BQ25570: Unstable VSTOR and VOUT output

Hello Jeff,

Sorry for the delayed response, we wanted to go through a bunch of other tests to see if we could figure anything out.  Thank you for the suggestions but they don't seem to address the issues.  The 0.1 uF VSTOR capacitor you talk about above we checked ours and compared it to the eval board and the distance seems about the same.  Here is our layout (C5 is the 0.1 capacitor):

We tried swapping C5 & C6 to get the 0.1 cap closer and no change, and then just to be sure we jumpered a 0.1u cap directly off pin 19 and still no change in behavior.

Also we tried removing and replacing the MOhm divider resistors and no change there either.  Also we have built 10 boards and found the problem exists on all 10, so given those 2 checks it would seem flux is also not the issue.

Again, the problem is that it seems the BQ2557 is having these momentary lapses where it seems to think the VBAT or VSTOR voltage or whatever is low and during these lapses it will pull VBAT_OK low, turn the VOUT buck regulator off and you'll see a spike on the VSTOR & VBAT lines.  Then it seems to take a good reading close to 100ms later and everything returns to normal.  You can see from our schematic we have U2 (load switch) connected to VSTOR for it's enable rather than VBAT_OK, this is a mismatch of the eval board but we made this change because of the issue we were having of VBAT_OK going low all the time during these lapses.

What we did find, is that if we remove our load from the 3V3 rail by removing U2 then the problem seems to go away, so maybe something is being caused by the load?  What we then found after putting our load back in, was adding an additional 4.7uF and 22uF cap on top of C9 made the problem not quite as bad, as in it still occurs but the temporary lapses seem to occur less often at most 4-20mA bus current levels.  But still if we find just the right current level on our 4-20mA bus we can get it to occur fairly frequently still.  We have tried tons of different value caps at various places all over the board and nothing alleviates the problem completely and only the caps on the load side seem to help at all.

Our only "solution" we have arrived at so far is to put a larger capacitor (470uF) on the load side instead of the 100uF cap and then we can ride through the lapses where the BQ2557 decides to turn VOUT off.  But this really isn't a correct solution and i am still worried at just the right conditions we could get VOUT waveforms like the one in the my original post where it is off so long that the 470uF can't ride through the dip....

Any other thoughts as to what could be the problem causing these bogus readings and lapses?

Thanks,

Jarrod

Hi Jarrod,

From your resistors, I compute VBAT_OV = 5V, VOUT=3.3V and BAT_OK = 4.3V-4.9V.  Is that correct?  What is the input source?  I see that there is a 5V TVS so I assume he input source at VIN_DC will always be below 5V so that the boost converter is always boosting?  Can you provide a scope plot with VIN_DC, VSTOR, VBAT and VOUT when this happens?  If you have a current probe, can repeat but replace VBAT with IVSTOR?

From the layout, C5=0.1uF ground return back to IC VSS pin is quite resistive as it has to go down one via to bottom layer and then back up another via to the power pad.  My original EVM had something similar and even that small amount of additional inductance prevented that cap from filtering the switching noise.  Can you try placing a 0.01-0.1uF capacitor across the VSTOR and VSS (NC) pin like is shown in the datasheet layout section?

Regarding the response time, the charger only checks the resistor dividers every 64ms so it can't respond to transients faster than that.

Regards,

Jeff

Hey Jeff!

Yes you are correct on the resistor settings.  The input source is a 4-20mA loop, and the voltage is always below 5V (level controlled by the voltage drop diodes) so yes always boosting.  Here is the 1st scope plot requested, i do not have a current probe currently so i could not do the second at the moment:

1 (yellow) = VBAT (5V)

2 (blue) = VOUT (3.3V)

3 (purple) = VSTOR

4 (Green) = VIN_DC

I had it at a point where the issue was happening pretty frequently in this plot, but i can get the spikes to be more spread out by adjusting the input current on the 4-20mA loop and also by playing with some capacitors on VOUT, but i have not succeeded in getting them to go away completely : /.

Thanks for the info on the VSS connection, i actually have to be out of the office until Tuesday but as soon as i get back in i will try tacking a cap with VSS very close to Pin 1 and see if that helps!

Thanks,

Jarrod

Jarod,

The only other known issue is the parasitic resistors across the large multi-ohm feedback resistors, caused by residual solder flux.  The datasheet  layout section explains more.  You can test by removing, cleaning and replacing the resistors but on their edge OR you can remove and replace with 10x lower resistance value resistors.  

Regards,

Jeff

Hello,

Just posting an update since it's been awhile, i was able to tack on a wire on a board from the GND side of the 0.1uF cap directly to a VSS pin on the chip and it did seem to rectify the issue.  I am working on getting a few more built up with the extra GND just to verify; hopefully those will be ready next week!  I'll post back on what i find for sure, but hopefully the GND problem like you suggested was the issue!

Thanks,

Jarrod