BQ24250: Increase load current from 2A to 4A with an external bypass diode?

The bq24250  is a narrow VDC (NVDC) charger with power path. The integrated power path is the feature that you want to implement. You don't need either D1 or D2. Please refer to the figure below. The current path 1 is one directional and has higher priority. The current path 2 is bidirectional to charge the battery or supplement the load current if the DC source is absent or the power not enough.  

For your application, the standalone bq25606 or I2C controlled bq25601, bq24190 are good options as well.  

Thank you for your response. Yes, I understand that there is power-path functionality in BQ24250, and that it will work without any external diodes for currents below 2A. I am attempting to power a load of 4A while charging battery at 2A, so a total of 6A of incoming current. Will BQ24250 support 6A of current going through it?

The datasheet says: "Absolute Maximum Ratings: Output Current (Continuous): IN: 2A, SYS,BAT: 4A. Recommended: Input current: 2A" My interpretation of this is that it can handle at most 2A while I need 6A. (Let's assume the input supply is ~5V). BQ24250 also has input current limiting functionality, with the maximum of 2A (would that limit not trip if the load tries to pull 4A?). Is my understanding of the datasheet incorrect?

What I believe will happen if I try without external diodes, is that the BQ will power the load with 2A from the input and with another 2A from the battery: so the current into the battery will be -2A (discharging) instead of +2A (charging).

One problem with BQ24250 I observed is that even the basic powering from battery is not adequate for the spikes in my load, most likely due to the high Ron. (I witnessed this problem in my alternative circuit without the BQ24250, with a discrete P-FET between battery and load: unless I use a FET with a Ron<12mOhm or so, Vload droops on spikes). BQ25895 has a lower Ron of 11mOhm, so it probably won't have this issue.

One source of my confusion was assuming 5V Vin. I know understand that I need to increase Vin such that I get enough power within the maximum input current limit. The BQ24250 is still too low current: Pin is limited to 12W and output current is still limited to 2A, so even going to Vin=9V won't help. However,  with BQ25895, despite the input limit of 3.25A, we can get the max 5A output, by setting Vin to 12V. That's sufficient for charging my application (at say 2-4A), but not sufficient for powering the load (additional 4A), so with BQ25895 my application will probably work, but the battery will charge very very slowly if load is at full power.

Perhaps using a BQ rated for 8A  to provide a 4A to the battery and 4A to the load is an option. However, I think a better option is to dedicate a buck to the DC input and another buck to the charger, instead having one buck for both, and an external power-path OR circuit. Because with a single buck, in the DC IN mode, the circuit will buck then boost, which is unnecessary. And, the 8A IC will be hotter. In two-buck solution, the power path feature of BQ remains unused.

I suppose the builtin power-path  feature is useful for systems with Vdd requirement of <3.5V, which can use the "merged" BQ output  from "buck OR battery" of 3.5V-4.2V directly, but not useful in systems where I need 5V, because the "merged" output still needs to be boosted to 5V. So, I am forced to boost back DC IN that was unnecessarily bucked. With an external power path circuit I avoid this inefficiency:



Note that the OR circuit in the dashed box is available in one package IC LTC4412.
For the charger, I'm now considering the simple charger with no built-in power path: LT3651