BQ25890: Device thermal runaway under 5A load

Jeff,

Thanks for confirming.  I realized I made a typo and meant RDS instead of RGS but you seem to have corrected my mistake.  Anyway the 0.5W dissipated only helps slightly, since the thermal characteristics assume that my ground plane is as good as the test coupon used.  I've confirmed this suspicion through http://www.ti.com/lit/an/spra953c/spra953c.pdf.  PCB design has a huge impact on the thermal constants.

As for faults there were none when I last tested but I will confirm when my sample order arrives and I can replace the dead chip. I can't change the board until my next revision but even then may not be able to switch to a 4 layer board (although my current design is tight with only 2 layers considering the largely broken ground plane).  I can heatsink the IC through the aluminum enclosure.  Will the IC dissipate enough heat through the top of the package? 

I am not nearly as worried about charging or operation from VBUS as I am limiting charge current to 1A and the 5A drawn would exceed the connector I've placed on the board for charging.  

With appreciation,

Robert

Jeff,

Thanks.  I'll see when I get the samples in and replace the IC.  There's not much I can do about the ground plane at the current moment but maybe heatsinking the top of the package can extract some heat. As you mentioned the top is a very good insulator so I don't think this will be as effective as increasing copper weight or number of ground planes.  The device will be drawing 5A continuously for up to half an hour so heat is of great concern. If the device limits current or enters thermal shutdown the voltage drop will be great enough such that the LDO for the microcontroller stops outputting.  Another solution I am considering is to have the load powered from VBAT and the microcontroller from SYS.  This way the only other resistance is that of the MOSFET.

Robert

Hi Jeff,

The samples came in and I was able to replace the IC to replicate my issue.  When I attach a dummy load of 1 ohm to SYS and only power the device through BAT after some time the device again ceases to work.  The REG10 bit 7 isn't flipped but the IC gets very hot.  After some time with the load attached (and drawing 4-5A) BATFET is disabled (REG09 bit 5).  After this behavior starts BATFET is instantly turned off when the load is attached.  

What's strange is that the MOSFET also dissipates a similar amount of heat and hardly heats up at all - in fact, I've left the MOSFET on with the load attached for up to an hour and it hardly gets hot.  Please let me know if you have anything else that I can troubleshoot in case my issue is somehow unrelated to thermal performance.  I may simply need to move to a second revision where I revise the layout to improve thermal performance.  However, I would like to confirm that this is in fact the underlying issue first.

Kindly,

Robert

My inductor is rated for 6.5A and the saturation current rating is 7A.  The part is an Eaton MPI4020V2-1R0-R.  I can potentially order an eval board but it will be difficult to hook it up to my circuit without bodging a number of wires. Nevertheless, if you suggest that's where I troubleshoot next I can see if this is possible without ordering a board specificically for testing.

Robert

Robert,

We can try to debug why your board's efficiency is so low.  If you load with SYS with a 5 ohm resistor and 5V on the input and maximum input current limit, what is the efficiency (Vsys*Isys)/(VBUS*Ibus)?  Should be in the ~ 90%. As you increase the load on SYS with 5VBUS, you will eventually need to start supplementing through the battery FET.  If you increase VBUS to approx 7V (as measured close to ICs VBUS pin), the charger should be able to provide the entire 5A load on SYS regulating to 3.7V (3.7V*5A/7V*3.25A) = 81%.   with Po+Ploss=Pin and Po/Pin=0.81 you can find Ploss which is the power that the IC and the inductor must dissipate.  Inductor losses are approximately DCR*ISYS, ignoring ripple. 

Regards,

Jeff