Inductive Ringing

Thread Starter

Gorden

Joined Oct 30, 2019
19
Hello,

Following on from my project driving a solenoid using PWM, i am now trying to improve ringing caused by the inductive load of the solenoid and potentially long driving cable (10M).

I am using around 1Khz 50% for testing, the ringing measured at the coil end is around 300mV decaying in around 250-300ns. The coil is around 400mA at 12V. I have a good, clean signal at the MOSFET gate and diode accross the coil.

I do not what to have to fit anything at the coil end, and expect that it will not be possible to improve on this noise. Of course if i fit the drive circuit at the coil end the ringing is greatly reduced, but this is not what i want to do, it just proves the point.

I just trying to understand and get some opinions if this result is expected, unfixable and acceptable in terms of EMC etc.

Thanks,
 

Thread Starter

Gorden

Joined Oct 30, 2019
19
Yes - i will post a copy later. It's a very basic circuit, microcontroller PWM into low side MOSFET switching the coil on cable, diode across the coil. The gate is driven directly with 820R and 10K pull down.
 

SteveSh

Joined Nov 5, 2019
105
I assume the ringing occurs when the MOSFET turns off? At least that's what I would expect with an inductive load.

Can you slow down the turn off speed of the MOSFET? In general, lower edge speeds == less ringing, overshoot, undershoot, etc.
 

Thread Starter

Gorden

Joined Oct 30, 2019
19
This is the basic circuit, as i said this is just a test setup - i have used multiple MOSFET's, all fully protected types, like VNN1N04.

Circuit.png

This is the waveform across the coil...

DS0000.png

This is a a waveform from 0V at MOSFET end to the +V at the coil end...

DS0001.png

And then the voltage across the power supply...

DS0002.png

As you can see the the effect happens at turn on and turn off -

On closer inspection i see the gate is not quite clean as first thought - But would this really account for the measurements at the coil end.

DS0001.png

The turn on wave form, 2.5us..

DS0002.png

And turn off wave form.

I may fit a gate driver to see how this works.
 

SteveSh

Joined Nov 5, 2019
105
Nice schematic, helps a lot.
OK. Looks like you have the classic case of an unterminated transmission line. What's the distance from the MOSFET to the relay? Is it just a wire hanging in the air, wire over ground, twisted pair?

300 mV ringing for a 12V signal is really not all that bad. If I were to scale that for 3.3V digital, it would be something like 83 mV, which most digital designers would consider very good performance.
 

Thread Starter

Gorden

Joined Oct 30, 2019
19
The distance of the cable connected to the output of the MOSFET to the coil could be upto 10M, it's just a 0.5mm2 control cable.

Driving 12v 400mA at 1Khz over 10 metres of cable - i surpose it's not to bad as is. I am in the process of adding a gate driver to see how this helps.
 

TeeKay6

Joined Apr 20, 2019
572
The distance of the cable connected to the output of the MOSFET to the coil could be upto 10M, it's just a 0.5mm2 control cable.

Driving 12v 400mA at 1Khz over 10 metres of cable - i surpose it's not to bad as is. I am in the process of adding a gate driver to see how this helps.
As the gate driving resistance (now 820R) increases, the notch in the gate voltage waveform may increase a bit; likewise, when driven from a low resistance source (e.g. a gate drive IC), the notch will likely be less. In either case, the notch is normal and is due to capacitive feedback within the MOSFET. Unless you can see some bad effect in the drain current due to the notch, I would ignore it. Read an article about how MOSFETs function to understand better. I agree with @SteveSh that 300mV is not much flyback. Don't forget that the diode has a forward voltage drop.
 

crutschow

Joined Mar 14, 2008
24,299
Is that ringing really a problem for the solenoid?
What is your concern?
I would expect the current through the solenoid would not see a significant current ripple due to that small voltage.
 

Thread Starter

Gorden

Joined Oct 30, 2019
19
Thank you all,

I think I see now that the ringing is not an issue for the solenoid itself, I was concerned about the amount noise in the cable driving the solenoid. However, I think I see this is not so bad.

I did add a MOSFET driver, although this cleaned the gate square wave slightly, no real change in the MOSFET output slopes. But this led me to more questions - I searched "MOSFET Gate Resistor" and found hundreds of results that just made me to be more confused.

If I understand everything correctly: The gate resistor for the VNN1N04 is specified as 330r minimum, if I was choosing this I would have worked on 20mA (assuming the input would be initially shorted to ground - mcu output pin 25mA) so 250r.

So if 330r is the minimum, why use anything else? The VNN1N04 is a logic level device, if I use anything bigger, like the 820r I am using, in theory this would slow the switching slopes.

I think I see that in a different application with a different MOSFET, higher Id, higher frequency my driving device may not be able to meet the drive requirements, and I would have to use a MOSFET gate driver.

Am I understanding this MOSFET correctly, why would you increase the gate resistor?

I can't find an LTSpice model for this device, so I have not been able to test this.

Thanks
 

MisterBill2

Joined Jan 23, 2018
5,159
If someone who can explain the purpose and function of the gate resistor will do that it will help indeed. Guesses are not welcome at this point, it needs to be an answer from real knowledge. I can provide a number of guesses with no promise that any of them are right, but they would all sound good. And it would not help at all, so I will not offer them.
 

SteveSh

Joined Nov 5, 2019
105
If someone who can explain the purpose and function of the gate resistor will do that it will help indeed. Guesses are not welcome at this point, it needs to be an answer from real knowledge. I can provide a number of guesses with no promise that any of them are right, but they would all sound good. And it would not help at all, so I will not offer them.
Which gate resistor? R1, R2, or both?
 

MisterBill2

Joined Jan 23, 2018
5,159
Which gate resistor? R1, R2, or both?
The series resistor, R2 is the one that needs explanation. R1, the resistor from the gate to the source, is a pull-down resistor to remove the gate charge, thus removing the gate to source bias, switching the FET off, if the driving source does not include a means to remove the gate voltage. That part is not a guess. The value of that resistor, 10 K ohms, seems a bit high to me, and it may be related to the slightly slower turn off time shown. But that is an educated guess.
 

SteveSh

Joined Nov 5, 2019
105
In the circuit shown, neither resistor is really needed.

R1 is used to limit the peak current into the gate of the FET, due to its large gate capacitance. With an ideal voltage source as shown, R1 is not needed. If the gate is being driven by the 5V output of a microcontroller, as is implied by the comment, then R1 is used to keep the output current from the microcontroller within acceptable bounds. In many of our applications, R1 is much smaller in value, tens of ohms.

R2 is not needed if the output of the voltage source can drive down to GND. Depending on what the outputs Vol specification is, if it's (say) 400 mV, then R2 might be needed to ensure that the FET is really off. In many of our applications a resistor like R2 is not needed.

Where R2 is needed is if the device driving the gate of the MOSFET is open drain. In that case without R2, the gate would float, not a good idea. The MOSFET would also turn off slowly, or not at all because of charge buildup on the gate.
 

cmartinez

Joined Jan 17, 2007
6,751
In many of our applications, R1 is much smaller in value, tens of ohms.
I'd like to add that a low value R2 is normally used when switching at high frequencies (about larger than 40KHz) precisely to prevent ringing of the gate's Mosfet, and hence spurious undesired triggering. Also, R2 can be much larger if a gradual turn on of the Mosfet is desired, such as when switching large inductive loads. This to minimize possible EMI, or too abrupt a current draw. I speak from experience on the latter case. The drawback in this scenario is that the transistor will tend to run hot.

But normally, when switching of a load at low frequencies is desired, there is absolutely no need for R2. It can be driven directly from an MCU's pin, if it's gate input voltage qualifies as logic-level, of course.
 
Last edited:

SteveSh

Joined Nov 5, 2019
105
I'd like to add that a low value R1 is normally used when switching at high frequencies (about larger than 40KHz) precisely to prevent ringing of the gate's Mosfet, and hence spurious undesired triggering. Also, R1 can be much larger if a gradual turn on of the Mosfet is desired, such as when switching large inductive loads. This to minimize possible EMI, or too abrupt a current draw. I speak from experience on the latter case. The drawback in this scenario is that the transistor will tend to run hot.

But normally, when switching of a load at low frequencies is desired, there is absolutely no need for R1. It can be driven directly from an MCU's pin, if it's gate input voltage qualifies as logic-level, of course.
That's a good point about R1 being used to damp ringing. Most of our SMPS's switch at 400 KHz or higher. Also we are very concerned about switching losses, which impact overall power supply efficiency, so we want the switching MOSFETs to turn on and off as quickly as possible, hence the reason for low value of R1 in our applications.
 

cmartinez

Joined Jan 17, 2007
6,751
That's a good point about R1 being used to damp ringing. Most of our SMPS's switch at 400 KHz or higher. Also we are very concerned about switching losses, which impact overall power supply efficiency, so we want the switching MOSFETs to turn on and off as quickly as possible, hence the reason for low value of R1 in our applications.
Also, now that I remember, a technique I've seen used when switching at such high frequencies, is to install an inverse parallel diode along with R2. This allows a for a quicker gate discharge when switching the Mosfet off. It is my understanding that the undesirable ringing phenomenon only manifests itself when the mosfet is being turned on, and not off.
 
Last edited:

Thread Starter

Gorden

Joined Oct 30, 2019
19
Misterbill2, i agree R2 was not actaully fitted to pull down the gate, but more as protection to remove charge, as you say - it perhaps is a bit on the high side.

The reason R1 is 330R is becuase that's the minimum series input resistance specified in the datasheet, this may be due to the fault monitoring on the input - this is a fully protected device.

cmartinez, i think has made the point, increasing R1 in value would slow down the slopes, i was not taking into account the potential benfit of doing this when driving large inductive loads.

Perhaps i am making to much of this, due to my over-reading the subject.

For this application, 1Khz 0.5A 12V - The datasheet seems to tell me if i fit a 330R resistor on the gate, i obviously wil not overload the MCU pin, i will be able to achieve the Id and the switching slopes will be fine, i can see this on the above traces.

No need for drivers.

But in another application i would have no choise but to understand this fully, to ensure the gate was driven fully to switch the required load, at the required speed.

Is it possible to create a MOSFET LTspice model, or modify an exisiting model for this device?

Thanks
 
Top