.

Willing to try anything for learning purposes. I attached the .zip. Also, I was using this diode for testing on the perfboard S5B-E3/57T. It's rated 5 amps. I did not have a spice for it so I dropped the 1n4148 without due diligence. Thanks for your input.

 

Try this. I know that the MOSFET and the Schottky diode are beasts, but I have the models and the simulation works repetitively. If you want the turnoff to be faster, you're going to need an active driver.


View attachment 262361

Thank you I will examine this closely after work and explore the simulation.

 

As requested, post you LTspice files.

Noticed a couple of concerns:

Why are you using a small 1N4148 diode to carry the 4.5A inductor current during the turn-off transient.
Its Absolute Maximum surge current rating is only 2A.

The Zener diode you are using has an impedance at its test current of 26Ω.
That could be causing the voltage peak you are seeing.

The Zener turn-on time is not a factor here.

What is the pulse frequency?

Thank you for your input. I was using this diode for testing on the perfboard: S5B-E3/57T. It's rated 5 amps. I did not have a SPICE model for it so I dropped the 1n4148 without due diligence. I will look more closely at the 26 Ohm impedance at the test current. Pulse frequency would be 10 Hz typical and I would like to design for 20 Hz as a reach goal.

 

My simulation with a 33V Zener I had a model for:

R2 damps the observed oscillation.

The turn-off time is slightly less than 0.5ms, as the TS desired.

A Schottky diode is not need for D2 since the turn-on time of standard diodes is quite fast (which is what is needed here).
The diode turn-off is slow but that has no effect on the inductive spike suppression.

View attachment 262366

@crutschow This looks promising. Thank you and I will simulate it tonight after work. I would like to try to get it working with the SiS176LDN because my micro can switch it with 3.3V.

 

There is a problem with your Zener model:
Below is the simulation of the 1smb5936bt3 diode (yellow trace) and a DFLZ33 33V 1W Zener (green trace).
As you can see, the 1smb5936bt3 diode has a sharp rise in its voltage above 55mA current, which is not normal.
In contrast, the DFLZ33 voltage stays quite constant, which is normal.

You need a correct Zener model.

As far as your failure with the real circuit, do you have an oscilloscope to measure the voltage at the MOSFET source terminal?
If so, start with a lower supply voltage (say 10V) and go up from there.

 

There is a problem with your Zener model:
Below is the simulation of the 1smb5936bt3 diode (yellow trace) and a DFLZ33 33V 1W Zener (green trace).
As you can see, the 1smb5936bt3 diode has a sharp rise in its voltage above 55mA current, which is not normal.
In contrast, the DFLZ33 voltage stays quite constant, which is normal.

You need a correct Zener model.

As far as your failure with the real circuit, do you have an oscilloscope to measure the voltage at the MOSFET source terminal?
If so, start with a lower supply voltage (say 10V) and go up from there.

View attachment 262392

Ok something up with the model. That does explain a lot and makes me feel less incompetent.
I did recently buy a siglent sds1104x-e so I will scope and report back.
What am I looking for on the nmos source ? I would have thought to probe the drain?

Also the resistor damper you proposed - is that frequently used to suppress Zener snubber ringing? It looks like it works great, I'm just hoping I can find some literature to read more about the theory there to better understand it.

 

What am I looking for on the nmos source ? I would have thought to probe the drain?

Yes, I misspoke. I meant drain.

Also the resistor damper you proposed - is that frequently used to suppress Zener snubber ringing?

The resistor lowers the Q of the inductor-stray capacitance resonant (tank) circuit to suppress the ringing (by dissipating the inductive energy).
You will likely have to experiment to get the correct value in the actual circuit, due to the unknown value of stray capacitance.

But the ringing likely doesn't cause a problem in your circuit, so you probably don't need the resistor.

 

Yes, I misspoke. I meant drain.
The resistor lowers the Q of the inductor-stray capacitance resonant (tank) circuit to suppress the ringing (by dissipating the inductive energy).
You will likely have to experiment to get the correct value in the actual circuit, due to the unknown value of stray capacitance.

But the ringing likely doesn't cause a problem in your circuit, so you probably don't need the resistor.

Ok thanks! I will report back on the physical results.

 

Yes, I misspoke. I meant drain.
The resistor lowers the Q of the inductor-stray capacitance resonant (tank) circuit to suppress the ringing (by dissipating the inductive energy).
You will likely have to experiment to get the correct value in the actual circuit, due to the unknown value of stray capacitance.

But the ringing likely doesn't cause a problem in your circuit, so you probably don't need the resistor.

All I had on hand was a 43V Zener (1SMB5940B-13) but I replaced the nmos (SiS176LDN) and scoped the drain while everything was working.

The Zener seems to be clamping around 62V which is good but it is taking more time than expected for the voltage to start dropping and then drop. Maybe that's because of the Zener impedance?

I am trying to find a Zener with a lower impedance that can still withstand the short but high wattage.

 

Maybe that's because of the Zener impedance?

I am trying to find a Zener with a lower impedance that can still withstand the short but high wattage.

The Zener impedance mostly affects the peak flyback voltage.
It should have little effect on the voltage dropping back down.

Is the circuit essentially the same as the one you posted in your first post or did you add a resistor to dampen the ringing?

 

The Zener impedance mostly affects the peak flyback voltage.
It should have little effect on the voltage dropping back down.

Is the circuit essentially the same as the one you posted in your first post or did you add a resistor to dampen the ringing?

Yes, the physical test circuit is consistent with my original LTspice circuit, but with a 47V Zener ( 1SMB5940B-13) and diode ( S5B-E3/57T). The circuit scoped was the same as the one that fried during my original testing. I'm not sure what is different this time, maybe my soldering or maybe that I have not turned the solenoid on for extended periods (5+ seconds) this time. I have not tried the resistor damper yet and I do not see any ringing presently. I am not sure why the first one fried but I have not tested the new one harshly. I wanted to scope it before any mosfet failures. I will solder on a resistor in the morning and scope it again.

How would one reduce this period? And what is going on in this region?
edit @10:33est: clarification - why does the voltage reduce gradually and then plummet to Vcc? Does that falling edge denote when the inductor stored energy is completely dissipated?

 

How would one reduce this period? And what is going on in this region?

That is the period where inductor current is going through the diodes to dissipate the inductive energy.
If you want that time to be shorter, than you need to used a higher Zener voltage to dissipate the energy faster (which, of course, will require a MOSFET with a higher voltage rating).

For example with a 51V Zener, the turn-off time is reduced to about 0.35ms (below).
Notice the diode current (red trace) during the turn-off.