This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LMG5200: Switch Node ringing

Part Number: LMG5200

I understand the absolute maximal voltage on the switch node to be 80V.  However, I am seeing ringing on my switch node during the turn on time of either top or bottom FET, which I'd like to understand how bad of a problem it is.  I understand no amount of ringing is good, so will certainly look into why there is ringing in the first place (I followed all of the layout guide-lines to the T).  In my application, Vin can vary anywhere from 5V to 60V.  When Vin is ~52V, the ringing is as high as 66V for less than 2ns so I stopped raising the voltage to avoid permanently damaging the FETs.  See the attached image for an example of my switch node waveform.  

Is the 80V maximum voltage a hard limit?  Meaning if it is 80V for any finite amount of time irreversible damage will occur?  Or is ~2ns of 80V probably going to be OK? 

I understand there is no body diode in GaN FETs, is that true with this package?  I understand due to the construction of the FET, there is "body diode like" properties, and do believe I see this taking over in some circumstances in my waveform, but that doesn't clamp until the voltage is a couple voltage above Vin or below Ground for more than ~10ns.  So assuming this body diode like property is somewhat slow, relatively speaking.  Meaning the 2ns spikes I'm seeing happen too fast to get caught.  To that end, I'm wondering if really fast and low capacitance schottky diodes would help clamp these spikes I'm seeing.  Thoughts?

Thanks in advance,

Ryan

  • Hi Ryan,

    Thank you for your question. For switching transient events of short duration, like switch node ringing, the 80 V rating is not a hard limit. For these type of short transients, the device can handle up to 100 V. In your application, since the highest Vin is 60 V, there should be no concerns due to the switch node ringing.

    From looking at the waveform, it seems like the switch node voltage may be ringing down to a negative voltage as it is turning on. In order to avoid damaging the device, this transient event should not exceed -5V on the SW  pin, as specified in the abs. max of the data sheet. To control this transient, you could try clamping the voltage using Schottky diodes, as you suggested.

    Regarding the body diode of GaN technology, you are right - GaN devices do not have body diodes as traditional silicon FETs do. However, the device can still conduct in the third quadrant, meaning that the device conducts a current from source to drain with a relatively high forward voltage.

    The ringing in the waveform is likely not due to the third quadrant conduction of the device. Most of the time, it is related to parasitic loop inductance in the high di/dt power loop.

    A follow up question - what sort of topology is this device being used in? This will help us understand the ringing events better.

    Best Regards,

    John

  • Hi John,

    Thank you for the swift and thorough response!  What a great Christmas present.

    I will add the schottky's to the design to ensure I don't damage the FET's.  Good to know the transients won't hurt the device up until about 100V.  I also am suspecting my measurement isn't accurate and the ringing isn't real, so working on perfecting the measurement now.

    The topology is a class-D full bridge amplifier, so there are two of these devices used to make up the full bridge.

    Thanks again, and happy holidays,

    Ryan

  • Hi Ryan,

    No problem - glad to help. If using in a class D amplifier, ensure a load is connected. We have had other customers report a similar ringing issue at the switch node when they are switching only with the LC filter. This filter may add unwanted resonance at the switch node if used on its own.

    Best Regards,

    John

  • That is great to know.  In my case the load was connected.  My load is highly capacitive so assuming the ringing is real and not a measurement error, that could explain why it is ringing.