BQ24297: BQ24297 Failures. Any ideas?

Hey Harry,

I'm going to need a little bit more information, like input voltage applied, battery voltage, current setting, and a power on waveform, and maybe details about the failure more. Under what condition do they fail?

From there, we can look at what may be causing this.


Regards,
Joel H

Hi Joel

The input voltage is 5V. Battery voltage could be from 4.1V down to where the battery internal cutoff starts, about 2.5V.  The peak charging current for the battery is about 1.3A with a 2A limit (chip default).  The charging adapter is actually rated 1A (nameplate).  The "overload" of the adapter has not been an issue. The system load current when operating is from about 700mA to 1.5A (peaks).

We do not know under what condition they fail.  I have not seen one fail personally, only that they are given to me "not working anymore".  We test these units with a Keithley Model 2306 Battery charger/simulator. The charging voltage is set to 5V, the battery simulation voltage is set between 4.1V and 3.0V during test (to simulate the battery at full charge and near empty.  During the test the battery voltage may be switched off as well, which is not different from the condition if the battery internal cutoff circuit disconnects the battery. It is also possible that the charging voltage might be switched off, the same condition as if the system is disconnected from the battery charger.  The currents are limited at 2.75A, well above the normal system currents so that the tester will not shut off except in a gross fault condition. 

What circuit points would you like me to provide scope photos of.  I assume VBUS (+5V), SYS, and BAT. Anything else ?

Thanks for your assistance

Harry Bissell

Hey Harry, 

Thanks for the info. Based on that, I think there are some waveforms we would like to capture:

For all waveforms, capture VBUS, SYS, BAT, and either one of the following depending the waveform below (IBUS, IBAT, or SW):

1) Adapter plug-in event: Capture IBUS in one waveform and SW separate.

2) You mentioned a scenario when the battery test voltage is switched off. Please monitor IBAT in this one. 

3) Any other waveform where a peak transient load occurs, so the transition to 1.5A peak that you mentioned. Monitor SW or IBAT

4) Any waveform with load sequencing or load switching

Lastly, for all of these waveforms, test between 3.8V and 4.1V of your range.

Regards,

Joel H

Hi Joel:  I have not been able to get the scope on this yet, the battery charger/simulator is living at the vendor right now. I was there this morning and found they have set aside about 20 units with the problem I'm describing.  Bad units have excess quiescent current on battery only with all the other circuitry powered off, we expect less than 100uA. Good units all meet this specification.     I removed four of the "bad" chips and replaced them, our boards work correctly after they are replaced.  I metered the pins of the IC directly, from Battery (pins 13-14) to SYS (pins 15-16) read 1.9 ohms. From SYS (pins 15-16) to PGND (pins 17-18) read 8.8 ohms.

I'll attach the schematic.   

Is there any way I could damage the chip by how I sequence the battery and charger. For example, I can turn the battery "on" or "off" and set the simulation voltage,not lower than 1.8V or higher than 4.2V in our test sequence. The charger is either set to 5V, or 0V (on or off).   I expect that its OK to have the battery connected (and set to some voltage) without the charger connected (happens all the time in normal operation).  I could have the battery disconnected (not normal in operation) and have the charger connected. This could happen if there was a deep discharge cycle that shuts off the internal battery protection, the BQ24297 will tey to apply a low current to bring the battery out of this condition).

I have seen no evidence of solder bridges etc under the chip. The vendor checked with x-ray and did not see any problems.495-0551_BATTERY.pdf

Hi Joel:

I'm having the vendor check with an ohmmeter from Vraw to ground BEFORE they apply power for the first time to see if we can see the problem after manufacture, or if we are causing the problem when we apply power.

Hey Harry,

Based on your information, there could be few things to look for. Once is when your turn on the battery simulator. I would verify if there is any voltage spike on either the BAT or SYS pin when you turn it on at the higher voltages (4.0-4.2V). If the regulation response of the simulator is slow, you could see some transient overshoot from its output that could inadvertently damage the charger if it is too high.

The same goes for VBUS. I would check what the transient of turning on VBUS looks like.

For both of the scenarios I described, long wires, thin traces, and lack of a sufficient number of vias could add a good portion of inductance to your lines, making any voltage spikes more aggressive.

And lastly, a few more comments and questions on your schematic:
1) What is the voltage rating of C28?

2) Your BTST cap is lower than we recommend. You have 10nF (10,000pF). We recommend 47nF.

3) We also recommend at least 10uF very close to the BAT pin of the IC. Right now you have a 1uF and 0.1uF.


Regards,
Joel H

Hi Joel

Thanks for the ideas.  C28 is a 25V capacitor.  I can increase the value of the BTST cap. It would be better to change it later (running change) as I have a number of assembled boards now. If you think it is a potential for failure, we can change it now.  Likewise I can get the capacitance up at the BAT pin. I will send some print of that section of the layout. C7 (.1uF) is located right at the battery pin on the top side (pins 13 - 14) and the ground via of that cap returns to the internal ground plane directly beneath the IC. It is an 0402 to keep the inductance down, otoh the via adds some. C8 is located at the connector for the Battery itself. I have the larger cap there.  I could get into a 10uF 0402 at C7, but I'd be limited to a 6.3V part in that case size. Probably enough for the Li-Ion battery in use.

The transients might indeed be an issue. I have a Kelvin connection of sorts to the battery connector, there are 14ga wires to minimize voltage drop over about a three foot run between the simulator and the UUT.  The sense wires go out to near the connector, but for the last six inches I transit to single 24ga wires. The battery leads are only three inches long in the product, but I'm limited as to the size of wire I can put in the connector for the longer wire runs in the test harness.  I will look for overshoots there.

I'll check the battery charger/simulator manual to see if I can adjust the rise times of the supplies. Its possible I might be able to control them. More concerning is that the supply is capable of delivering more than the battery voltage, and the test operators might inadvertently switch the unit incorrectly. Setting the simulator requires you to move one digit at a time, so if the operator went from 3.8V volts to 4.1V... they might get to 4.8V by switching the volt digit first, then the tenth-volt digit down (4.8, 4.7, 4.6....).  Of course I've instructed them not to do that. I'll tell them to turn the simulator channels OFF before they switch voltage, then back ON after they have completed the setting.  It will take a little longer, but nowhere near what it takes to replace a blown chip.

I'll bring the scope with me next time I visit the vendor (probably tomorrow) and look for voltage transients.  I have not personally blown out one of these ICs, but i generally use the real battery with the three inch leads, not the test harness.  

I copied the test procedure from one written by a former engineer. Looking at the manual for the Charger/Simulator suggests that the channel assignment he picked is wrong. I'll change that too.

Thanks for the tips.

HarryBq24297_Layout.pdf

Hi Joel

I'm changing the BTST capacitor to 47nF X7R 16V  (same as evaluation board) and the BAT pin capacitor to 10uF  X5R 6.3V.   The Eval board used 10V part, but my layout is for an 0402 part, 6.3V is the best I can get in that package.  The battery voltage is 4.2V max so I think this will be OK.  

I found that the battery charger works much better as the previous engineer configured it, contrary to the manufacturer's instructions.  I saw immediately fatal overshoot when configured per the manufacturer.   I have added some external capacitance at the output of the charger/simulator, we have not had any failures since I added them.

We will have new production with the changed parts on Monday, I'll advise if this solves the issue. As always thanks for your consideration.

Harry

Hey Harry,

6.3V rating for BAT should be fine. Your battery voltage max should always fall way below this. 6V is also the absolute max of the pin on the charger.

And I do want to note that 10uF is a minimum, so if you find that the voltage rating is low enough to cause a DC derating, then consider increasing capacitance value beyond 10uF. And this goes for any recommended capacitance for the chargers.


Regards,
Joel H

I was able to resolve my issue.  I had two capacitors (see thread) which were undersized. The main problem was that my Battery Simulator / Charger could overshoot when changing programmed voltage, possibly exceeding the device maximums. I added small capacitors at the output terminals of the simulator to snub the overshoot and have had no unusual failures since that time.  Thanks Joel, for your invaluable assistance.

Harry