TPS7A30: TPS7A3001 Humidity Qalification Failure

Hi Greg,

Firstly, we do not hear about too many issues with this type of test, which usually indicates that our LDO's pass when customers subject them to these tests.

1. I would check the capacitors to confirm they are properly rated to maintain stability.  Use the capacitor manufacturer tools and verify at minimum AC applied voltage (0.01 Vrms is usually the lowest) and DC applied voltage to assess the capacitor meets the stability requirements for the LDO.
   
   
2. Verify the PCB is clean and free from contaminants.  If this issue exists then a leakage path may exist to the NR/SS capacitor, which we see multiple times per year from engineers on E2E.  This will raise or lower the NR/SS voltage based on the leakage path.  If the PCB was hand soldered then contaminants can exists even if you cannot see them, even with a microscope.  Isopropyl alcohol or a PCB board cleaning solution followed by a DI bath is an option to fully clean the PCB.

Do you have any underfill or surrounding bonding material on this component?

Do you see any signs of corrosion on the failing LDO when viewed under a microscope?

How many failures have you observed?  Are they all from the same lot?  If you replace them with an LDO from a different lot and retest, do the failures remain?

Thanks,

Stephen

Hi Stephen,

Thank you very much for your response.  

Please find my response below.

1. I would check the capacitors to confirm they are properly rated to maintain stability.  Use the capacitor manufacturer tools and verify at minimum AC applied voltage (0.01 Vrms is usually the lowest) and DC applied voltage to assess the capacitor meets the stability requirements for the LDO.
   1. 1. The capacitors used are Kyocera KGM series X7R dielectric. 
      2. Capacitors tested operating at 90 to 95%RH for 500hours per data sheet with less than 12.5% variation in capacitance. 
2. Verify the PCB is clean and free from contaminants.  If this issue exists then a leakage path may exist to the NR/SS capacitor, which we see multiple times per year from engineers on E2E.  This will raise or lower the NR/SS voltage based on the leakage path.  If the PCB was hand soldered then contaminants can exist even if you cannot see them, even with a microscope.  Isopropyl alcohol or a PCB board cleaning solution followed by a DI bath is an option to fully clean the PCB.
   1. 1. We use a washing machine which measures the cleaning agent contamination levels and does not terminate the cleaning cycle until a specific requirement is met.  This process gives me confidence that there are no contaminants. 
      2. If you feel the NR/SS capacitor could be the failure, would you be able to suggest a capacitor series for our application?
      3. The parts are all installed by automated assembly pick and place machines.

Do you have any underfill or surrounding bonding material on this component?

There is no underfill or bonding material.  The only additional material used after board cleaning is Humiseal conformal coating at ~1.2mils

Do you see any signs of corrosion on the failing LDO when viewed under a microscope?

No.  The parts, after humidity soaking, look fine.  They just do not have any output until they are baked out and power cycled.

How many failures have you observed?  Are they all from the same lot?  If you replace them with an LDO from a different lot and retest, do the failures remain?

We are currently in the process of collecting the lot code information from the devices.  This failure has happened on at least five to six different assembles which we had subjected to the high humidity environment.  We have no reason to feel it wouldn’t happen on every assembly that we have built currently.

Hi Gregory,

Can you please review your capacitance using Kyocera's online tools at AC bias (0.01 Vrms or whatever the lowest value is) and your DC bias (Vout)? Let's confirm that it meets the minimum effective capacitance necessary for stability (2.2uF).

Can you conduct the same test without conformal coating?  This is still a leakage path.  It's notable that the EVM does not have conformal coating and does not fail in the testing.

Can you confirm the handling of the PCB sufficiently prevents contaminants after wash but before conformal coating?  I've seen instances where the technician or engineer taking the assembled, washed PCB's to the conformal coat machine handled the PCB's and left an oil residue on the PCB (after wash but before conformal coat).  The oil residue was left on the PCB during conformal coat and the failure occurred that way.  Please confirm controls are in place and this is not an issue.

Can you (carefully) replace the NR/SS capacitor with another capacitor (take the one from the EVM, for instance, if a trained technician can carefully remove it and the PCB can be carefully cleaned).  Then try the humidity test and see if the LDO still misbehaves.

If you take the failing LDO, carefully remove it and place it onto an EVM, (and after a thorough cleaning of the EVM) can you rerun the humidity test to see if the LDO fails again?  This would help isolate the issue to the LDO and we would be interested in the lot numbers from this batch of LDO's.

Thanks,

Stephen

Good Morning Stephen

        Please find my responses below.

• Can you please review your capacitance using Kyocera's online tools at AC bias (0.01 Vrms or whatever the lowest value is) and your DC bias (Vout)? Let's confirm that it meets the minimum effective capacitance necessary for stability (2.2uF).
  1. The Spicat.kyocera tool does not have an AC bias characteristic, it shows DC bias amongst others.

• Can you conduct the same test without conformal coating?  This is still a leakage path.  It's notable that the EVM does not have conformal coating and does not fail in the testing.
  1. This had been previously completed. Without conformal coating the device fails sooner. 

• Can you confirm the handling of the PCB sufficiently prevents contaminants after wash but before conformal coating?  I've seen instances where the technician or engineer taking the assembled, washed PCB's to the conformal coat machine handled the PCB's and left an oil residue on the PCB (after wash but before conformal coat).  The oil residue was left on the PCB during conformal coat and the failure occurred that way.  Please confirm controls are in place and this is not an issue.
  1. Yes, I can confirm this. I absolutely agree about the contaminants as we have made many efforts to measure the impedance of the water droplets showing how contaminated water measures dramatically lower impedance.

• Can you (carefully) replace the NR/SS capacitor with another capacitor (take the one from the EVM, for instance, if a trained technician can carefully remove it and the PCB can be carefully cleaned).  Then try the humidity test and see if the LDO still misbehaves.
  1. This test is in process.

• If you take the failing LDO, carefully remove it and place it onto an EVM, (and after a thorough cleaning of the EVM) can you rerun the humidity test to see if the LDO fails again?  This would help isolate the issue to the LDO and we would be interested in the lot numbers from this batch of LDO's.
  1. The two footprints of the device failing on our assembly are and the EVM are different. Our assembly uses a BGA while the EVM is a MSOP. 
  2. For this we have wired the devices onto their opposing assemblies and ran testing.
  3. Testing is showing very poor performance for the BGA lot code 25FJ. Can you please review if there have been any issues with these parts? 
  4. Further BGA testing of lot code 364X has been completed with improved performance. These ICs are not completely passing but remain on after humidity. 

Thanks Gregory.  I have passed on the information regarding the lots numbers to our PE manager.  I will get back to you as soon as I have an update on next steps.  I'm looking forward to the testing with the NR/SS capacitor swap.

Thanks,

Stephen

Good Afternoon Stephen,

We have been doing design of experiments to support our root cause of this -5v failure in a humid environment.  The experiments are leading us to many concerns of flux being trapped underneath the part and when introduced to water, the flux offers a high enough level of contamination to make the water more conductive.  Could you please advise what cleaning process you would suggest, for us to try to achieve our highest chance for success, with these QFN devices?  Is there a time constraint in which the components need to be mounted to the board and cleaned within?   

Thank you for your help.  We look forward to your response.  

Hi Gregory,

We enforce using only water soluble flux on our PCB designs. We then use a distilled water bath to fully clean the PCB.  This has been successful for us. 

There are too many types of flux from many manufacturers for us to offer any guidance on timelines.  In general, I have found that the longer the flux remains on a PCB the harder it can be to remove and eventually, flux cleaning solution may be necessary.  I would consult the manufacturer of the flux for guidance on how best to clean it, assuming it is something other than water soluble flux.

I'm still waiting on feedback from our PE manager who is more familiar with these types of concerns and I'll let you know if they have any feedback to offer when I have it.

Thanks,

Stephen

Hi Stephen,

We have been doing lots of research and tests here in high humidity purposely contaminating the water with flux.  Regarding my below statments from my original post, specifically number 3, we have learned that this failure mode happens when we bridge the Enable and the NR/SS pins with ~2Mohms of resistance.  

This likely is steering us towards Ionic contamination underneath the device which is not being cleaned well.  Could you kindly follow up with your PE manager to solicit any suggested guidance for how we clean your QFN device?  Alcohol baths with Deionized water baths have helped, however, any additional information you can add would be greatly appreciated.  

Thanks for your help.  

1. Regulator configured for -5v output per table in specification sheet with supporting input/output 10uF capacitors and 10nF Cbyp and Cnr/ss capacitors.
2. Failure is no output after soaked in humidity whereas the -5v goes to zero volts
3. Upon Failure it is noticed that the NR/SS voltage measures ~-6VDC where in normal operation it measures -1.2VDC

Hi Gregory,

We would recommend contacting the manufacturer of the flux and solder used in assembly for guidance on cleaning the PCB.  They will probably be your experts on this since they are most familiar with the materials.

Thanks,

Stephen

Hi Stephen,

I hope you enjoyed your weekend.  This problem with humidity still presists.  I imagine this is not the first time this device or similar ones have encountered such issues by your customers.  Could you please share any supporting information which lead them to be successful?  We had ran many tests regarding the Vss capacitor and are planning to run a test without it on Monday, is operating this device with no Vss capacitor concerning to you?  

Unfortunately, we are getting quite backed into a corner and unfortunately, if we do not figure this out, likely this week, we will be forced to design this part out.  Any information you have to get us across the finish line without a respin would be greatly appreciated.  

Thanks for all of your help.

Hi Gregory,

From before it looks like you were becoming convinced that the issue was leakage across the NR/SS capacitor.  Is this still your hypothesis or has something changed?

We do not have many customers ask about humidity testing causing Vout to shift, and none for this LDO that I'm aware of. 

We have plenty of history with leakage on the NR/SS capacitor causing Vout to shift.  Most of the time this is not an issue but with contaminants that can cause the leakage, it becomes a problem.  We have many LDO's without the NR/SS capacitor and we can offer suggestions for a new LDO if you give me your most recent list of requirements.

If you remove the Vss capacitor then the LDO will turn on faster.  However you have a Cff installed across the top setpoint resistor so this will protect you against most of the issues you may encounter with turn on (I.E. I don't think you'll have an issue here).  Your noise and PSRR performance will get slightly worse but that's about it.

Thanks,

Stephen