• State TI Thinks Resolved
  • Locked Locked
  • Replies 9 replies
  • Answers 3 answers
  • Subscribers 63 subscribers
  • Views 181 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.

Original question:

BQ25616: Schematic review requested, thermal runaway suspected

BQ25616: BQ25616: Schematic review requested, thermal runaway suspected

Jon Fick1
Prodigy 90 points
Part Number: BQ25616

Tool/software:

I posted several weeks ago and haven't seen this come through, so posting again.  This pertains to my locked post "BQ25616: Schematic review requested, thermal runaway suspected."

=========================

I posted a few months ago regarding our use of the BQ25616J and referred to a few failures of the chip. You did a design review that revealed no misconnections and two instances where you advised decoupling caps be moved closer to the chip. We will include that in the next board spin.

We've just had four units returned from the field, and the BQ25616J in each of the four units exhibits identical symptoms. Essentially the chip draws 170mA from the USB source, has a very low output voltage, does not charge the battery, and gets hot.

Going down the right side of the Functional Block Diagram in the datasheet, these are measured voltages in the four chips.
PMID 4.9V
REGN 4.9V
BTST 5.0V
SW 0.5V
SYS 0.5V
BAT 0.3V
TS 2.9V

These and other measurements seem to confirm that the ACDRV charge pump is alive (although we don't use it), the internal 5V LDO is alive, PMID is alive, but SW and SYS are 0.5V. Disconnecting the downstream voltage regulator from SYS makes no difference. Disconnecting the battery from BAT makes no difference. Replacing the BQ25616 is what resolves the issue.

A failed BQ25616 exhibits a waveform of 8% duty cycle at SW resulting in 0.5V at SYS, whereas a working BQ25616 exhibits an 85% duty cycle at SW resulting in 4.4V at SYS.

Looking at the inductor, it's rates for 0.8A and the saturation current is 0.8A. We are charging at 400mA, so by datasheet formula I think our inductor saturation current provides margin.

We charge from 5V-only USB sources, not PD or QC sources. Nothing is connected to VAC.

So, why are we experiencing these failures?

  • 0
    TI__Mastermind 40635 points

    Hello,

    Are these failed boards on the new board or the old design? Have you tried to use our EVM with the same setup to see if you can make it work? What is the event that causes the chip to be damaged? When replaced, are there any issues? Is it possible handling could be affecting the performance?

    Regards,

    Mike Emanuel

  • 0 in reply to Michael Emanuel35
    Prodigy 90 points

    Mike,

    The failed boards are of the original design layout.  We have not used the EVM as a template for layout.  We can't lay a finger on an event.  The units are sealed, meaning there is no probing going on.  A unit might be running properly...and hours later suddenly the unit is running hot and won't charge the battery.  It doesn't seem to be related to a plug or unplug event, or full charge or fully charged event.  We have no other symptoms at this point in time.

    We were hoping that the measured voltages would reveal a particular area of the chip that is suspect.  Replacing the chip, and being careful to include solder paste under the center chip pad, is what resolves the issue.  Everything just picks up and works after that.

    If we re-spin the board design following the EVM layout, we are still in the dark about the failure mode.  In other words, we have to wait until the customer field tests the units.  If we are in that mode, we will likely look at alternative chips given that the unknowns will be the same.

  • 0 in reply to Jon Fick1
    TI__Mastermind 40635 points

    Hello,

    We really need to see if our EVM causes the same conditions. It sounds like "being careful to include solder past under the center chip pad" resolves the issue. Is it possible this is a manufacturing issue?

    After looking at the layout again, it looks like there is a thermal pad, but the GND node is not connected to the PCB GND. If you look on the red layer, Pins 9, 17, 18. and thermal pad are tied together but not connected to PCB GND. On the blue layer, the GND pour does not connect to anything. Please connect the PCB GND to the chip PGND.

    Regards,

    Mike Emanuel

  • 0 in reply to Michael Emanuel35
    Prodigy 90 points

    Mike,  

    Solder paste under the chip is standard and is not a manufacturing issue with this product.  We checked.  My comment was to ensure you when I replaced the chip I made sure to include paste under the tab.

    I don't know how we could see if the EVM causes the same conditions.  Run it for six months?

    You can see in this screen shot that our top layer connects pins 9, 17, 18, and the thermal pad.  All of that connects to the next layer underneath, which is a ground layer, through five vias.  Thus, pins 9, 17, 18, and the thermal pad are grounded.  Those five vias extend through the board to the bottom layer (blue layer) where there is a small pad area intended to provide a bit more heatsink for the vias.

    We just had another board failure today.  Our programmer wrote, "the charge chip is really hot and the board is not operating now."

    And 

  • 0 in reply to Jon Fick1
    TI__Mastermind 40635 points

    Hello,

    As said previously, I do not see the connection where the thermal pad, pins 9 ,17, and 18 connect to the GND pour for the rest of the board. The GND pad connects to nothing on the bottom layer. In the picture above, they are indeed all connected together, just not to the GND plane itself. Please demonstrate where the thermal pad node connects to the GND node for the rest of the board.

    Regards,

    Mike Emanuel

  • 0 in reply to Michael Emanuel35
    Prodigy 90 points

    Mike,

    Sure.  I think this image will help.  I've thickened the 3D view of the PCB in Altium so that you can see between layers.  There are four layers in this PCB (top, GND, VCC, and bottom). The top layer and next layer down (GND layer) are shown.  You can see that there are five vias extending downward through the BQ25616 thermal pad.  These five vias connect to the GND layer below.  You can also see in the GND layer other vias that aren't grounded have clearance around those vias.  So, the five vias are grounded, which makes the thermal pad grounded, which also connect pins 9, 17, and 18 to ground.  Does that clarify the topology?

    Jon

  • 0 in reply to Jon Fick1
    TI__Mastermind 40635 points

    Hi Jon,

    Understood, it may help to share all the gerbers so I can see the full picture.

    I am starting to wonder about if the inductor current rating is not high enough. With 400 mA ICHG and VSET of 4.208V, let us try a sample ripple calculation. With 5V input and 3.5V output, we get a ripple of 0.7A. With 400 mA ICHG, that gives a peak current of 750mA. This assumes no system load, which I believe you have. Even 50 mA pushes the inductor current into saturation. Please replace the inductor with a higher rated inductor.

    Regards,

    Mike Emanuel

  • 0 in reply to Michael Emanuel35
    Prodigy 90 points

    Mike,

    Thanks!  We'll look into a component swap.  I'm not quite ready to close this, but I wanted to respond immediately.

    Jon

  • 0 in reply to Jon Fick1
    TI__Mastermind 40635 points

    Hi Jon,

    Looking forward to hearing back.

    Regards,

    Mike Emanuel