BQ24702: Alarm pulse triggered in a strange way

Hi Lex,

Based on the fact that ALARM only goes high for a single pulse, it sounds like there may be some fault condition which is causing ACFET and BATFET to not match their programmed states (please see the Autonomous Selection Operation section on page 22 of the datasheet for more details).

In order to gain a deeper understanding of why ALARM is going high, I would suggest analyzing the ACPRES, ACSEL, ACDRV, BATDRV, and ALARM signals, similar to the example shown below:

Best regards,

Angelo

Hi Lex,

This appears to  be a fault condition as described in the "Selector Operation" on pages 21-22:

Based on your description it seems that the AC adapter removal creates a fault condition that forces a switch to battery backup, and this pulse is an indication of that forced switch as described in the "note" section above.

Regards,

Steve

Hi Angelo,

I made further measurement like the following waveform.

Two questions:

1. When keeping ACSEL high, after disconnecting the AC adaptor(ACPRES goes low), I understand the ALARM will go high, but why ALARM goes low after 91ms? Is that means the alarm signal active under the condition that selector input is not equal to selector output? so this waveform we saw is a pulse alarm?

2. If it's a certain pulse alarm, what's the typical time duration? Is the 91ms an acceptable time?

In the previous waveform I sent (re-attached below), I connect ACSEL to ACPRES, so ACSEL will go low when disconnecting the AC adaptor.

The ALARM signal in this waveform is more like a pulse to me, but ALARM was not triggered after disconnecting AC adaptor until 81ms later.

You can see there is no signal changed when signals ALARM. 

I want to describe more detail about our project. We have an external battery and an internal battery.

The priority to power the system is AC > external battery > internal battery. Which means we're not using /BATDRV pins since we have the dual battery.

Could you also help to confirm the schematic we made? Thanks!

Hi Angelo,

Could you help to confirm my schematic?

I'm not sure if it's OK to make /BATDRV floating.

Thanks!

Hi Lex,

I've reviewed your schematic. It's okay to leave BATDRV floating in your application, but my guess is that this is the reason why your ALARM signal pulses. Since BATDRV isn't directly controlling any FETs, the BQ24702 may detect that the selector MOSFET switches don't match the expected states based on the ACDRV and BATDRV pins.

For the most part, your schematic looks correct, but my main concern is your logic circuitry for connecting EXT_BAT to EXTBAT_VSYS and INT_BAT to INTBAT_VSYS (pictures shown below). With your implementation, a change in the ALARM signal could cause the FETs to turn off and disconnect your system from the battery, thus losing system power. (Correct me if I'm wrong about this since it's definitely possible that I missed something while flipping through multiple schematics.)

Here's another possible approach. You could remove the logic circuitry connecting EXT_BAT to EXTBAT_VSYS and use EXT_BAT to drive the logic circuitry used for connecting INT_BAT to INTBAT_VSYS. This approach could still prioritize EXT_BAT over INT_BAT while removing the need to use the ALARM signal to drive any FETs, which may be unreliable based on your application. The example circuit shown below is highly simplified and incomplete, but hopefully it gives you an idea of the basic functionality I'm describing.

A more integrated solution is to simply use a power mux IC such as the TPS2121, which can detect and seamlessly transition between two power sources. The TPS2121 is capable of prioritizing your external battery when present and quickly switching over to your internal battery if the external battery is absent. It looks like this device would eliminate the need for five FETs and two diodes in your design. The TPS2121 would also remove the need to use the ALARM signal. The downside of this approach is that the TPS2121's RDSon is about 56 mΩ, but if this is acceptable for your application, then this would be the simplest, most integrated solution.

Finally, here are a few discrepancies that I wanted to draw to your attention just to make sure that these are intended design choices:

1) VREG = 13.75 V is unusual for Li-ion battery charging. Are you charging a different battery chemistry with your external battery charger circuit?

2) For your external battery charger circuit, you have a 100 µH inductor, while you have a 33 µH inductor for your internal battery charger circuit.

Best regards,

Angelo

Hi Angelo,

Thanks for your patience and quick response.

But I still looking for the answer to the questions I asked in my previous reply(two waveforms).

If this indeed a pulse, could you tell me the typical pulse width?

And how long it takes after the LT1013 detect there is a mismatch in the selector and signal the pulse?

To answer your concern, (BTW, thanks for your contribution to such a detail review), we will look into the possible changes to TPS2121 in our future project.

But now we still have to figure out the reason why ALARM been signaled since we're struggling in our production line.

For the discrepancies that you mentioned, its because the external battery is the lead-acid battery, like the car battery, and it has the capability with 12V, 17.2Ah.

The internal battery is the pack of several 18650 batteries (3-series, 3-parallel)with 10.95V, 7800mAh capability.

I made an experiment to connect /BATDRV to the select switch PMOS that control the battery output, just like the application in the LT1013 datasheet, and the result is showed in the following waveform.

You can see it still signals a pulse on the ALARM pin.

Another waveform shows there is nothing abnormal on the BATDEP pin, so we can rule out the possibility that the battery is depleted.