This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

BQ25505: VBAT_OV - ROV1-ROV2 inconsistency

Part Number: BQ25505
Other Parts Discussed in Thread: BQ25504

Assistance greatly appreciated.  I have been thwarted in my efforts to arrive at a consistent VBAT_OV for a Lithium Iron Phosphate 18650 charge voltage.  I seek 3.5-3.6 volts.  Attempts at using ROV1 and ROV2 of equal value, which should result in a charge termination of 3.6 volts have failed.  We consistently arrive at 3.3 volts and no more.  Many circuit boards have been tested.  Manual replacement of the resistors (0402 size) with exquisite attention to cleaning before and after placement on rare occasion results in a different VRDIV waveform, so it appears that ROV1/2 choice will influence this voltage.

Using ROV1 and ROV2 of equal value leads to a VRDIV 2/3rds voltage of 2.44v, which when multiplied by 3/2 should result in a charge termination of 3.66v or so.  Can't get there.  I have VBAT_OK set at 2.6 volts with VBAT_OK_HYST of 3.0 volts. This wide range is driven by a 50 ma load that will dip system voltage to trigger VBAT_OK if the range is too narrow. Furthermore, the device is in sun-induced sleep hence we want maximum charging without load during daytime in an attempt to get back above 3.0 volts for nighttime function.

Question:  The data sheet suggests that the sum of ROV1 and ROV2 should be 13 megaohm nominal, with a minimum of 11 megaohm .  In reviewing this forum I note a suggestion of reducing these resistor values by 10x.  What are the consequences, other than unintentional power cost, of using lower values than the sum of 11-13 megaohm?  I assume that the idea behind lower values is to reduce the percentage impact of flux-created parallel resistance altering the design-specified resistance.  According to this suggestion, I will place 1 megaohm resistors to see if this works.

Thank you for your assistance.

Jim

  • Hi Jim,

    Most likely, it is the parasitic resistor formed by solder flux across your 402 footprint resistors.  If you replace with smaller valued resistors (and accept the higher power dissipation) and the regulation returns to normal, then parasitic resistor is the problem.  There is no other problem than increased power dissipation when using smaller resistors.  

    Regards,

    Jeff

  • Thank you Jeff.  We tested a number of boards these past few days. With 1.3 megaohm resistors all is well.  Charges precisely as predicted to 3.67 volts.  I tested 3 megaohms only to find max charge at 3.3 volts as before. This despite careful cleaning before and after reworking the 0402 resistors.  So, 1.3 megaohms it will be.  At 3.6 volts I am burning an extra 2 microamp or so.  I can live with this.

    This seems to be a common theme amongst those working with the BQ25505 and seeking feedback on this forum.  We fought this issue for months until I considered the flux resistance, and then found the same on this support forum.

    Jim

  • Jeff, I thought that this was resolved, but challenges persist.

    We built 50 boards with 1.3 megaohms for ROV1 and ROV2.  Approximately half charge to the appropriate 3.6 volts.  The other half remain at 3.20 - 3.25 volts.  All have been cleaned thoroughly.

    Questions:

    1) If the waveform on DRDIV reveals 2.42 volts for the 2/3rds VBAT_OV, may I assume that the system will charge to 3.6 volts?  Stated alternatively, regardless of the resistors chosen for ROV1 and ROV2, if the shunt parallel resistance caused by flux results in a 2/3rds voltage of 2.4v, am I correct in assuming that the charge VBAT_OV will be 3.6volts?

    2) As we are not achieving 3.6 volts, is there another resistor impacted by shunt parallel resistance (flux induced) that might terminate charging?  My understanding is the VBAT_OK is an output that optionally drives a GPIO to control an external microcontroller to disconnect load.  ROK1-3 are subject to the same impact of flux resistance.

    3) VBAT_OK does not control when the internal PFET opens or closes the connection between VSTOR and Battery. Correct?

    4) Does charge oscillation at LBOOST stop when VSTOR (hence battery voltage) achieves VBAT_OV?  If so, then LBOOST oscillation suggests that the system does not believe that it has achieved VBAT_OV. Correct?

  • Hi Jim,

    Regarding 1, yes.

    Regarding 2, no.  The OV resistors are the only ones that affect OV termination/regulation.  The only other termination/regulation issue we have found is with noise injection on VSTOR pin, which is the reason for the CBYP noise filtering capacitor needing to be placed close to IC VSTOR.

    Regarding 3, correct. The PFET is opens if VSTOR=VBAT_SEC (first pulse on VRDIV) drops below VBAT_UV ~= 2.1V.

    Regarding 4, yes and yes.

    Regards,

    Jeff

  • Jeff,

    The continuing saga.  So, testing 15 or so boards all with the same resistors (1.3 megaohms ROV1 and ROV2) finds some charging to 3.6 volts (as designed and a good thing) while others stop charging at 3.2-3.25 volts.  Thorough cleaning of all.  VRDIV equal on all at 2.41 volts (approximately).  Have swapped solar panels and batteries to rule out variations in these components (or defective panels or batteries).  Bright sunny days in southern California.  Charged over 2 days.

    Despite staring at these boards and my oscilloscope screen for 1 week straight, I am unable to identify why some charge correctly while others do not.

    Thoughts?

    Jim

  • Jim,

    With 402 footprint resistors, you are much more susceptible to parasitic resistor from solder flux.  To put this in perspective, our original BQ25504 EVM had 805 footprint resistors in 10Meg range and no issue.  When I made the BQ25505/70 EVMs, I used 603 resistors in the 10 Megohm range, had the issue you are seeing on some but not all boards.  I had mixed results when I tried to hand clean my boards with 603 resistors. Eventually I had to change to a board assembly vendor with highest board cleaning capability to prevent the issue. 

    If you are still think that the flux parasitic resistors is the issue you can lower the resistor value (for example, 1/2 of 1.3Megaohm) or you can turn the 402 resistor up on its side or force a 603 resistor on its side onto the 402 pad and retest. 

    Regards,

    Jeff 

  • Jeff,

    Challenges persist. Boards are now with 730k each for ROV1 and ROV2 :::>  2/3rds VBAT_OV at VRDIV is 2.41 - 2.42 volts.

    1) Will conformal coating spray impact VBAT_OV resistor divider with R1/R2 in the 700k ohm range? TO be clear, these boards tested are not sprayed.  Just asking whether in production this is an issue.

    2) Will quality of VREF_SAMPLE capacitor impact VBAT_OV?  If I understand correctly this might impact MPP set point and efficiency, but not ultimate voltage achieved during charging; this should be set by the VBAT_OV divider.

    3) Are you able to recommend a part number for the 10nf capacitor at VREF_SAMPLE?

    4) I have equal number of assembled boards charging to 3.6 volts (per design) while others will not charge higher than 3.25.  I have switched batteries and photovoltaic panels from the good guys to the bad, only to find the bad boards continue to be bad.  So it is not the panel or battery.  Any ideas?

    Thanks for your continued assistance.

    Jim

  • Hi Jim,

    Regarding 1, I have never tested with conformal coating.  The coating will have a finite resistance which could be less than 10Megaohm range.

    Regarding 2, no, unless the MPP value is so low that you aren't getting enough input power to both charge the battery and provide the load current at VSTOR.

    Regarding 3, the only capacitor that I have ever tested is the one of the EVMs (GRM188R71H103KA01) which I believe is soon to be discontinued.  You want a 0.01uF capacitor with the highest insulation resistance (lowest leakage) so it needs to be COG or X7R, with much higher voltage rating than you would expect (i.e. 50V) and in at least a 603 package.

    Regarding 4, is it possible that there is solder flux (or conformal coating) under the IC itself that is creating a parasitic resistor to the power pad?  If you move a bad IC to good board or one of our EVMs, does it function correctly?  If not, then I suggest returning a couple of the ICs through whomever your bought them from.  They will eventually get back to TI for failure analysis.

    Regards,

    Jeff

  • Jeff,

    Can VBAT_OK ROK_1, 2, 3, be reduced to 1 meg range as well (similar to what we have done with VBAT_OV resistors)?  No consequence save for increased power cost?

    Jim

  • Hi Jim,

    You are correct.  The increased power cost is the only issue.

    Regards,

    Jeff

  • Jeff,

    I studied the boost waveform on various boards; those achieving 3.6 volts and those unable to.  I notice that the exponential decay of the ringing for those with higher charging voltage seems to be centered on a higher voltage.  Indeed, the ringing seems to be centered on 3.65 volts for those charging to 3.65 volts, while those only charging to 3.25 volts finds the ringing centered on 3.25 volts.  I assume that the energy in the inductor passed to the battery is at that center-point voltage. All of these boards are identical - same BOM and assembly house.

    What is puzzling is that all of the boards demonstrate the same VRDIV 2/3rds voltage - 2.41 volts.  I have carefully cleaned the VREF_SAMP capacitor, ROV_1,2, the component it self, etc.  I have cleaned flux and whatever off the inductor.  All components are 1% tolerance (resistors).  Capacitors are 10% and inductor is 20% (22 uH).

    As an after thought of the previous assistance you have shared, if unintentional parallel resistance resulting from contaminants was the culprit for off-design charging voltages, wouldn't this effect be evident in the VRDIV voltage?

    Is there any information to be gleaned from the VBAT_OK resistor divider? Specifically, the waveforms seen at OK_PROG and OK_HYST - the center points of the OK resistor divider?

    Jim

  • Hi Jim,

    If the VRDIV is correct, then I would expect correct regulation.  Are you certain that there is no other loading on VSTOR or VBAT_SEC that is causing power limiting?  Did you have a change to solder a "bad" IC onto a "good" board or TI EVM and tests its operation?

    The VBAT_OK circuit is an isolated comparator that only affects the VBAT_OK output.  It should not affect regulation.

    I had another thought.  If you remove the CBYP = 0.1uF from VSTOR on a bad board, does regulation change?  If you change to 0.01uF, does that improve regulation?

    Regards,

    Jeff