Massive ringing with IPP320N20N3

Thought I had found the ultimate MosFet for class D
Have been experimenting IPP320N20N3 instead of my usual IRFB5620.

I really though one of my boards was fried as I en the first test got massive ringing.

I mounted another board up with IRFB5620. Only limited ringing (before mounting of snubbers).
Then changed to IPP320N20N3 ...... see pictures!!!

Boards based on IRS2092 with ZXGD3003E6 buffers and 4R7 gate resistors.
Pre + post filter fb.

Data sheet would suggest this is a real good MosFet, fast, lower Rds(on), higher current capability, low Ciss ... all in all better than IRFB5620 !!

Anyone having experience with these Fets?
What am I missing here?
 

Attachments

  • IPP320N20N3.JPG
    IPP320N20N3.JPG
    118.9 KB · Views: 389
  • IRFB5620.JPG
    IRFB5620.JPG
    105.1 KB · Views: 384
IPP320N

We use these MOSFETs with no issues.
However we use 22 ohm gate resistors with 10 ohm + 200pF snubbers at each MOSFET.

The problem with the IRS chips is that you have no control of the rise and fall times of the chip whereas we use our in house designed discrete circuit where we control the rise and fall times of the driver circuits.

Some of our older designs use the IRS20957S chip with a LM360 comparator up front BUT we use feedback from after the L-C filter.
 
Hi MOER
Thanks for you feedback.
Also using post filter fb.
My implementation of the buffers only uses one gate resistor ... could have implemented using two to have better control of rise and fall times. Will do this for the next version.
Will try to increase the gate resistors. (already tried 10 ohm, without much difference though).
The design also have room for snubbers, but I want to see the ringing come to something more normal before applying that.
 
Hi Reactance

No, it is not. Do you think this is critical? If you have a solid output like the one from IRS2092 I whould not think this would make any difference.

Anyway, I actually tried to omit the buffer all together, taking it out of the circuit and connecting IRS2092 and gate through a 10 ohm resistor ..... same result

Its hard to tell without actually probing the pcb trace at HO/LO, but... I would assume an inductive trace from the IRS2092 could cause turn on/off spikes so a pull-down base resistor would help with floating or possible inductive signal bouncing.

Even though a better FET was introduced, timing characteristics will generate faster edges and that will need to be accounted for. (Do have have resistor/diode taming at the gate.)

Lastly, what is your probing method ? spring loaded ground clip ect..?
 
Last edited:

kASD

Member
2004-12-25 12:24 pm
N/A
All the superjunction mosfets [Optimos/SJ/MDMESH/X2/X3] are trash when used in Class-D applications. They are good in PFC/FORWARD-Converter/Flyback etc. The culprit is abrupt change in CRSS & COSS wrt VDS Graph.

Go with Planar Mosfets instead.
 
Last edited:
Find an attachment with parasitics simulation, the file was adapted from eva's electronic workbench example.

Change the values to match your mosfet model, and see if the resonance fits your expected snubbed values.
 

Attachments

  • snubber_simulation.asc
    5 KB · Views: 19
  • parasitics_model.PNG
    parasitics_model.PNG
    29.3 KB · Views: 342
All the superjunction mosfets [Optimos/SJ/MDMESH/X2/X3] are trash when used in Class-D applications. They are good in PFC/FORWARD-Converter/Flyback etc. The culprit is abrupt change in CRSS & COSS wrt VDS Graph.

Go with Planar Mosfets instead.

"
Infineon's 200V OptiMOS™ products are performance leading benchmark technologies, perfectly suited for synchronous rectification in 48V systems,DC-DC converters, uninterruptable power supplies (UPS) and inverters for DC motor drives.
"
:D
 

kASD

Member
2004-12-25 12:24 pm
N/A
"
Infineon's 200V OptiMOS™ products are performance leading benchmark technologies, perfectly suited for synchronous rectification in 48V systems,DC-DC converters, uninterruptable power supplies (UPS) and inverters for DC motor drives.
"
:D

:D
When presented my findings to Infineon engineers in discussion, they themselves accepted the fact that OPTIMOS mosfets are very noisy when compared to their IRF counterparts which are planar in their own Class-D applications. Hence you will never see any serious use of OPTIMOSmosfets in their class-D gate driver application note.

Samething IXYS also iterated that their X3 class mosfets[irony is that they themselves pushed them into class-d apps] are good for low frequency switching under 100k but not suitable for 250k upwards because the gate in SJ mosfets is very noise sensitive and you cannot get away with increasing gate resistors and controlling dv/dt.

From IXYS team lead on ultrajunction mosfets

"i had a discussion with my team again this week, it seems like the way Ultrajunction or superjunction MOSFETs are made, they cannot operate at high frequencies so it is not recommended to be used for class D applications":eek:
 
Last edited:

kASD

Member
2004-12-25 12:24 pm
N/A
Find an attachment with parasitics simulation, the file was adapted from eva's electronic workbench example.

Change the values to match your mosfet model, and see if the resonance fits your expected snubbed values.


Even when using gate beads and heavy snubbing, the ringing reappears when high current conduction happens, no matter how tight the pcb layout is with parasitics kept at minimum.
 

Attachments

  • SJpl.png
    SJpl.png
    56.3 KB · Views: 215
Last edited:
Hi kASD

Thanks. Just the information I needed, as the question was how are these fets different from the usual IRF parts I use, where I see no problems like these. I did not realize how different the the mosfets where before this. Need to do some more reading on SJ and UJ fets ;)

I'll ditch them for my designs, switching at around 400 kHz.

As for snubbing, it is no good to use devices which requires heavy snubbing, resulting in large extra components and a lot of heat and power loss.

If you look at page 85 in this selection catalogue you will actually find the IPP320N20N3 listed for audio amplification!?!?
https://www.infineon.com/dgdl/Infin...N.pdf?fileId=5546d4625607bd13015621522aa012cb

@Reactance, no first measurements not done with spring loaded probe, so they show somewhat worse. But the big question here was why the inherent difference. For spring loaded probe see attachment.

Overall I learned something today .... not bad :up: :)
 

Attachments

  • IPP320N20N3 22R spring probe.JPG
    IPP320N20N3 22R spring probe.JPG
    227.7 KB · Views: 211
I use IPP320N20N3 on my IRS2092S class-d with no ringing.

Try to put antiparallel diode + 2.2 ohm resistor on the gate drive for faster gate turn off. Also increase the width of gate drive pcb traces. I use some of the gate drive pcb plane to cool down the zxgd sot23 drivers.
Interesting!

Tried antiparallel diodes on gate resistors. No change.

What is your switching freq?
All with no snubbers?
So you also use gate buffers in this design?
Post a picture?