BQ24780S: The CMPOUT rising itself when CMPIN is still triggered

It appears from the trace that the CMPIN drops almost all the way back to the original 2.440 volts at the second transition. This may be the reason for the transition of the output. Your max and min are very close to the reference point of the comparator, and there is also a hysteresis in the comparator, so this makes interpreting the capture a little difficult, but it still seems to me that the second transition in the output is consistent with the transition at the same point in the input.

What is a little strange in this graph is that the input begins to slowly rise back towards that 2.680V point that originally triggered the comparator low, and yet the comparator output never flips back. Still, the input never quire reaches 2.680V in this graph and that, in addition to hysteresis, may explain this. I would need to see the input rise above the 2.680V point before I would be suspicious.

As to whether the comparator output could be affected by OVP -- the only affect that the external system should have on the comparator (other than the input voltage at CMPIN!) is if something causes one of the register settings that affect the comparator to be reset. OVP would not do this, but it is possible that it could cause something to happen that would. For instance, OVP will cause ACOK to be de-asserted. If this were to in turn to reset the device, it would set comparator register settings back to POR values.

Also, there is a power path latch-off feature that is enabled at POR that turns off the ACFET/RBFET when the comparator is triggered. I don't know your system well enough to know what effect this latch-off might have. Is there any chance that this is related to what you are seeing? (Reference charge option 1 register, bit 3 in the datasheet for more detail.)

The best way to determine if a system event is leading to a change in behavior of the comparator (due to resetting register parameter) would be to read the register values that affect the comparator . Most of the values that affect the comparator are in charge option 1, but if possible, you should take a full register dump before and after the above system event, compare them, and analyze any differences.

Regards,
Steve

I placed a resistor at CMPIN, which is controled by CMPOUT, when CMPOUT is high, the resistor will connect with original resistor divider, and CMPIN will be pulled down a little bit, so that can we ensure CMPOUT won't be false triggered.

So, the transition of CMPIN is due to the transition of CMPOUT, not the other way around. And I catched another figure which is shown below. the blue channel is CMPIN, and the purple channel is CMPOUT. I've drawn a longer time scale here. In this figure we can see that CMPIN is always higher than 2.92V after first transition. So, there must another reason cause the falling of CMPOUT.

As for the POR which is caused by OVP, since I didn't change the register of independent comparator, I don't think there are any affect.

And I don't use the function of path latch-off.

So, is there any new suggestion about this problem?

I was able to verify that when no battery is attached to EVM and input voltage trips the ACOVP, the comparator is disabled.  Since it is OD output with pull-up, it will pull high in this condition.  In the below experiment, input voltage is very close to ACOVP margin so that the ripple actually triggers ACOK low for two short pulses.  You can see that the comparator output is released in a 1:1 with the ACOK signal.  When this occurred, I also verified that the registers were not affected.

(CMPIN is shown AC-coupled in the below.  It is ramping up and down between 2.2 and 2.5 V.)

Note that with the battery attached, ACOK/ACOVP do not have this effect on the comparator, i.e. it retains expected output even at 30V.

Is this consistent with what you are seeing?  Can you verify that you do not have battery attached (or it is deeply discharged) and that you do not see the issue if battery is present?

I will find out more background on why comparator is disabled in this situation and will reply back with this information.

Several days pass by, may I know when we can get the final result?