BQ24630: Charge State Pulsing when Battery Charged

Hi Steve,

Sorry for the delay. We can start by probing the signals. Using a scope, take a look at VFB, STAT1, STAT2, and VCC. I am looking to see if there is anything wrong with the input power which may be causing the toggling. If STAT2 is not switching, it means that it is going from fault state to charging state. We need to find what and where is the fault located.

If it is indeed a fault, you want to look at SRN, CE, VCC and REGN.

Regards,
Steven

Thanks for the info, Steven.  I'll reply with more info when I take measurements on a board, but here is what I do know:

• Stat1 is the signal that's pulsing at around 1 second period.
• Stat2 appears to always be off - monitored by SW, not a scope, so it's possible it has some activity.  I'll check with a scope.
• CE is SW controlled.  When CE is low, Stat1 is off.  When high and battery needs charging Stat1 is solid low/on.  When the battery is charged, that's when Stat1 starts blinking.  

I'll take a look at SRN, VCC, VFB, and REGEN and report.

- SteveP

Can you send me your schematic? There could be a problem with how the battery is detected, so anything additional on the battery side may be affecting the charger operation and creating the 1 second cycle pulse. A scope of VFB, SRN and battery voltage may give more info on what is the problem in this case.

The charger does have a recharge threshold after the battery is fully charged, but it shouldn't take 1 second for the battery to go below the threshold unless you are discharging a lot of current.

Regards,
Steven

Sure, I have a PDF of the charger page, but how do I attach it to the post?

Hi Steve,

There should be an attachment button when you reply to this post. If I recall correctly, you may have to click "enable HTML" for the toolbar to appear.

Thanks,

Nabeel

That's it - select "Use rich formatting".  Thanks Nabeel.  Schematic attached.

- SteveP

P08_ Battery Management.pdf

Nabeel, thank you for explaining how to attach.

Steve,

Thank you for sending the schematic, appreciate it.

There are a few things on the schematic which are recommended, but I don't think are the root cause of the problem.
I suggest you separate the system cap into 4x10uF.
For input cap you should have 2x10uF and 1x1uF. The 1uF should be as close to Q26 drain as possible for the best filtering, followed by the 2x10uF.
Same for the battery cap, 2x10uF and 1x1uF. The 1uF should be as close to R99 as possible for the best filtering, followed by the other 2x10uF.

R96 resistor value is recommended to be 100k, changing R110 accordingly.

With the schematic available we can now narrow down the possibilities:
1. Pull up TTC to VREF to disable timer.
2. Pull up CE to VREF, removing the Q17 functionality.
3. Probe VFB, VTS, VCC and REGN on the scope.

Regards,
Steven

Thanks Steven.  Can you elaborate on splitting the caps and changing the VFB R?  Adding the 1uF caps makes sense, and any further details (or reference to an app note) would be appreciated.

Problem Summary:  when charging disabled, all seems as it should be.  When actively charging, all seems as it should be.  Battery voltage and current is around where they should be.  When the charge is "complete," the charge on/off pulsing starts.  I'm assuming the charge is complete since I see the constant current phase, then the constant voltage phase for a while before the pulsing starts.  When it's pulsing and I disable charging and re-enable it via CE, the pulsing starts immediately.  

Per your item 3 above, I collected some data from the lab and here's what I found:

Signal	Level/Behavior
STAT1	Tied to an LED to Vin, which is 18V for the measurements below. High (off) when not charging (i.e. when CE is low).   Low (on) when charging. Pulsing when "charge complete" - On for 140ms in a 1.6 second period.
STAT2	Always 3.37V (off, pulled to 3.3V).  No dynamic behavior seen.
CE	3.29V when enabled.  0V when disabled.  No dynamic behavior seen.  No need to disconnect.
VFB	1.8V steady state, except for a blip to 1.9V corresponding to the pulsing (STAT1 low). Same when charging, even though STAT1 is not pulsing.  I didn't measure the 1.9V width, but it looked the same.  The period may have been slower during normal charging vs. the pulsing state.
VCC	17.5V static.  No dynamic behavior.
REGN	6.0V static.  No dynamic behavior.
TS	2V static.  No dynamic behavior.

I did not get a chance to try item 1, but will do so.  Any flags showing in the data above?

- SteveP

Splitting the caps would allow for less ESR on the caps, and less ESR is better for the response of the converter, faster current capability and more reliable because if one cap somehow gets damaged, you still have another in place.

It is suggested to change the R96 resistor to 100k because this is the value we used in our evaluation. It doesn't mean that other values will not work, but it was tested with 100k and 22pF feed forward cap.

As for the problem at hand:
It seems that this could be a problem with battery impedance. As the charging starts, VFB goes above regulation and then shuts down the charging. But then when it shuts down, the battery voltage is below the threshold to recharge, and so it will start to charge again.

See if you can probe both SRN and the voltage at the battery cell. The impedance between the charger and the battery may be too large and will cause this problem. You can also lower the termination current so that you don't have such a big jump in voltage between charge enabled and disabled.

Regards,
Steven

Steven,

Thanks for your info and suggestions.  I believe you are correct in the oscillation mechanism here:  when charging stops, the VFB drops below the charging threshold so charging starts, and then shortly stops because the current is below the charge complete current.  I didn't get a chance to measure SRN, but did some other measurements/experiments to add clues.  First, I changed the termination current from 300mA to 160mA but got the identical results (charge-oscillation).  I was surprised since I expected this change to make a difference.  More measurements yielded the following:

• Vbatt is at 14.6V during the CV charging phase, which is the voltage it's set to.
• During oscillation, Vbatt is 13.3V between charge pulses and 14.8V during charge pulses.  This should yield a 184mV delta at VFB, which is greater than the 125mV charge threshold so this confirms the oscillation mechanism.  Measurements on VFB confirm this (but are less accurate due to the low voltages).
• All these tests are with a battery pack designed for our application.  The results are the same on all 3 packs we've tried so far.  The packs use the TI BQ40Z50-R1 battery manager chip and the BQ40Z50-R1 parameters were set using the TI parameter tool by the battery pack vendor.
• We tried an off-the-shelf battery pack with identical specs and it does NOT oscillate!  So the pack may or may not be the cause, but the pack does influence the problem.

The cell (no pack-level spec) impedance rating is:  "<= 20m ohm at 1kHz AC after 50% charge" per the pack vendor.

My next experiment will be to lower the charge voltage to ~14.0V to see if making it closer to the nominal pack voltage will prevent the oscillation.  I do not understand why the oscillation is happening, or re-phrased why this oscillation doesn't always happen.  I need detailed explanation on why it should not oscillate this way, and therefore how to prevent it.  Thanks.

- SteveP

Steve,

It seems as though the gauge is acting on some protection mechanism and opening the charging FETs. This can increase impedance and could cause the big jump. See if you can get the schematic on the battery pack and bq40z50-r1.

Regards,
Steven

Steven,  Battery pack intervention is certainly possible, but I would not expect the event to happen just as the battery becomes fully-charged.  TI, the pack vendor, and I have reviewed the parameters and haven't found any likely candidates yet, but that doesn't mean there isn't one hiding among all the others.  

Testing update:  I changed the charge voltage to 14.0V and the problem went away!  Still need to find the root cause though.  While making this change I also followed your suggestion of changing the ground-side voltage divider resistor to 100k and then the other to 680k.  So I actually changed two things:  the charge voltage and the resistance/impedance of the battery voltage divider.  The much higher Rs may change the dynamic behavior of the charger some.  

I'm waiting for permission to post/communicate the pack schematic, but I believe it's close to the suggested design in the BQ40Z50 datasheet and has a single pre-charge/charge FET and single discharge FET.  As I mentioned, I have no idea what's in the off-the-shelf pack.

I reviewed the layout carefully and there are no current bottlenecks to/from the battery pack and the battery current sense R has a good Kelvin layout with short traces and the filter caps as shown on the schematic.  

One clarification I need to add - the schematic I posted is the future rev 2 design, which shows the comp value changes I've made (all discussed along this forum thread).  But I should have mentioned that the current design has a BQ34Z100 fuel gauge chip on the main board.  We decided to use the fuel gauge chip in the battery pack instead, so removed it in rev 2 (the posted schematics).  The only impact of the BQ34Z100 is that its current sense R of .01 Ohms is in the pack ground path.  Could this be contributing to the charge oscillation problem?

I don't think that the sense resistor in the ground path would cause any issues. Lowering the charge voltage fixing the issue is good data and could mean that there may be a limit to charge voltage on the battery pack. Can you get a scope shot of SRN and VFB with both charge voltage regulation limits (one at 14V and the other at the original value?

This could give a clue that the issue is behind the gauge itself and not on the charger configuration.