Author Topic: MOSFET: RC snubber vs. slower gate drive  (Read 7698 times)

0 Members and 1 Guest are viewing this topic.

Offline TimNJTopic starter

  • Super Contributor
  • ***
  • Posts: 1720
  • Country: us
MOSFET: RC snubber vs. slower gate drive
« on: February 13, 2019, 06:38:32 pm »
Hi all,

For a MOSFET switching a load with some inductance, voltage overshoot appears on the drain (AKA switch node) when the circuit is opened up. With the right (wrong?) parasitic elements present, this overshoot can lead to switch node ringing.

In my experience, there are two basic ways to cut down overshoot and ringing, without changing the MOSFET.

1.) Add RC snubber from drain to source
2.) Increase gate drive resistance to slow down turn off

In my understanding, these two methods sort of do the same thing from different angles. The RC snubber is sort of a tuned circuit intended to have a low impedance at the problem frequency. Energy at that frequency will be dissipated in the snubber, cutting down on overshoot/ringing. A higher gate drive resistance slows down the turn off, reducing the peak overshoot voltage.

Now, to me, it seems like both these things kind of cut down on overshoot/ringing by slowing things down. The snubber must also slow down the transitions too, right?

My question is: When is a snubber more appropriate? When is bumping up the gate drive resistance more appropriate?

Thanks!
Tim
 

Offline dzseki

  • Frequent Contributor
  • **
  • Posts: 524
  • Country: hu
Re: MOSFET: RC snubber vs. slower gate drive
« Reply #1 on: February 13, 2019, 08:12:39 pm »
My limited experience in the field tells me usualy it is the best to use both. For switch off ringing the drain RC snubber is the best. But proper gate resistance is also needed, because too fast ramp up on gate will cause overshoot at turn-on. Sometimes the gate resistance needs to be asymmetrical, ie with series diode shunt.
HP 1720A scope with HP 1120A probe, EMG 12563 pulse generator, EMG 1257 function generator, EMG 1172B signal generator, MEV TR-1660C bench multimeter
 

Offline station240

  • Supporter
  • ****
  • Posts: 967
  • Country: au
Re: MOSFET: RC snubber vs. slower gate drive
« Reply #2 on: February 13, 2019, 08:22:35 pm »
The way to think about it, is the spike itself is a waveform, and the RC snubber is tuned for that frequency.

Lets say the switching is 100kHz squarewave 50% duty cycle, 10us p-p.
The spike appears on the scope as 0.25us, or 1/20th of the positive part of the main waveform.

f = 1/ time;
  = 1 / 0.25us
  = 4Mhz
So the spike is 4Mhz, and you design the snubber for that.


Anyway the trade off between faster risetime, and not needing a snubber comes down to:
a. Mosfet is most efficient when either fully on, or fully off, in-between states produce heat as Drain-Source is higher resistance.
b. Slower rise time means less EMI issues, and less expense in snubber components.
c. The voltage overshoot is also an issue for the poor mosfet, as you need to over specify it's voltage to cover any spikes no matter how short.

The practical RC snubber networks I've been looking at lately (flyback SMPS), all have a diode as well, presumably to reduce ringing.
 

Offline TimNJTopic starter

  • Super Contributor
  • ***
  • Posts: 1720
  • Country: us
Re: MOSFET: RC snubber vs. slower gate drive
« Reply #3 on: February 13, 2019, 09:32:29 pm »
Thank you both.

Dzeski, that's  usually what I've noticed too. Typically, we tune both the gate resistance and the RC snubber.

Station240, that's basically how I look at it too. The overshoot is basically half of a cycle at some frequency. Maybe a couple of MHz. Design such that the snubber will present a sufficiently low impedance at that frequency.


Both the RC snubber and gate turn on/off speed affect efficiency. In an RC snubber, there is always power dissipated in the resistor. And slowing down the transition speed (increasing gate drive resistance) drives up switching losses...Why spread the inefficiency between the MOSFET and the RC snubber? Why not just slow down the MOSFET sufficiently and eat the efficiency loss in the MOSFET alone? Why add extra two parts for the snubber?

I guess those are my big questions. I know my understanding is incomplete/wrong, but I'm curious to know why we usually use a combination of the two.

Thanks.
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: MOSFET: RC snubber vs. slower gate drive
« Reply #4 on: February 13, 2019, 10:19:06 pm »
Why spread the inefficiency between the MOSFET and the RC snubber? Why not just slow down the MOSFET sufficiently and eat the efficiency loss in the MOSFET alone? Why add extra two parts for the snubber?

Snubber is more power-efficient than slow switching. Power controller IC manufacturers happen to have appnotes regarding subject. First one in the search result list:

https://www.maximintegrated.com/content/dam/files/design/technical-documents/design-solutions/DS32-Correct-Snubber-Power-Loss-Estimate-Saves-the-Day.pdf
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: MOSFET: RC snubber vs. slower gate drive
« Reply #5 on: February 13, 2019, 11:55:26 pm »
What the gate resistor is, is: you're still doing an RC, but the C is Cgd, which is small and nonlinear, so you might want to add a fixed C in parallel to swamp out the nonlinearity a bit (beware to avoid making a Colpitts oscillator this way!).  Or use an ancient part (say, IR HEXFET) where Cgd is relatively large.

I guess I don't feel strongly one way or the other.  I think I'd rather sidestep the issue entirely and use a protected switch -- they have built-in slew rate limiting and voltage clamping, no worries. :)

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline TimNJTopic starter

  • Super Contributor
  • ***
  • Posts: 1720
  • Country: us
Re: MOSFET: RC snubber vs. slower gate drive
« Reply #6 on: February 14, 2019, 02:15:39 am »
Why spread the inefficiency between the MOSFET and the RC snubber? Why not just slow down the MOSFET sufficiently and eat the efficiency loss in the MOSFET alone? Why add extra two parts for the snubber?

Snubber is more power-efficient than slow switching. Power controller IC manufacturers happen to have appnotes regarding subject. First one in the search result list:

https://www.maximintegrated.com/content/dam/files/design/technical-documents/design-solutions/DS32-Correct-Snubber-Power-Loss-Estimate-Saves-the-Day.pdf

Thanks for the link. It seems that the snubber does not affect the dv/dt of the transition as much as I thought.

I did some simulation in SPICE with a MOSFET driving a load with some series inductance. Removing the snubber had almost no impact on the rising edge dv/dt at turn-off. With the snubber in place, almost all ringing was eliminated, though overshoot was still there. To achieve the same peak overshoot voltage without a snubber, the dv/dt had to be reduced by about 20% (by changing gate drive resistance). Ringing was still quite strong.

Just simulation, but I kind of see why they are typically used in tandem.
 

Offline TimNJTopic starter

  • Super Contributor
  • ***
  • Posts: 1720
  • Country: us
Re: MOSFET: RC snubber vs. slower gate drive
« Reply #7 on: February 14, 2019, 02:27:32 am »
What the gate resistor is, is: you're still doing an RC, but the C is Cgd, which is small and nonlinear, so you might want to add a fixed C in parallel to swamp out the nonlinearity a bit (beware to avoid making a Colpitts oscillator this way!).  Or use an ancient part (say, IR HEXFET) where Cgd is relatively large.

I guess I don't feel strongly one way or the other.  I think I'd rather sidestep the issue entirely and use a protected switch -- they have built-in slew rate limiting and voltage clamping, no worries. :)

Tim

Thanks, Tim.

So, what difference does it make that Cgd is non-linear, in terms of switching performance?

And I had no idea 'protected switches' were even a thing. Neat!

https://www.onsemi.com/PowerSolutions/parametrics.do?id=819&tab=products
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: MOSFET: RC snubber vs. slower gate drive
« Reply #8 on: February 15, 2019, 12:53:03 am »
The difference is that dV/dt depends on Cdg, so it can be quite high even for large Rg.  Which means an inductive load is more likely to be incompletely discharged by the time it crosses zero.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: ogden

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: MOSFET: RC snubber vs. slower gate drive
« Reply #9 on: February 15, 2019, 02:02:01 pm »
1.) Add RC snubber from drain to source
2.) Increase gate drive resistance to slow down turn off

3.) Some gate drivers actively measure dI/dT and dV/dT and adjust  the gate drive to limit them.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: MOSFET: RC snubber vs. slower gate drive
« Reply #10 on: February 15, 2019, 03:49:00 pm »
Ooh. Examples?

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: MOSFET: RC snubber vs. slower gate drive
« Reply #11 on: February 15, 2019, 05:28:41 pm »
Ooh. Examples?

Linear Technology's low noise series of switching regulators like the LT1683 and LT1738 which use external power MOSFETs and integrated devices like the LT1533 and LT1534 have independently adjustable current and voltage slew rates.  I think I remember a power MOSFET driver also but good luck finding it through Analog Devices' broken selection guides.
 
The following users thanked this post: T3sl4co1l


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf