BQ25504 - Vbat_ok drops down on Vbat_ov level

Hello Michal,

Could you take the same scope shot but also include the VBAT signal?

Also have you noticed if the changes on VSTOR are related to some other component in your system? They three deviations that you should do not seem to have consistent timing so I am curious if anything in your system could be running on a similar timing scheme.

The 4.7uF capacitor on VSTOR will first try to absorb any system load transients and then the higher impedance super cap will provide current through the PFET that connects VSTOR and VBAT pins.  If VSTOR is drooping due to high load transient, thereby triggering VBAT_OK to go low, then you can do one or both of the following:

1.  Increase the 4.7uF ceramic capacitor on VSTOR so that it can handle the transient without dropping below the VBAT_OK threshold.  CSTOR > I_TRAN *TRAN_TIME / (VBAT_OK_HYST - VBAT_OK) might work if you use VBAT_OK signal to enable system loading.  Only consequences are slightly longer startup time due to larger cap on VSTOR. 

2.  Connect the system directly to the VBAT pin, thereby bypassing the internal PFET's undervoltage protection, which may not be necessary for supercap.  This eliminates the impedance of the internal PFET.

Hello Jeff and Ryan,
thank you very much for your answers and tips. Unfortunately, it did not solve the issue. I perform another testing. I this case, a static load (resistor aprox. 500 OHM) is connected directly to supercaps as Jeff suggest. When Vbat reaches overvolage level then Vbat_OK starts oscillating in the same way as current from harvesting source (in my case: Agilent power source – 2.8V constant voltage). Also, the circuit behaves this way with or without any load. I am attaching a new figure.

Vstor (yellow), Vbat_OK (green), Vbat (blue)

Vstor never drops under undervoltage level and then I do not understand the reason why Vbat_OK signal is still dropping. Vbat stays always constant. Is it possible that BQ25504 is not able to charge such large supercaps (two 60F in series) due to its internal resistance?

This is speculation: When the circuit stops charging supercaps due to an overvoltage level, the supercaps voltage drops and an internal connection between Cstor and supercaps is disconnected and indicated by Vbat_OK signal. Is it possible?

I also try to eliminate internal supercaps resistance by parallel capacitance (10-30uF), but I was unsuccessful.

Thank you for your help.

Michal

How well was your board cleaned of flux?  When I modified an EVM without cleaning it, I see a similar problem (VSTOR overshoots and BATOK drops out momentarily).  If I blow on the board, the problem gets worse (does it more frequently) due to the mositure in my breath.  There is a warning in the layout section of the bq25505  or bq25570 datasheet about how residual flux can form parasitic resistors/capacitors that resize the >1Mohm voltage setting resistors.  Can you try re-cleaning the board?  If that doesn't work, then reduce the size of your resistors by a factor of 100 and see if you still have the problem (assuming your input source can supply the current for those resistors). If that fixes the problem, then residual solder flux is the issue.

The board is professionally manufactured and well cleaned. I tested 4 boards and they behave same way. Maybe I failed on layout level (thin connection between Cstor-C17, Inductor-L3 and IC6).

I will try change resistors values first and then I have to change the layout.

Thank you very much for your help.

Michal

Hi Michal,

I have exactly the same problem on a BQ25570.

For now, I haven't a solution, but I noticed that my problem occur only when VSTOR is greater than ~4.5V.

Did you have already modified your layout? Have you still the problem on VBAT_OK?

Best regards.

I know it sounds unscientific but try blowing on the board.  If the overshoot occurs more frequently during or shortly after you blow, then the problem is residual flux.  The impedances at the VBAT_OV, VBAT_OK and VBAT_OK_HYST pins are very high and they are connected to very large resistors.  Any stray current path can  alter the IC's regulation points.

Alternatively, you can lower all of the resistors by a factor of 10 to 100 and see of if you still have the problem.  If not, then that is problem.

The EVM board vendors that Texas Instruments uses were also skeptical of residual flux creating parasitic resistors  but I finally convinced them and now they perform a special cleaning just for those EVMs. 

Good new everyone!

I solved the problem with Vbat_ok dropping. My devices now have new 4-layer PCB design. I am attaching the top layer as a reference. Mainly, I used PCB polygon to connect inductors and I optimized other layout. The Vbat_ok signal is now stable at fully charge and the problem is fixed.