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.

bq2002C Problem

Other Parts Discussed in Thread: BQ2002C, BQ2002, BQ2002T, BQ2002F

I have designed a standby power supply that includes a bq2002C which keeps a 6-cell NiMH battery (i.e. 6 AA batteries, each rated 2000mAH) charged to maintain the system during any mains power failure.  When I tested a prototype all appeared well, so we built an initial production batch and soak tested them.  I was extremely concerned when the battery pack on one unit literally exploded after several weeks operation (with a constant mains supply)!  I have returned the batteries to the manufacturer for investigation and closely examined the charger.

Since then I have read up on NiMH battery technology, closely examined the bq200C datasheet and carried out extended tests on one of our units.

Our circuit comprises a switched mode supply that generates 12VDC, and is capable of sourcing at least 3A.  It also powers a 5V regulator that drives the bq2002C, and a switching regulator that is configured the act as a 1.4A current source.  The bq2002C enables this regulator as required. VM on the bq2002C is set to mid-rail.

First Problem:

The bq200C datasheet states that it can be used to charge both NiCd and NiMH batteries.  However, battery manufacturers specification recommend trickle charging NiMH batteries at around C/300. The bq2002C trickle charges at C/32, almost 10x the recommended value.  I am trying to speak to an engineer at Duracell to see if this cn have any detrimental long-term effects.

Second Problem:

The State Diagram shown as Figure 2 in the bq200C datasheet shows that there are three main states, fast charge mode (LED on), charge pending mode (LED flashing) and trickle charge mode (LED off). The states can be summarised as follows:

  1. At the initial application of mains power the device checks the battery voltage if greater than 2V it enters trickle charge mode, otherwise:
  2. If the battery voltage is less than 0.84V it enters Charge Pending mode, otherwise:
  3. If VTS < VCC/2 it enters Charge Pending mode, otherwise:
  4. (This is not included in the State Diagram, but is describe in text).  If VBAT < VLBAT or VBAT > VMCV or VTS < VHTF(where VLBAT =  0.175 * VCC,  VMCV = 2V and VHTF = 0.6 * VCC) it enters Charge Pending mode,otherwise
  5. It enters Fast Charge mode and lights the LED. Fast charge terminates on -dV/dt, PVD or out-of-range battery volts, then:
  6. The device then enters trickle charge mode and the LED switches off.  The state diagram suggests that it will stay in this state until the battery voltage is less the 2V (which it should be) when it will revert back to state (2) above.
  7. In Charge Pending mode it will continually check VBAT and VTS. If VBAT > 0.84V and VBAT < 2V and VTS > VCC/2 it will revert state (5) above.

This suggests that it will continually loop between Fast Charge, Trickle Charge and Charge Pending.  Since it repeatedly fast charges the battery, even when the battery is fully charged, it will eventually destroy the battery.

My design has proved this to be the case.  At power-up the device enters fast charge mode (LED on).  I believe that it terminates on -dV/dt, but by then the battery temperature has risen to about 400C. The RT/NTC I am using give TS = 0.6 * VCC at 350C and TS = 0.5 * VCC at 470C. The device then enters trickle charge mode, and the LED flashes.  This indicates that it has indeed passed through the (trickle charge / LED off) mode back to checking for under voltage and excess temperature, as described above.  Since VTS > VHTF it enters charge pending mode, still trickle-charging the battery but flashing the LED.  When the battery eventually cools below 350C (this typically takes over an hour) it reverts to fast charge mode.

My application therefore proves that the State Diagram is correct, and experience has shown that repeatedly fast-charging the battery does eventually destroy it.

WHAT'S THE SOLUTION?  I THINK THE bq2002c IS NOT SUITABLE FOR A BACK-UP POWER SUPPLY, BUT NOTHING IN THE DATASHEET SUGGESTS THIS!

Corollary

I thought I might be able to overcome the problem by using a different variant of the bq2002.  The bq2002T is the only device that can trickle charge at around C/300.  However, it proved unsuitable on a number of grounds:

  • It terminates Fast Charge on rate of change of temperature or timeout.  If the battery is not completely flat when power is first applied it will fully charge before the timeout expires, so must terminate on rate of change of temperature.  However, because it carries out additional tests on temperature, relatively low value resistors are needed around the thermistor, reducing its sensitivity.  On fast charge the battery temperature rises, but too slowly for the rate of change sensor to detect.  When the battery reached 600C I decided to terminate the test.
  • Trickle charge mode pulses the current source on for only 286us.  The switch mode circuit I use takes about 150us to start, so all duty cycles are about half the specified values
  • First Problem:  TI suggests to follow the battery manufacturers recommendations..with that said....there are 100s of millions of cordless home phones that use NiMH batteries and use the C/10 slow charge method and as long as the phone is hung up the battery gets charged at C/10.......thus it is hard for me to believe that any catastrophic event happened due to c/32 trickle charge.  The only down side to the continuous C/10 charge is that the pack may last only 3 years instead of 6 years.

    Second Problem:  The flow you describe is basically correct, but one misconception.....once the charge terminates and enters trickle charge, the battery has to be removed for the output voltage to go above 2V and then when a battery is reinserted the voltage is pulled below 2V and a new charge cycle is implements, otherwise the charger should stay in trickle charge indefinitely. 

    The battery can be over charged if abnormal circumstances happen and then need extra circuitry to solve the issue.

    If the input source or battery is repeatedly unplugged and plugged in, say every 10 minutes, the battery is repeatly recharged.  If this is the enviorment, then I would suggest a external circuit comparitor, that monitors the pack voltage and will disable the charger IC if the pack voltage is greater than ~1/3V/cell.

  • Charles

    Thanks for your response.  I would like to make the following comments:

    First Problem:  Are you implying that products like cordless phones simply trickle charge the batteries?  If that is the case all that is needed is a series resistor to limit the current, and you are arguing against the need for any TI components!  All battery manufacturers advise against simply trickle charging, but I am trying to ascertain the reason.

    Second problem:  I assumed the symbol in the State Diagram for exiting Trickle Charge mode meant Vbat < 2V.  My tests confirmed this assumption in that the bq2002C transitioned only from LED on to LED flashing and never achieved the state of LED off to show it was in trickle charge mode. You imply that it means that Vbat has exceeded 2V then returned to a value below 2V.

    I have carried out yet more tests on my design. The battery remains connected at all times.  The bq2002C is powered from a 78L05 regulator with a 1u cap across the output. RB1 = 390K, RB2 = 100K with a parallel 10n cap. Rt = 3.6K with a 100K/10n RC filter between it an pin TS.  All tracks are less than 20mm long.

    I have monitored Vcc, TS, BAT and CC with an oscilloscope and can see no significant noise. I have a digital thermometer on the battery which reads around 420C, Vcc = 5.02V and TS = 2.97V (gradually rising as the battery cools).

    The bq2002C again cycles between the Fast Charge mode (LED on) and Charge Pending mode (LED flashing) and at no time enters Trickle Charge mode (LED off).

    HOWEVER.....

    I have tried fitting a 10u tantalum bead directly across the power pins of the bq2002C (pin 4 = Vss, pin 6 = Vcc).  Lo and behold, when Fast Charge terminates it goes into Trickle Charge.  This is repeatable on at least two power supply modules.  It appears therefore the power supply decoupling is critical.  A 1u capacitor 20mm away is not sufficient.

    I will try alternative values, and work out a way the fix can be implemented on the supplies we have already built.

  • An addendum to my previous post:

    Although the fix described above works, the battery is trickle charged at C/32, compared to the battery manufacturer's recommendation of C/300.  This keeps the battery temperature about 50C above ambient, which I suspect will shorten the battery life.

    I have changed the device used from bq2002C to bq2002F.  Note I have set TM = Vcc/2. The datasheet shows the significant differences to be:

    1. bq2002F adds a top-off charge of C/32 when fast charge ends. The top-off charge terminates after 160 minutes or when the maximum temperature is exceeded (Ts < 0.5 * Vcc).  This is in fact recommended by the battery manufacturers.
    2. Trickle charge rate for bq2002F is C/64 vs C/32 for bq2002C.  This is a little closer to the battery manufacturer's recommendation.
    3. The trickle charge pulse rate for bq2002F is 286us every 9.15ms vs 73ms every second for the bq2002C.

    However, I am using a constant current source designed around an LTC1771 switch-mode regulator.  I have found that there is a delay of about 150us after raising RUN before it starts, so the trickle charge on-time is reduced from 286us to around 150us.  This reduces the trickle charge rate to around C/120, which is even closer to the recommendations. It also reduces the top-off charge rate to about C/64 (which shouldn't have any undesirable impact), but doesn't change the fast charge rate.

    I believe this design is close to the optimum for NiMH batteries.

    Note, however, that is is still essential to have a 10u decoupling capacitor directly across the supply pins.

  • That is hard to believe, unless your cells are very small.  the temp rise is probably not linear, but if a C/32 rises 5C then a C charge would be 160C rise.

    Another way to look at it is C/32 = 44mA, this squared times the battery impedance which is probably in the 100mOhm region is 0.2mW.

    I would not think one could even measure a temp rise from the power dissipation.  The battery must be self heating from other circuitry, or there is a significant discharge current that is heating the cell, or the cell is damaged.