mosfet gate ringing issue

Go To Last Post
32 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hy there community,

 

I have the following setup: mosfets IRFP4468 (3 phase bridge), IR2110 fet driver, 22 ohm gate resistors, 3000uf +100nf caps on supply close to fets (fets 30v ,fet drivers are on 12v), atmega328 controlled; on power side 10sqmm cables , on command side 0.22 sqmm wires about 8cm long between gate drivers and fets (there is absolutely no way to shorted these wires).

Now the issue; there is ringing on the rising edge of the low side fet gates freq is about 20-23Mhz (low side pwm-ed, freq 16Khz) , only on the rising edge (for the length of the miller plateau, 600ns long) but not on the falling edge, now the ringing is constantly there when the duty cycle is small about 7us pulse, and the motor is not turning yet, as soon the motor starts to turn the ringing goes away but not all the time on some very few pulses there is but on most of others there is not; if i load the motor up to 6 amps, the ringing is more often about on half the pulses, if i disconnect the motor, ringing stops, if i put a 100ohm resistor in series with the motor (motor stopped) very small ringing. 

I tried 50 ohm gate resistor almost no difference in ringing, i tried 1 ohm gate resistor small increase in ringing, i tried with 20cm wires between drivers and fets to see if wire length affects ringing but very small difference.

The probing was made with small ground lead (NOT the 3" crocodile terminated wire) directly on the fet gate , all three low side ring but none of the high side (high side have isolated supply, no bootstrap).

I read a ton about fets, watched almost every tutorial about fets and ringing, but i can not figure this behavior. Again i can not shorten (shorter than 7-8cm) the wires between drivers and fets.

I attached some snaps from oscilloscope sowing ringing and not ringing

Attachment(s): 

This topic has a solution.
Last Edited: Tue. Dec 3, 2019 - 12:35 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 1


Post your schematic!

 

It's a lot easier if you make the images visible in the post - see Tip #1 for instructions.

 

For example:

 

Top Tips:

  1. How to properly post source code - see: https://www.avrfreaks.net/comment... - also how to properly include images/pictures
  2. "Garbage" characters on a serial terminal are (almost?) invariably due to wrong baud rate - see: https://learn.sparkfun.com/tutorials/serial-communication
  3. Wrong baud rate is usually due to not running at the speed you thought; check by blinking a LED to see if you get the speed you expected
  4. Difference between a crystal, and a crystal oscillatorhttps://www.avrfreaks.net/comment...
  5. When your question is resolved, mark the solution: https://www.avrfreaks.net/comment...
  6. Beginner's "Getting Started" tips: https://www.avrfreaks.net/comment...
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Maybe you need an RC high pass filter to GND, to conduct the noise away?

I think you can get some good ideas from here: https://micro.rohm.com/en/techwe...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Inductance in the 0V current path?
What is happening on the drain? Is it the load coupling back via gate/drain capacitance?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Since you say it goes away when you unplug the motor, I'd join Kartman on this, because a motor is a giant inductor (considered a certain way).

 

You mentioned caps on one rail - 3000uF & 100nF - but then you mentioned there were two rails (motor power and driver power).  Do both have the same capacitance handy?

 

Finally - Ground, Ground, Ground.  Or Earth, Earth, Earth, depending.  Seems like you've got a lousy return path.  That would be the first place I would be looking at.  Best!  S.

 

Edited to add:  Depending upon how large your motor is, 3,000uF might be a bit small.  Try 33,000uF.  S.

Last Edited: Thu. Nov 21, 2019 - 01:48 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 2

When turning on the mosfet, (it begins to conduct) Vds drops very quickly and that causes the charge in Cgd (aka Crss or feedback capacitance) to get dumped into Cgs which fights the gate driver and forces Vgs toward zero tending to momentarily turn the mosfet back off.  You can mitigate this effect by adding capacitance between the gate and source.  This will have a twofold effect, first it will make Vgs less sensitive to the charge dump from Cgd and second it will tend to reduce the dV/dt of Vds lessening the impact of inductance between the gate driver and the gate.  However, you may find that the required added capacitance will reduce the dV/dt of Vds to the point that the mosfet  spends too much time in the linear region thus negating the advantages of the switch mode topology.  If this is the case, the best solution would be to choose a different mosfet.  The key here is the ratio between Ciss and Crss.  You want Ciss/Crss  to be as large as possible.  For the IRFP4468 Crss=540pF and Ciss=19860pF which gives a ratio of 36.8 to 1.  I would suggest finding a suitably rated mosfet where this ratio is 400 to 1 or higher.

Letting the smoke out since 1978

 

 

 

 

Last Edited: Thu. Nov 21, 2019 - 03:08 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The setup is battery powered from 8 li-po cells, and from that supply with 3000uf on it goes to motor power and thru buck converter then linear regulators to the driver and command modules,

i'l look carefully at the ground traces and cables again, i also suspect that is something on the ground, but i'm baffled by the fact that nothing else (three micro controllers, sensors, comms, nothing) in the circuit is affected by this ringing , all woks great, witch makes me ask myself what level of ringing can be acceptable?....

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ok, but then why only on low side? the high side is in the the same way but not referenced to the ground and isolated supply, and then why load affects the severity of the ringing, it must be something on the ground path since is affecting only the low side....

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

i'l scope in the morning the gate and the drain together and i'l post a snapshot.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

i scoped the gate and drain together same ringing behavior, i made a few snapshots of this,

and i made a snapshot of the layout of the gate drivers (copper fill is removed),

i scoped again the high side gate with respect to Vs (Vs for driver and source for fet) and no ringing at all.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

A 4-layer PCB with a solid ground plane would improve the ringing significantly.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

well, i don't want to be rude, but i'm very sure with high tech i't easy to achieve a lot, this is not the point here, i'm trying to squeeze every micro gram of performance with simplicity, hence the command module is driven by 8 bit AVR not the 32bit ARM.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

 

I hope you aren't running your gate drivers with only measly 0.1uF caps...the drivers are made to supply AMPS of high current pulses into the gates...give the chips some serious caps for energy storage.

However, you ALSO need ceramic caps, as many electrolytics have awful high freq response.

 

Also beware of scope probe issues...even if well-placed  systemwide gnd dips can play funny tricks against a grounded scope...if you can use a batttery powered scope, that can be helpful.

 

Also, if the fets are connected by wire to the drivers add  a ferrite bead RIGHT AT the gate pin...this will make a huge difference, I guarantee it. Your problem will be solved.  And you will never have another problem surprise

https://www.digikey.com/products/en?keywords=1934-1015-ND

 

you will go from

 

to this

 

 

 

 

 

 

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Last Edited: Fri. Nov 22, 2019 - 10:50 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

What does that have to do with the PCB design?

 

This is basic physics: you need good low-impedance, low-inductance paths - or you will get ringing!

Top Tips:

  1. How to properly post source code - see: https://www.avrfreaks.net/comment... - also how to properly include images/pictures
  2. "Garbage" characters on a serial terminal are (almost?) invariably due to wrong baud rate - see: https://learn.sparkfun.com/tutorials/serial-communication
  3. Wrong baud rate is usually due to not running at the speed you thought; check by blinking a LED to see if you get the speed you expected
  4. Difference between a crystal, and a crystal oscillatorhttps://www.avrfreaks.net/comment...
  5. When your question is resolved, mark the solution: https://www.avrfreaks.net/comment...
  6. Beginner's "Getting Started" tips: https://www.avrfreaks.net/comment...
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The problem is you have poor layout -  physics rules. I had a case where the mosfet was oscillating at 70MHz - I could have made a remote control transmitter with it! It was poor board layout. 

 

Understand where the current is flowing and make sure it flows where it should. All your paths have inductance - seek to minimise this to ensure they don't become tuned circuits.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

yeah it's like i was blind, so fixated , good point there, fresh eyes looking at the circuit.

the high side has each own 10uF cap (between Vs and Vb), wile the low side does not, i'l put some 100uf and see what it does,

thank's for pointing that out!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You still haven't shown a schematic.

 

Kartman wrote:
All your paths have inductance

and remember that includes the wires which connect from the board to the power supply, motor, etc.

 

hence some good, clear, in-focus photos of your setup would also help.

 

 

Top Tips:

  1. How to properly post source code - see: https://www.avrfreaks.net/comment... - also how to properly include images/pictures
  2. "Garbage" characters on a serial terminal are (almost?) invariably due to wrong baud rate - see: https://learn.sparkfun.com/tutorials/serial-communication
  3. Wrong baud rate is usually due to not running at the speed you thought; check by blinking a LED to see if you get the speed you expected
  4. Difference between a crystal, and a crystal oscillatorhttps://www.avrfreaks.net/comment...
  5. When your question is resolved, mark the solution: https://www.avrfreaks.net/comment...
  6. Beginner's "Getting Started" tips: https://www.avrfreaks.net/comment...
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Don't forget to wear your best beads to the party:

 

It has been found that a ferrite bead combined with a resistor on each MOSFET gate eliminates parasitic oscillation while minimizing switching losses...adding a ferrite bead is more effective than using gate resistance alone because the impedance of the ferrite bead is directly proportional to frequency....impedance of the ferrite bead to oscillation noise is 25 to 125 times ... This high impedance is extremely effective at blocking drain to gate noise current...parasitic oscillation can be completely and reliably eliminated... bead can also be used with a single MOSFET that is not connected in parallel with any other MOSFETs. The effect is the same. 

 

https://www.microsemi.com/docume...

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


  I suspect that you have a hard switching for the low side and this can be due to two reasons:

- bad dead time when the low side turns on

- capacitive working mode

 

  Look at this picture (from post #10)

  There is only a small decrease in drain voltage when gate voltage is above the threshold ~ 4.5V. This tells that the high mosfet is still on, or that the internal drain source capacitance plus diode capacitance keep the voltage high. Anyway, you have a huge current in there. Only when the drain voltage reaches 8V, (it is not the 8V that makes the drain voltage to drop, it is the time in between the two points I am talking about) the drain voltage drops and with it the current. This huge change in current makes everything to oscillate, including your probe and all that measurement. It shows drain negative voltages, in reality there are no such multi volt negative voltages in there. It is that big electromagnetic field that interefers with your measurement. Running like this the mosfets get damaged little by little.

  It is not easy to drive a three phase mosfet bridge at 16kHz with an mega328. Also, try and keep your bridge in inductive mode.

 Capacitive mode happen if you try to break with the motor. However, if the pulses are right, you would see hard switching on both arms of the bridge. Therefore, look at the gate pulses in terms of width and dead time. You say that at low duty cycle, maybe your mega328 is not able to achieve such short pulses correctly.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

On the snapshot that you describe, the motor is stopped, so the high side on one half bridge is on and the low side on other half bridge is on with pwm on it, so there is no cross conduction, the dead time on the fets is about 700ns i checked that.

The phenomena that you describe is due to what digitalDan described upper in the comments " When turning on the mosfet, (it begins to conduct) Vds drops very quickly and that causes the charge in Cgd (aka Crss or feedback capacitance) to get dumped into Cgs which fights the gate driver and forces Vgs toward zero tending to momentarily turn the mosfet back off.

There is no ringing on the high side because Vds goes to V+ when fet turns on and Cdg has to go from V+ to 12v above that and that change is with same polarity, but on on low side Cdg has to go from V+ to 0 and the switch polarity up to 12v, and that huge change causes this ringing.

I tried every thing, caps on MGD-s, resistors beads on gates , i even shorten the wires to about 2cm (just for testing) and still no joy, the ringing goes away in to cases: 1-st is i remove the load witch meas Vds is always 0 and Cdg is also 0, second is when: low side was on in the before commutation step (and still on in the present) and in between these 2 steps there is a 500ns dead time (time where high side switches from on half bridge to another) and during this time Vds does not go all the way from 0 to V+ and because of that Cdg does not charge all the way to V+ and then when it has to flip polarity has a lower voltage to swing from. 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I tried every thing, caps on MGD-s, resistors beads on gates

Just to be extra clear, the beads are ferrite beads, not resistors (though a resistor often is placed between the gate driver & the bead).  Have you actually tried ferrite beads right at the gate pin? 

 

Next time you layout the board, use the bottom layer as a gndplane & route all your traces on the top layer.  Since this is such an easy board, there will only be very very few places you need to put a short trace on the bottom layer. You will end up with a bottom layer that is 90% gnd

 

If you want more help, a schematic will really be needed!  

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Last Edited: Sat. Nov 23, 2019 - 10:00 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

avrcandies wrote:
If you want more help, a schematic will really be needed!  

+1

 

  I don't have a six channel scope, you may have. If not ,I would use a logic analyzer to see all six gate voltages at the same time. It would be helpful to see the timing of all six when the issue is present.

 

Mihai_F wrote:
There is no ringing on the high side because Vds goes to V+ when fet turns on and Cdg has to go from V+ to 12v above that and that change is with same polarity, but on on low side Cdg has to go from V+ to 0 and the switch polarity up to 12v, and that huge change causes this ringing.

  The Vds goes to zero.

  I am convinced that the two mosfets in a half bridge are in the same situation (providing that they are correctly driven).

 

  Only the low side mosfets are doing pwm ?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yep ferite beads on gate pin, and resistor in just before the pin, unfortunately the schematic is not available at this time, due to various reasons. I wil try to a a cap between gate and source like digitalDan suggested, and see what happens.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Nope they are not, this is what i thought, but after a bit of analysis i figured thai is not, becouse on high side Vgs is floating and is referenced to Vs witch swings from Gnd to V+, and Vds goes to V+ not to Gnd, so Cgd does not have to witch polarity. draw it on paper, it will show up

Last Edited: Sat. Nov 23, 2019 - 11:19 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

More on that and the ratio Cgs/Cgd in

MOSFET Gate Drive Circuit Application Note | Toshiba

[page 18]

4.4.3. Gate malfunction prevention

...

If dv/dt [dvDS/dt] has a very sharp slope, a voltage is applied to the gate of a MOSFET, depending on the ratio of the gate-source capacitance to the gate-drain capacitance. If this occurs, self-turn-on might occur. Self-turn-on could also occur when a rapidly changing voltage is applied to a MOSFET in the off state during diode reverse recovery. There are three ways to prevent the self-turn-on phenomenon:

(1) Adding a capacitor between the gate and source terminals

...

(2) Miller clamp circuit

...

(3) A turn-off gate voltage can be driven to a negative value so that it will not exceed Vth.

...

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 1

Well, i added a 10nF cap between gate and source, an half of the ringing was gone, then i put a 22nF cap and almost all the ringing was gone , only 2 small spykes remained, now the turn on time doubled, but i might be able to reduce that by reducing gate resistor and hopefully not getting back the ringing. i'l investigate more in the next days.

i added a snapshot of the gate with almost no ringing (with motor stopped as before when there was lots of ringing)

in the next days i'l ad a few pictures of the controller and the motor to show you on what i'm working.  

Anyway, thank you all for your time and help, it was greatly appreciated!!

almost no ringing, motor stopped as before when there was lots of ringing

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It is not clear from your jumble of statements what you are tying to do right now at the moment.  Are you concentrating on driving ONE low side fet on/off & get that working properly (leaving all the other fets alone--keeping them on or off)?  If you can't drive each fet by itself properly FIRST, then having them switch together properly is out of the question.   Only the fet to be investigated should be switching at this time.  Then move on.

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

yep only ONE fet is investigated , that is why motor is stopped  witch means high side one fet is constantly on and one fet low side is pwm-ed.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Why not keep the high side off during your test?  Also, you must mean two different leg, or else you will have a dead short when the low side comes on (if the hi is always on)

Show a diagram of your multi-fet hookup to make sure there is no misunderstanding.

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

hy i'm back, i'v been busy with work, now, i made a few modifications, 33nf cap and 11ohm gate resistor at this point the ringing is almost almost gone just a few small wrinkles, and the turn on/off are about the same (as when no cap and 22ohm gate r), (gate rise time now is about 500ns before was about 400ns ). 

all the tests were made on test setup with only 6 fets and last tests were made on 80V

here are a few pictures.

scope snapshots, power stage layout, test board and test power stage, motor, and fully assembled controller. (motor has 218mm diameter, power:25kw)

This reply has been marked as the solution. 
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

  Nice, impressive, I wish you be successful with your project.

 

  One note, you have 18 mosfets, where one of them is more expensive than the Mega328. For such a hardware, I would use a controller with peripherals made especially for such application.