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.

  • TI Thinks Resolved

CSD87350Q5D: CSD87350 Buck controller application damage

Part Number: CSD87350Q5D

Hi Sir,

Our customer used the CSD87350 with TPS40170 for buck application,

But have some CSD87350 Damage, High side MOS is good Low side MOS D to S is short.

We are check VSW have overshoot 35.2V, We will Fine-tuning Rboot/Rgate/Snubber with customer.

But end customer need some data to Clarify this issue need your support, Thanks.

Q1:If this issue is from Avalanche Breakdown, 

How many VDS will damage CSD87350? 30V or 27V or other level?
If VDS over this level How many time of this Pulse will damage? please share the Calculation formula for me.

Q2:If this issue is from temperature, How many temperature of flash time will damage? or Pulsed current of time will damage?

Q3:If this issue is form body diode Damage, can you share some damage condition for me?

Q4:If this issue is form ESD, Can you share the some suggest for me?

  • Hugo,
    The absolute maximum allowed voltage on the drain to source of either FET in the power block is 30V so anything above that can cause damage to the device, particularly if it is a repetitive event.

    While some vendors try to provide a guaranteed "repetitive" avalanche rating, the science behind predicting how much energy or repetitive avalanche events a MOSFET can handle before failing is murky to say the least. To my knowledge, there is no way to predict when a failure will occur, and exceeding the 30V absolute maximum rating is something we never recommend in application.

    As such, I believe it is highly likely this is the cause of the failure, but if you debug your system so that the ringing is <30V and are still seeing failures, feel free to come back and either reopen this thread or start a new one and I will try to help you resolve the issue.

    Best Regards!

    Brett Barr

    Product Marketing Engineer

  • In reply to Brett Barr1:

    Hi Sir,

    Does TI have similar CSD95472's VIN to VSW(10ns) & VSW to PGND(10ns) data of CSD87350? End customer tell we, they are  can't found out this data in CSD8750 datasheet.

    Our customer tell me bandwidth setting 20MHz the ringing under 24V. what's BW setting of Oscilloscope for your Suggest?

    AS I Know, Voltage Margin is broken down into two sections: VDC and VAC, What's VDC & VAC value of CSD87350? 

    If this product need re-layout,  What's VDC & VAC value is Safety? 80% De-rating for standard?

  • In reply to Hugo Liu:

    Hugo,
    In general, we recommend at least 20% derating between Vin and the absolute maximum breakdown conditions of the FET (in this case 24V). You can in theory operate higher, but that will usually require a snubber.

    I will ask our applications team members if they would permit a small voltage pulse above 30V for the order of a few ns, but in general, if you are approaching the 30V BVDSS and ever going above it during ringing, this will be avalanching the FET and is very likely the source of your failures.

    I will also have to ask about the oscilloscope bandwidth. What switching frequency is your customer operating at?

    Best Regards!

    Brett Barr

    Product Marketing Engineer

  • In reply to Brett Barr1:

    Hi Sir,

    Our customer is used the 20MHz bandwidth for verify,  But as I Know Measure SW need use the full bandwidth or 250MHz bandwidth right.

    Hugo 

  • In reply to Hugo Liu:

    Hugo,
    We usually use 1GHz Bandwidth with 20GS/s. The customer should use the full bandwidth possible to see most accurate result.

    Best Regards!

    Brett Barr

    Product Marketing Engineer

  • In reply to Brett Barr1:

    Hi Sir,

    Do you get pulse above 30V for the order of a few ns from your team?

    Hugo
  • In reply to Hugo Liu:

    Hi Brett

    Sorry for the confuse

    We saw CSD87335Q3 have Vsw to PGND (10ns) voltage rating in datasheet ,But CSD87350Q5 doesn't have it 

    Could you kindly help to confirm the voltage rating of CSD87350Q5 for 10ns?

    Or CSD87350Q5 10ns voltage rating is the same with CSD87335Q3?

    Thanks 

  • In reply to Kai Hsiao:

    Hugo,
    Our apps team said you can use the same 32V that we highlight in the CSD87335Q3D datasheet.

    They again pointed out that the 35V spike you saw in the waveform above, and continuous pulsing of that overshoot is all but guaranteed to damage the device.

    Best Regards!

    Brett Barr

    Product Marketing Engineer

  • In reply to Brett Barr1:

    Hi sir,


    We know that if Vds overshoot over spec. that TI will not guarantee MOS works fine without damaged and we'll try reduce the overshoot under the spec.

    But since we try to increase the frequency that cause overshoot more and test many times (15pcs over a week ), it still can't duplicate MOS damaged.

    Because we can not test for a long time like half a year to get the result. MOS is a semiconductor process. We would like to know is there any reliability

    data of CSD87350Q5D that can set parameter then get the result. For example, set some conditions like temperature, voltage, current...etc. then it can

    quickly get the life cycle of the MOS. The same, is it possible that we input the overshoot value then get the lift cycle data?


    regards,

    Jack   

  • In reply to Jack Liang:

    Jack,
    There is no way to calculate lifetime based on repetitive avalanche. The long term reliability of devices that are repetitively subject to this destructive event is not something that is well understood.

    It could fail after 100 events, or it could fail after 100,000. This is why we do not provide a repetitive avalanche condition in our datasheet.

    Best Regards!

    Brett Barr

    Product Marketing Engineer

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.