TPS561208: Catastrophic failure

Just to be clear, this is not occurring on ALL units, I believe it has happened on 2 (and unsure of the number of total units, but much more than 2) of the units tested.   We are evaluating for our end customer

Just to be clear, this is not occurring on ALL units, I believe it has happened on 2 (and unsure of the number of total units, but much more than 2) of the units tested.   We are evaluating for our end customer

Hi Ken,

1. Is TPS561208 damaged during start-up, or steady state?

2. Every pin of TPS561208 has a body diode, could you test every pin by using multimeter to find which pin is short or open?

Regards,

Jiandong   

Hi Ken,

1. Do you have waveforms(Vin, Vout, SW, IL) during start up? Normally input voltage spike will case this kind of damage.

2. Do this test on damage device, and the reference is GND. Using multimeter's Diode Test function.

Regards,

Jiandong

I may have some difficulty measuring pins 2 & 3 : 

I will have a little difficulty on pins 2 & 3 :

A input voltage spike is exactly the thought I had as well.  Is there a suggestion for mitigation using a zener or transient diode ?

Hi Ken,

TVS surely could help suppress the voltage spike, but it also increases cost. I suggest that you capture the input voltage spike during start up by oscilloscope, to see the amplitude of spike. Then we can discuss solutions.

Regards,

Jiandong.

Hi Ken,

Jiandong will feedback to you early next week.

Hi Ken,

That's really wired Vin waveforms, because at the beginning Vin is very stable but suddenly it has so much spikes. To find solutions to this issue, please answer below questions:

1. What kind of power do you use to supply 12V Vin? An adapter or DC source? 

2. At which pinot of schematic do you measue Vin? It's suggested to measure voltage of C131.

3. Please also test Ven and Vout waveforms just as below figure does. 

Regards,

Jiandong

Sorry Jiandong,  the pictures were not taken at the power up sequence.  Those are actually when the adapter is being unplugged.

This is a "wall wart" adapter, and is unregulated.  Some initial testing was that it was producing 16.55 volts unloaded (nominal), and when loaded with a 51 ohm resister, 13.45 volts.  The "advertised voltage is 12 VDC with a current capacity of 1500 mA.  This unit should draw approximately 150 mA nominally, and a peak of somewhere around 300 mA (at 12 VDC).

Measurements (so far) are at the input to the device.

Will be going back to the office in the morning and will measure as requested.

Some additional questions I have :

1) is there anything that can disable or bypass the hiccup feature ?

2) the failing regulator is for 4v, there is a 3.3v regulator that does not seem affected, what could explain this beaviour ?

Hi Ken,

1) hiccup can't be disable.

2) As far as I know, Vin damage due to voltage spike has to relation with output voltage. As we don't have large database of test results, it could be a coincidence.

Regards,

Jiandong

Vin is Ch 1 (yellow); Ven is Ch 2 (blue) and Vout is Ch3 (pink)

Keeping with the previous setup : Vin is Ch 1 (yellow); Ven is Ch 2 (blue) and Vout is Ch3 (pink), I set the scope for any voltage exceeding 17.8 (17 v + 0.7 diode drop gives an absolute recommended 17.7v).  The spike on Vout is frightening.

Hi Ken,

There's no issues for start-up waveforms. For Vout spike waveforms, it occurs at the moment when Vin start drop, did you unplug the adapter at this timing?  And you can see from below waveforms, the Vout spike is exactly the same as Vin. It indicates Vout is coupling with Vin at the moment when Vin start drop.  It's suggested to check waveforms of SW voltage and inductor current when Vout spike occurs, to verify the root cause of this issue. 

Regards,

Jiandong

Yes, I am seeing the spikes at the unplugging of the power supply.  The indication was that this occurred at power-up.  It now appears that the previous power don may indeed be the culprit

Setup : Vin is Ch 1 (yellow); SW is Ch 2 (blue) and Vout is Ch3 (pink)

Did not capture as large of a spike on Vout after several runs

Hi Ken,

The SW waveforms are basically good. I agree with your comments, a TVS placed in Vin could help clamp spikes and protect IC from damage.

Regards,

Jiandong 

Do you have a specific TVS suggestion ?  We need a nominal 13.8 volts and it would appear a very quick response time