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.

BQ76920: bq769x0 Device Reset

Part Number: BQ76920
Other Parts Discussed in Thread: BQ76930, BQ76940
I note, as others have in the past, the absence of a RESET pin on the device while it is quite complex with state machines and a sequencer. This matter hasn't had a straight answer, so I am going to be quite direct:
Does the possibility of the part “locking up” actually exist, or has it got internal countermeasures to prevent this? Do we need to worry about it?
 
If it fell back into SHIP MODE, it would be fine as the MCU can reboot it. If the I2C bus locked up, it is normally possible to recover by clocking it 9 times etc, but if the part became completely unresponsive, i.e. no more I2C communications and no result from a BOOT pulse, then it would need to be able to be reset by a power cycle. I have a high-reliability application and I would like to get an idea of whether measures beyond the application schematics need to be considered.
Thanks and best regards,
Eric
  • Hi Eric,
    The bq769x0 family does not have a reset pin. Reset is defined with the voltage levels on the BAT inputs, for the bq76920 it is the BAT pin.
    The part is designed to work properly with well defined inputs. The I2C SDA stuck low is a standard protocol technique in any I2C system and not specific to the design. The bq76920 is not designed to lock up, but unexpected faults can occur in electronics systems, the bq76920 is not designed to detect and recover from unexpected system faults. Notice there are no self test or redundancy descriptions in the data sheet and no reset pin.
  • Yes. I haven't had the part ever locking up, but I have seen it falling back into ship mode under adverse and abnormal conditions during development. The MCU was able to wake it up again.

    The interesting part of your answer is about the voltage on the BAT pin when read in conjunction with this older thread. According to the data sheet, the BAT pin feeds an internal 3.3V regulator tied to an external capacitor on CAP1. The purpose of this internal regulator is not stated, but briefly pulling CAP1 to ground as alluded has been used to force a reset. There would appear to be issues doing this with the bq76930/40 versions due to their cascaded internal architecture, but is it a valid option with the bq76920? Should the BAT pin be pulled down instead and can it be done safely?

    The data sheet also indicates that REGSRC must be powered even if REGOUT is not used as it supplies power to the core logic. Can you elaborate on what is being powered from BAT vs REGSRC? Until now I guessed that the level shifter were powered from BAT and the core logic from REGSRC. This wouldn't preclude the internal reset from being generated from BAT.

    For the sake of completeness, powering off REGSRC to achieve a reset produced unexpected results as described here and doesn't appear to be a pathway at all.

    Thanks

  • Hi Eric,
    I was mistaken, disconnecting BAT does not allow it to fall to reset the part, there are internal leakage paths. As you indicate it may not be safe to pull BAT down to reset it. The diodes used on the bq76940 and bq76930 keep the BAT pin near the VC5 level, a limit is not published. While the EVM shows the REGSRC and BAT sharing a filter, and some movement of the net is expected with load, there will be some common mode effect where the input will not measure correctly with low BAT and the user would limit operation. I would not recommend pulling down BAT to VSS.
    In the data sheet block diagram 7.2, note that CAP is connected to an internal 3.3V regulator. This operates most of the IC and does not have specification for other use. REGSRC is used for the FET drive and REGOUT. REGOUT is used for the current measurement circuitry and the communication interface. Note in the section 6.5 Supply Currents the substantial current from REGSRC for the CC.
    Momentarily shorting CAP1 will force the internal 3.3V supply and circuitry to be restarted, but note in the block diagram it does not force a reset of the REGSRC/REGOUT circuit. It will determine that it does not have inputs and shut down, but it is not a well defined process.
    Removing REGSRC will cause the supported regulators and circuitry to turn off and the internal (CAP1) circuitry will re-start the REGSRC circuitry when REGSRC returns, but it is not a reset of the device.
  • Many thanks for your very informative answer. I want to mention that I once accidentally destroyed a bq76920 when all the inputs from the battery were transposed in spite of all the protections of the application schematic being present, so BAT and REGSRC were grounded and VC0 had battery voltage while VSS was still grounded correctly as the board supply didn't come from the cell connections. I have since been thinking about adding diodes to always feed the highest of all cell voltages into BAT to try ruling out this damage scenario, but it would involve making a new board or ruining an existing one and then testing it and risk destruction again...

    Back to resetting the part, CAP1 has no voltage until the IC is booted. An experiment I conducted showed that pulling down CAP1 causes a steady few mA to flow out, so the 3.3V LDO remains active during the short, but it does shut down immediately afterwards. This suggests that the part indeed restarts as the 3.3V supply returns, but then enters SHIP MODE. REGOUT is also turned off in the process and the experiment shows that a boot pulse is then able to restart it.

    Would it be correct to deduct that, while the process does not technically begin with a full internal reset, it causes a reset of the core logic followed by a shutdown of both the internal 3.3V LDO and REGOUT and therefore indirectly resets the part this way?

  • Hi Eric,
    VC0 has a 3.6V abs max. Keeping BAT as the highest voltage is good for the supply but may not be sufficient to prevent damage.

    You are correct, with the supply gone the logic in the part must re-start. It is not a well defined operation, it is a side effect rather than a feature.
  • I had VC0 protected by a 3V Zener to ground, so I didn't expect damage from there.

    Thanks for the great discussion about the reset.