I did the calculation for you just so you could see how close your 100 ohm guess was. The parasitic capacitance of the FET in this circuit is 3nF. To damp to Q=1, your snubber would be R=183 ohm, C=10nF. To critically damp, the snubber would be R=91 ohm, C excessively large, say 10uF (for true critical damping, it would have to be infinite). So your 100 ohm and a still excessively large 100nF is close to critical damping. Rather than deleting all your traces every time, just use F4 to label the important ones, then they won't change.
THX for calculating the optimum value. Critical damping at Zeta=1 translates to Q=0.5 en.wikipedia.org/wiki/Damping_ratio Do you have a suggestion for a website where the calculation of the values for optimized components of a snubber is easily explained? I found a few but they are all not very good for a beginner.
I did have a really good PDF explaining RC snubbers, RCD soft voltage clamps, hard voltage clamps and RCD dV/dt limiters, but I can't find it. RC snubbers are easy enough to calculate, the hardest thing is knowing which parasitics are resonating. It's almost always an obvious inductance such as a transformer's magnetising inductance or leakage inductance or just an inductor, ringing with the parasitic capacitance of whichever semiconductor is currently turned off. You can measure inductance, look at the frequency of resonance and calculate the parasitic capacitance from the formulae for inductive reactance and capacitive reactance. The resistance required for Q=1 will be equal to either reactance at the resonant frequency, which is also equal to SQRT(L/C). Q is proportional to R, so to get Q=0.5, multiply that R by 0.5. In the real world, you usually design snubbers for Q=1 to 0.7. Snubber capacitor will be 3-5x the parasitic capacitance. If you don't know either the inductance or capacitance, you can add capacitance in parallel to the parasitic capacitance until the resonant frequency is halved, then you know the total capacitance is 4x the parasitic capacitance, so the parasitic capacitance is 1/3 the capacitor you just added, use that to calculate R and keep the capacitor you just added for the final snubber.
THX. By the way: Did you delete your comment on the first video you placed a few hours ago? I can´t find it anymore. You were absolutely right on the breakdown-voltage of the MOSFET giving the Back-EMF of 27V. But I think that I am still right on the other things mentioned in the video (SRF of the tank-circuit formed by the parasitic capacitance of the inductor etc.)
No I didn't delete my comment. I can still see it. The youtube comments section is buggy as hell. If you look at the frequency of the ringing you can calculate the capacitance which is needed for that inductor to ring at that frequency. With just the inductor and the FET in circuit, the ringing is at 257kHz. C=1/(L*(2*pi()*F)^2) = 3.83nF The default parasitic capacitance of an inductor model is not going to be near that high. In fact I think the only parasitic in the inductor that has a default minimum is series resistance as SPICE dictates there must be resistance between every node on the model (presumably to avoid divide by zero errors). If you look at the model definition of the PSMN0R725YLD NMOS model, you'll find "Cgdmin=518p Cgdmax=3.682n Cgs=7.8n Cjo=9.16n". That's what's ringing here's how you can tell: Again, if you look at a different FET, you'll see a different ringing frequency. The IRFP2907 which has a high capacitance rings at 175kHz, so C=8.27nF. The BSS123 which has a very low capacitance rings at 3.147MHz, so 25.6pF. Incidentally, these 3 NMOS models act completely differently to each other and e.g. don't clamp the voltage to their Vds max value. That's because the FET you used has breakdown voltage specified, Bv=27.5 and the others don't. Bv is a parameter normally seen in diode models and only actually used for zener models (The spice model for all diodes is the same for all types of diode, just with different parameters, Cjo is also a parameter of the body diode), so 1 model treats the body diode as a zener and the others don't. That shows why I avoid NMOS and PMOS models in SPICE whenever I can.
Yes I noticed that yesterday also, when I rechecked my simulation du to your comment/critics :-) But when I replace the chosen MOSFET with the "ideal" default-MOSFET of LTSPICE ("nmos"), then I get exactly the resonant frequency of the inductor with the 10pF parallel-capacitance shown as C1. So we are kind of both right. To demonstrate the self-resonant effect it would have been better to use an ideal MOSFET (or the voltage-controlled switch of LTSpice). As soon as you use a real-world FET the much larger drain-source capacitance overwhelms the (here) 10pF parallel to the inductor ans shifts the resonant-frequency to a much lower value. But thank god all of this doesn´t interfere with that I basically wanted to demonstrate here about the damping of an switched-off inductor with a diode-resistor-combination with different values of the damping resistor. Without your knowledge of LTSpice I would have never known about the Zener breakdown-effect implemented in the model of the chosen FET.
Unfortunately, the screen capture program doesn't capture the cursor changes, so we can't see the different LTSpice probes as you are describing them. I'm sure anyone who has used LTSpice can imagine how the cursors are changing on your screen, though! Thanks for these videos, I like how you are using the simulator to help describe a real-world phenomenon and ways to deal with it.
1N4007 ? By the way...if you double click any net to show its signal the LTspice will automatically delete all the older traces. There is a better way to stop this oscillation. Need to do a test first then can confirm you.
Eagerly waiting for your suggestion if there is anything better than a freewheeling diode with a series resistor. Remember that we not only have to suppress the oscillation but also suppress the back-EMF to a reasonable value for all values of inductance between 10nH and 1H.
I think there is a possible solution. First of all, I have used the BSC16DN25NS3 NMOSFET in order not to have problems with the breakdown voltage of the FET. Description of the circuit is like this: The L1 inductor has in parallel a frewheeling diode 1N4148 in series with a 10 V zener (I have used the BZX84C10L). In order to suppress the unwanted oscillations , we use an additional RC snubber in parallel with the inductor (and hence, in parallel with the diodes network). RC snubber values are as follows: For L1=100 uH => Rsnubber=288 ohms and Csnubber=3nF For L1=10nH => Rsnubber=2.88 ohms and Csnubber=3nF Now the question that arises is how to find a snubber which works for the range 10nH to 1H. 2 things to think about and decide on the trade off: 1)We can use the 2.88 resistor and make the Csnubber capacitor very large to damp oscillations for high inductance values, e.g., for 100 uH, Csnubber=3uF seems to damp well oscillations. 2)Use the 288 ohm resistor and Csnubber 3nF and accept the little oscillation when we have low inductance values. The ringing dies away very fast.
Thx for educating your viewrs, is there a way I can communicate with you ,, i want to clear some of my doubts regarding my circuit
You find an email-adress in our online-shop: www.ak-modul-bus.de/stat/ueber_uns.html
I did the calculation for you just so you could see how close your 100 ohm guess was.
The parasitic capacitance of the FET in this circuit is 3nF.
To damp to Q=1, your snubber would be R=183 ohm, C=10nF.
To critically damp, the snubber would be R=91 ohm, C excessively large, say 10uF (for true critical damping, it would have to be infinite). So your 100 ohm and a still excessively large 100nF is close to critical damping.
Rather than deleting all your traces every time, just use F4 to label the important ones, then they won't change.
THX for calculating the optimum value.
Critical damping at Zeta=1 translates to Q=0.5
en.wikipedia.org/wiki/Damping_ratio
Do you have a suggestion for a website where the calculation of the values for optimized components of a snubber is easily explained?
I found a few but they are all not very good for a beginner.
I did have a really good PDF explaining RC snubbers, RCD soft voltage clamps, hard voltage clamps and RCD dV/dt limiters, but I can't find it.
RC snubbers are easy enough to calculate, the hardest thing is knowing which parasitics are resonating. It's almost always an obvious inductance such as a transformer's magnetising inductance or leakage inductance or just an inductor, ringing with the parasitic capacitance of whichever semiconductor is currently turned off.
You can measure inductance, look at the frequency of resonance and calculate the parasitic capacitance from the formulae for inductive reactance and capacitive reactance. The resistance required for Q=1 will be equal to either reactance at the resonant frequency, which is also equal to SQRT(L/C). Q is proportional to R, so to get Q=0.5, multiply that R by 0.5. In the real world, you usually design snubbers for Q=1 to 0.7. Snubber capacitor will be 3-5x the parasitic capacitance.
If you don't know either the inductance or capacitance, you can add capacitance in parallel to the parasitic capacitance until the resonant frequency is halved, then you know the total capacitance is 4x the parasitic capacitance, so the parasitic capacitance is 1/3 the capacitor you just added, use that to calculate R and keep the capacitor you just added for the final snubber.
THX.
By the way: Did you delete your comment on the first video you placed a few hours ago?
I can´t find it anymore.
You were absolutely right on the breakdown-voltage of the MOSFET giving the Back-EMF of 27V.
But I think that I am still right on the other things mentioned in the video (SRF of the tank-circuit formed by the parasitic capacitance of the inductor etc.)
No I didn't delete my comment. I can still see it. The youtube comments section is buggy as hell.
If you look at the frequency of the ringing you can calculate the capacitance which is needed for that inductor to ring at that frequency. With just the inductor and the FET in circuit, the ringing is at 257kHz. C=1/(L*(2*pi()*F)^2) = 3.83nF
The default parasitic capacitance of an inductor model is not going to be near that high. In fact I think the only parasitic in the inductor that has a default minimum is series resistance as SPICE dictates there must be resistance between every node on the model (presumably to avoid divide by zero errors). If you look at the model definition of the PSMN0R725YLD NMOS model, you'll find "Cgdmin=518p Cgdmax=3.682n Cgs=7.8n Cjo=9.16n". That's what's ringing here's how you can tell:
Again, if you look at a different FET, you'll see a different ringing frequency. The IRFP2907 which has a high capacitance rings at 175kHz, so C=8.27nF. The BSS123 which has a very low capacitance rings at 3.147MHz, so 25.6pF.
Incidentally, these 3 NMOS models act completely differently to each other and e.g. don't clamp the voltage to their Vds max value. That's because the FET you used has breakdown voltage specified, Bv=27.5 and the others don't. Bv is a parameter normally seen in diode models and only actually used for zener models (The spice model for all diodes is the same for all types of diode, just with different parameters, Cjo is also a parameter of the body diode), so 1 model treats the body diode as a zener and the others don't. That shows why I avoid NMOS and PMOS models in SPICE whenever I can.
Yes I noticed that yesterday also, when I rechecked my simulation du to your comment/critics :-)
But when I replace the chosen MOSFET with the "ideal" default-MOSFET of LTSPICE ("nmos"), then I get exactly the resonant frequency of the inductor with the 10pF parallel-capacitance shown as C1.
So we are kind of both right.
To demonstrate the self-resonant effect it would have been better to use an ideal MOSFET (or the voltage-controlled switch of LTSpice).
As soon as you use a real-world FET the much larger drain-source capacitance overwhelms the (here) 10pF parallel to the inductor ans shifts the resonant-frequency to a much lower value.
But thank god all of this doesn´t interfere with that I basically wanted to demonstrate here about the damping of an switched-off inductor with a diode-resistor-combination with different values of the damping resistor.
Without your knowledge of LTSpice I would have never known about the Zener breakdown-effect implemented in the model of the chosen FET.
Unfortunately, the screen capture program doesn't capture the cursor changes, so we can't see the different LTSpice probes as you are describing them. I'm sure anyone who has used LTSpice can imagine how the cursors are changing on your screen, though!
Thanks for these videos, I like how you are using the simulator to help describe a real-world phenomenon and ways to deal with it.
Yes. I noticed this only after editing the video because I can´t see the video in full resolution during editing.
Try Schottky with snubber
1N4007 ? By the way...if you double click any net to show its signal the LTspice will automatically delete all the older traces. There is a better way to stop this oscillation. Need to do a test first then can confirm you.
Eagerly waiting for your suggestion if there is anything better than a freewheeling diode with a series resistor.
Remember that we not only have to suppress the oscillation but also suppress the back-EMF to a reasonable value for all values of inductance between 10nH and 1H.
I think there is a possible solution.
First of all, I have used the BSC16DN25NS3 NMOSFET in order not to have problems with the breakdown voltage of the FET.
Description of the circuit is like this:
The L1 inductor has in parallel a frewheeling diode 1N4148 in series with a 10 V zener (I have used the BZX84C10L).
In order to suppress the unwanted oscillations , we use an additional RC snubber in parallel with the inductor (and hence, in parallel with the diodes network).
RC snubber values are as follows:
For L1=100 uH => Rsnubber=288 ohms and Csnubber=3nF
For L1=10nH => Rsnubber=2.88 ohms and Csnubber=3nF
Now the question that arises is how to find a snubber which works for the range 10nH to 1H. 2 things to think about and decide on the trade off:
1)We can use the 2.88 resistor and make the Csnubber capacitor very large to damp oscillations for high inductance values, e.g., for 100 uH, Csnubber=3uF seems to damp well oscillations.
2)Use the 288 ohm resistor and Csnubber 3nF and accept the little oscillation when we have low inductance values. The ringing dies away very fast.