• State TI Thinks Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 2 answers
  • Subscribers 33 subscribers
  • Views 437 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

DS90UR903Q-Q1: Field failures of serializer.

Oliver Dumas
Prodigy 20 points
Part Number: DS90UR903Q-Q1

Description of the issue:

We have had several field failures of the serializer (PN: DS90UR903QSQE/NOPB), where the 1.8V power is shorting to 3.3V power inside the chip after failure.  When the boards are diagnosed, all devices connected to the serializer are fine, the only device failing is the serializer itself, U34 in the schematic below.

Notes:

  • At first, we suspected ESD:  But we have applied up to 15KV from an ESD gun to all external pins and have damage other parts but we have verified that the ESD high voltage has never propagated through to the serializer.
  • Note that the serializer/deserializer pair below are used to cross an isolation barrier via the use of transformers.
  • Note that the deserializer, U32 is operating with exactly the same power rails and control signals, yet we have never seen it fail.
  • Note that the 1.8V and 3.3V power supplies (schem below) are not sequenced and simply come up based on the regulators feeding them, both regulators are fed by a 5V regulator.  Do you think that having the 1.8V come up prior to the 3.3V could cause the above mentioned failure i.e., 1.8V shorted internally to 3.3V?
    • Have you seen this failure before?

 

 Thanks,

Oliver

 

  • 0
    TI__Mastermind 29565 points

    Hey Oliver, 

    Thanks for the information. I'll review and get back to you.

    Regards,
    Fadi A.

  • 0 in reply to Fadi Abdulhameed
    TI__Mastermind 29565 points

    Hi Oliver,

    1. This is a legacy device, how old is this application?
    2. How many failures are seen so far?
    3. Do you have a failure rate?
    4. What are the top markings of some of these failing parts ?
    Oliver Dumas said:
    Do you think that having the 1.8V come up prior to the 3.3V could cause the above mentioned failure i.e., 1.8V shorted internally to 3.3V?

    Yes, we have seen issues with devices due to incorrect power-up sequencing. Typically if any specifications from the datasheet are being violated then we cannot guarantee the device's performance across PVT. It is also impossible for us to specify exactly what may go wrong but usually with power sequence issues that could lead to either incorrect startup that causes data channel errors, unexpected behavior in digital features such as I2C, GPIO, etc. or even damage to the device in some cases. As far as the PDB pin, if you have a transient spike on your power rails at power-up and your PDB is already high, the spike may cause internal damage, that is why the RC delay is needed. It's always helpful to get a scope shot for power up sequence with respect to all VDDs and PDB. 

    Please see section (8.2.2.3 Power Up Requirements and PDB PIN) in the datasheet - DS90UR903Q-Q1

    When power is applied, the VDDIO supply needs to reach the expected operating voltage (1.8V or 3.3V) before the other supplies (VDDn) begin to ramp. It is also required to delay and release the PDB input signal after VDD (VDDn and VDDIO) power supplies have settled to the recommended operating voltages. A external RC network can be connected to the PDB pin to ensure PDB arrives after all the VDD have stabilized.

    Regards,
    Fadi A.

  • 0 in reply to Fadi Abdulhameed
    Prodigy 20 points

    Hi Fadi,

     

    Thank-you for your fast response:

     

    Please see responses inline with your questions, sorry for my slow response, I responded immediately but just realized that I had to respond through the forum.

     

    Thank-you,

    Oliver

     

    1. This is a legacy device, how old is this application? Around 6 years old, but it may have been carried over from a predecessor product which goes back over 10 years. Has this device been updated, if so, what  are the  changes?  Could you tell me the new part number or revision?
    2. How many failures are seen so far?  It’s difficult to say exactly, I would say about 2%. 
    3. Do you have a failure rate? 2%
    4.  

    Oliver Dumas said:

    Do you think that having the 1.8V come up prior to the 3.3V could cause the above mentioned failure i.e., 1.8V shorted internally to 3.3V?

    Yes, we have seen issues with devices due to incorrect power-up sequencing. Typically if any specifications from the datasheet are being violated then we cannot guarantee the device's performance across PVT. It is also impossible for us to specify exactly what may go wrong but usually with power sequence issues that could lead to either incorrect startup that causes data channel errors, unexpected behavior in digital features such as I2C, GPIO, etc. or even damage to the device in some cases.  Yes, but my concern here is that I reviewed a data sheet from National Semi (2011) which did not even mention power supply sequencing, they only mentioned the PDB input has to come up after the rails are up.  The TI data sheet is dated 2014, is there a difference between the National Semi part and the one we are using?  It would seem unusual that National Semi would not mention power up sequence if it could cause a hard failure as we are seeing.  As far as the PDB pin, if you have a transient spike on your power rails at power-up and your PDB is already high, the spike may cause internal damage, that is why the RC delay is needed. It's always helpful to get a scope shot for power up sequence with respect to all VDDs and PDB.  Yes, as you can see in the schematic, we have the RC delay.

    Please see section (8.2.2.3 Power Up Requirements and PDB PIN) in the datasheet - DS90UR903Q-Q1 Yes, very familiar with this section.

    When power is applied, the VDDIO supply needs to reach the expected operating voltage (1.8V or 3.3V) before the other supplies (VDDn) begin to ramp. It is also required to delay and release the PDB input signal after VDD (VDDn and VDDIO) power supplies have settled to the recommended operating voltages. A external RC network can be connected to the PDB pin to ensure PDB arrives after all the VDD have stabilized.

  • 0 in reply to Oliver Dumas
    TI__Mastermind 29565 points

    Hey Oliver,

    Oliver Dumas said:
    Do you have a failure rate? 2%

    This failure rate for overall 6 years failures or is it a recent fall-out? Are all failures coming from a specific lot or is it spread out? 
    Is the entire fall-out field failures ? What is the mileage on these failing parts?

    Oliver Dumas said:
    Yes, but my concern here is that I reviewed a data sheet from National Semi (2011) which did not even mention power supply sequencing, they only mentioned the PDB input has to come up after the rails are up. 

    Incorrect power-up sequence can lead to incorrect startup that causes data channel errors, unexpected behavior in digital features such as I2C, GPIO, etc. but in rare cases it might cause EOS damage, however, with your reported fall-out, it makes me suspect that it might be related to a different issue.

    To your point, this change was done by TI as part of an internal improvement actions that took place years ago for this device family.  This was a specification change to the datasheet only to more accurately reflect the device characteristics. There were no changes to the actual device. TI would encourage to follow power up sequence recommended in the updated datasheet.

    To get a better idea on this failure mode root cause. If you have a few of these failing parts on hand and de-soldered, you could send a a few loose parts through the quality channel to investigate root cause of this failure mode. 

    Please click the following link  return request submission form to fill out a request to submit for quality investigation. 

    If above link isn't directly linking you to the request form, please follow the steps below:

    1. Click this link Customer Product Returns (CPR) 
    2. Click on the return request submission form highlighted below 

    Regards,
    Fadi A.

  • 0 in reply to Fadi Abdulhameed
    Prodigy 20 points

    Hi Fadi,

    Thanks for your response, we do have a device here that we just removed from the PCB.

    I will follow your product return process and look forward to hearing the results.  Do you know how long evaluation will take?

    Thanks,

    Oliver

  • 0 in reply to Oliver Dumas
    TI__Mastermind 29565 points

    Hey Oliver,

    Sounds good! Investigation timeline will vary depending on the complexity of the failure mode, HW setup it requires, testing site etc. For gross failures such as EOS/EIPD , hot-plug, etc. finding the root cause could be done quicker than investigating a soft functional failure. The quality team will provide you with all the information and timeline details related to this investigation once they receive your request. 

    Regards,
    Fadi A.

  • 0 in reply to Oliver Dumas
    Prodigy 20 points

    Hi Fadi,

    Sorry, I forgot some responses:

    This failure rate for overall 6 years failures or is it a recent fall-out? Not sure, but have noticed for last 3 years, fewer systems were sold initially.  Are all failures coming from a specific lot or is it spread out? No, it's spread out.
    Is the entire fall-out field failures? Yes What is the mileage on these failing parts? Not sure.

    Thanks,

    Oliver

  • 0 in reply to Oliver Dumas
    TI__Mastermind 29565 points

    Hey Oliver,

    No problem! Once your CPR request is submitted, quality team will take over the investigation for the device level analysis and will provide you with the information needed to move forward. Please let me know if anything else is needed from my side. Thanks!

    Regards,
    Fadi A.