BQ24125: STAT1 pin alternating between ON and OFF after first test

Hi Ubiratan,

How much resistance is between the top feedback divider resistor and the battery pack?  How old (resistive) is the battery? How high is your termination current setting?  From your data, it appears that there is (8.3-7.8)V/1.78A =281mohm of resistance between the charger sense point and the PACK+. The charger senses that the battery has reached termination voltage due to extra series voltage drop (281mohm*1.78A) then he charger current tapers down to termination current level. The charger then terminates but then the extra series voltage drop goes away, the sense voltage drops close to the actual cell voltage so the sense point drops below the recharge threshold of 100mV/cell and restarts charging.

Regarding the input capacitor, it's ESR is too high to be an input capacitor for a buck converter.  I recommend changing to ceramic or adding a ceramic capacitor in parallel to handle the high ripple current.

Regards,

Jeff

Hi Jeff,

Please find some additional information.

About your questions:

1. How much resistance is between the top feedback divider resistor and the battery pack? 

I am not sure if I understood your question. The top resistor of the feedback voltage divider is a 301k x 1% (R14) and it tied directly to a large copper that comes from the battery connector.

1. How old (resistive) is the battery?

I bought a set of 10 batteries 10 months ago and used only in this project.

1. How high is your termination current setting? 

The resistor connected to Iset1 is a 10k resistor. I need a 2A charge current.

The resistor connected to Iset2 is a 1k resistor.  

1. There are other two input capacitors in parallel with the electrolic capacitor. The input capacitors are:

C20 (22uF): https://www.digikey.com.br/pt/products/detail/w%C3%BCrth-elektronik/865060342002/5728114

C30 (10uF): https://www.digikey.com.br/pt/products/detail/samsung-electro-mechanics/CL31A106MAHNNNE/3886839

C29: 100nF x 25V

Is it important to advise that I am using a thermistor but it is not in contact with the battery. It is measuring air temperature in fact.

Also please find attached the following picture.

1. Stat1 and Vbat waveforms
2. Stat1 and Stat2 waveforms
3. board layout details

 Thank you

.

Hi Ubiratan,

I am referring to the board trace + wires + connector + battery impedance "parasitic" resistance from the sense point to the battery.  When charging, that point sees a voltage = ICHG*Rparasitic + Vbatterypack+.  If Rparasitic is high and/or Iterm is high, then the charger will terminate and then bounce back and forth between charging and termination until the cells are fully charged, which is what your data suggests.

Regarding the input capacitor, the electrolytic one has ripple current and ESR below:

which is not good for the input to any buck converter but your parallel ceramic capacitor with lower ESR should handle the majority of the ripple current.

Regards,

Jeff

Hi Jeff. I am not able to calculate the total parasitic impedance. 

Are you able to give me an advice based on my board layout?

Please look at the picture below showing the feedback traces.

Traces from connector SV2 to R13 are 1,27mm thick. Traces from R13 and R15 to BQ are 0,2mm thick.

I reduced the cable length that connects the batteries to the board. These cables were always soldered in the SV2 pins.

Thank you.

Hi Ubiratan,

The layout looks okay.  How does the input source and battery ground get back to the IC ground pins, through the powerpad vias?  I recommend added multiple vias at the IC PGND pins. The battery ground pin looks like it was thermal relief with small traces connecting to the bottom plane?  I recommend making that connection with thicker traces.

Also, I looked at the BQ24125EVM and it has a 0.1uF capacitor across the top FB resistor.  Also, there is mention of adding a ~10uF capacitor in parallel with the BAT and GND if the battery is connected with long cables.  You might try adding both of these and see if they make a difference, in case my original diagnosis is wrong.

Regards,

Jeff

Hi Jeff, how are you doing ?

Answering your questions.

1) How does the input source and battery ground get back to the IC ground pins, through the powerpad vias?

These are the traces of VCC input. It comes from an SMD connector, pass through three input capacitors and connects to BQ pins 3,4 and 6. Although I am not using the power input connector. I am soldering cables direct to + and - connector pads. See the picture below.

These are the TOP GND traces.

And the bottom GND copper plane. Please notice that the BQ GND vias also have small traces. To try to solve this I removed the green mask of these vias and the copper surrounding them and solder all together. 

I also added the capacitors as you suggested and also made some modifications to reduce the impedance of GND thermals. But the behavior is the same. 

Thank you !

Hi Ubiratan,

Ok.  I am still convinced that the problem is too much impedance between the charger and the battery. You can measure this quickly by setting the charger up to charge at a known charge current.  Then measure the voltage at the IC sense point and at the battery pack terminals.  Then disable charge and measure again.  the voltage drop = Ichg*rparasitic.

Do you have a BQ24125EVM that you can modify to your settings and test?

Regards,

Jeff

Hi Jeff.

Another question is if my battery and BQ12245RHLT are appropriate to work together. Please find below some battery info.

Thank you !

Hi Jeff, I don't have the eval board, but if necessary I can order one. Is it possible to charge the batteries I am using ? Please look at the battery info I posted some minutes ago.

Hi Ubiratan,

I assume you have 2 in series.  Max charge current is 2.17A.  I don't see an issue.  Testing with the EVM would quickly rule out board layout issues.

Regards,

Jeff

OK, I will order one board. Thank you 

Hi Jeff.

Another question is if the resistors R14 and R15 are well calculated.

Also note that the point of the circuit called Vsystem will be connected to the rest of the board that will draw significant power from battery or DC power source. Vsystem was not connected during the initial tests.

To create this schematic I mixed two BQ24125 datasheet schematics. One from Figure 22 and other from Figure 24.

Thank you.

Hi Ubiratan,

Yes.  The feedback resistors give 8.42V as the regulation voltage.  

Regards,

Jeff

Hi Jeff, 

what about the temperatures some components reach. 

"The temperature in the input capacitor reached 48°C. In the inductor reached 64°C and in the BQ24125 reached 88°C. (all measured in the upper surface)"

Thank you.

Hi Ubiratan,

The temperatures seem reasonable.  You can estimate the IC temp rise using the RthetaJA value from the datasheet multiplied by the Ploss = Pin-Pout.  Ploss in the inductor is inductor current ~= charge current squared times inductor DCR.  The inductor datasheet might give a curve to predict losses.  Which input capacitor is getting that hot?  I would expect the high ESR polarized cap to get that hot if it was the only capacitor.  But with the ceramic capacitor in parallel, it should be handling most of the ripple current so I would not expect it or the polarized cap to get that hot.   

Regards,

Jeff

Hi Jeff.

I bought two EVAL boards and they charge my batteries correctly, as expected. I will start checking why in my board is not working the same way.

I based my design in a datasheet example (figure 24) that shows a circuit that does not need a power path to feed the system. But in my board some led drivers will sink about 5A from Vin or from the batteries. So we will need some additional componentes to implement a power path selection, right ?

What do you suggest for BQ24125 ?

Thanks.

Ubiratan

Ubiratan,

5Amps?!?  That's alot.  The charger will go into some sort of protection:

Regards,

Jeff

Hi Jeff. 

So an external power path is the solution, right ? Any suggestion ?

Thanks.

Ubiratan

Hi Ubiratan,

The simplest would be to connect your system directly to battery, so other side of the R13 current sense resistor.  The drawback of that is the charger is not able to terminate charging accurately because any system load is measured across the current sense resistor and misread as charge current.  If you had a way to short about R13 only when the LEDs needed their current, that might work. A PFET across it with gate used to control the PFET might work. That would force the charger into cycle by cycle current limit, outputting max possible current plus battery supplement to the LEDs.  The charger's cycle by cycle current limit is meant to be a short term protection though so it might get hot and then enter thermal regulation, which also reduces current.  Also if the charger sensing the voltage going below VSHORT, it will assume the battery is shorted are drop current to ISHORT.

Regards,

Jeff

Hi Jeff. 

I think it would be safer using some external FET to implement a power path. The PG pin could be used to turn them ON and OFF, right ?

Regards. 

Ubiratan

Hi Ubiratan,

I don't understand your system well enough to confirm that.  /PG is open drain and pulls low when the input power to the charger is good.  If that is when you need power, then it might work. 

Regards,

Jeff