the unused cell input pins of bq77908

To protect the battery the '908A must be able to monitor the individual cells.

The IC can be operated with a resistor divider like the EVM, but this will not protect the cells and will drain the battery.

Connecting the inputs to BAT+ through pull up resistors will cause the part to protect from OV and open cell tests.  Exceeding the abs max through the pull resistors may damage the IC.

Thank you very much for your reply.

According to your reply, we might do the following two steps:

1)  use two resistors (R>1M ) to divide the voltage between BAT+ and BAT-  to generate a less than 4.2V moitoring voltage for all cell input pins (all cell input pins CV1-CV8 will be tied together).

2) disable the cell balance functionality by setting EEPROM CB_EN=0.

Is the above solution totally OK and safe for bq77908 and battery, except costing tiny current (several uA) ?

The bq77908A must be able to monitor each individual cell to work effectively protecting the battery.

If all VC1 to VC8 inputs are tied together the part will not work. All configured cells must be in working voltage with a suitably low resistance to the individual cell voltages.  This can be done with a low resistance divider as with the EVM in simple evalution, but this would not keep the battery safe.   For example if the 8 cell battery is charged to 32V with a resistor simulator, the protector does not know if the cell voltages are all 4.00V.  It could be that the battery had 7 cells at 3.9V and 1 cell at 4.7V.  It would not be safe.

I am working on a design where the BQ34Z300 and BQ77908A are used in conjunction and we want to minimize the power consumption. We have 2 problems :

1-In some cases, we will not have access to the individual cells voltage of the battery. We will only have access to VBAT+ and VBAT-. So we wonder if there is a way to AVOID cells simulation with resistors (to minimize consumption), perhaps for example by fully disabling the cell balance functionality (set EEPROM CB_EN=0). Will we still have the BQ77908A properly working and detecting all other faults not directly linked to cells balancing ? What shall we do with all cells input of the IC (left unconnected ? high value resistors dividers ? others....) ?

2-We have choosen high side FET switches configuration because the use of isolator on the I2C bus for BQ34Z100 would have involved  too much energy consumption (I2C isolator are not really ultra low power). Consequently, we need to understand how to drive DPCKN signal in order to have proper load detection. In the forum it is suggested to use some transistor or comparator circuit to sense the pack voltage and drive DPCKN. Do you have any application note on this or can you provide a schematic ?

Thank you

The bq77908A protector provides several features including cell based voltage protection.  If you can't connect to cells, the device can't provide that protection.  It does have other features such as current protection and FET drive which could still be useful over a discrete solution.

1. Cell balancing can be disabled by setting the mode to 0, however the open cell (open wire) detection can not be disabled, so all used cell inputs (4 cells minimum) must be provided with a nominal voltage with a moderate impedance to prevent the open cell test from failing.

2. The easiest solution to providing over current recovery is with the CHGST signal.  Using DPCKN would require a circuit, we do not presently have an apnote or recommended circuit since requirements can vary significantly. Pulling up the pin until the desired recovery condition is recognized after the fault would work.  Certainly this may begin to look like a sequential circuit.  If your host is connected by I2C and remains powered, you could recover from the I2C lines or an IO expander.

Thank you for your answers. I would like to get more details.

Point 1

You are talking about moderate impedance. What do you mean ?

With the evaluation board of the BQ77908A, you have a battery cell simulation board with 200 ohms (1) series resistors between Vbat+ and Vbat- . On the evaluation board itself, you also have 47 Ohm (2) resistors at each of the BQ77908A IC cell input plus some 1uF (3) capacitors between cell inputs.

For the voltage cell simulation, a 200 Ohm (1)  value for the resistors of the divider will drain a lot of current from the battery. Could we use a much larger value ? What would actually be the max value knowing that we just want to prevent open cell test from failing.

The datasheet recommends a max of 1k Ohm in place of the 47 Ohm (1) which is the minimum recommended. Is the 1K value the max possible value or could we still increase it ? What would be the max value ? 

 And what for the capacitor (3). What woulf be the best value in our case ?

Point 2

How do we provide over current recovery with CHGST ?

Using a pull up on DPCKN does not seem suitable because the load may only powered by the battery. How can we detect the disconnection of the load. ?

Hi Paul,

Point 1. 

Realize that you are trying to do something the part was not intended for and in a manner where it can not keep the battery safe.  If you are using an average cell voltage to determine over voltage or under voltage, the part can't tell what the individual cell voltage is and may not protect properly.  One cell could be much lower voltage than the others or much higher than the others.

We have used 1k divider resistors to operate the part, but this is still a lot of load on a battery.  The electrical characteristics table shows 100k as the max impedance, but really it depends on the voltage.  A voltage below 1 V during the open cell test interval will cause it to fail.  "To much" impedance can cause effects on other protections as described in the Open Cell section pages 24-26 of the datasheet.   Having a high impedance voltage divider followed by buffers for the individual inputs would conceptually provide a low current drain with low to modest impedance for the cell inputs. At some point you have to evaluate the amount of circuitry to bypass the unwanted protector functions and compare it to the circuits needed for another implementation of the desired functions.

The input capacitors are there to filter the inputs and keep the inputs away from abs max in transients.  As the input resistors become larger, the capacitors can be smaller.  The 1uF capacitors should be OK, but smaller may work also. As you move to larger input resistances, you are less likely to need clamp diodes on the inputs for transient protection.

Point 2.

To do over-current recovery with CHGST, ground DPCKN and set SOR =1 in the device configuration.  After the current protection fault, it will not recover until signaled that it is in the charger by CHGST high.  You must have a system input for the CHGST for this to work.  This setting does affect other recovery, see Table 2 of the datasheet page 20 and bit descriptions.

With the basic circuit concepts for the part such as figures 13 through 15 of the datasheet, after the discharge FET opens with a load on the pack, the PACK terminals pull together and DPCKN pulls up indicating load.  When the load is removed and the load is high impedance, the FET leakage, the internal R_{DSG_GND} and external R_{LDRM_DET} (when used) all pull down on PACK- and DPCKN to bring the pack voltage close to normal.  When it is within the load recovery threshold V_{OPEN_LOAD}, the part considers the load removed. This simple behavior lacks complex recovery decisions.  If the system where the battery is used does not fully release the load, the charger may need to be attached to bring the pack voltage back to normal and allow recovery. If the load is highly capacitive and the PACK voltage does not change quickly after the overcurrent, the part may recover quickly and cycle.  Using a definite signal from the charger (or system) that it is appropriate to recover could be better. 

With a high side switch, a circuit is needed to control DPCKN if load detect is desired for the system.  The simplest would be to consider load removed when the PACK+ is greater than some fixed voltage, or some small threshold near PACK+, such as PACK+ - 2V.  The fault cycling with capacitive load could be avoided with SOR and a signal on CHGST.  A more complex control of DPCKN could also be used without the CHGST, but it begins to need a sequential circuit.  Some behavior such as DPCKN is low while PACK+ is rising above ~ 30% of battery and DSG is low, otherwise DPCKN is high.

With high side switches, if you have signals from the system, control of DPCKN and CHGST should be straight forward.

Thank you for your answers. It is much cleared now.

However, I do not understand you latest paragraph about detecting load removal in a high side switch configuration.If the load is removed there is no PACK+ voltage as the reference is PACK- (and VBAT- through the sense resistor).

Right.  If the battery provides all power for the system it is important to keep the battery from protecting in normal operation so that the battery stays on.  The information from the gauge is important to know when to stop discharge, or if the battery is too hot.  Overcurrent or short circuit may be a serious malfunction. If it does protect, the FETs turn off and the PACK voltage collapses.  Whether high side or low side, they come together.  As you pointed out, it is easier for communication with the gauge if it is the high side.

When the part protects with the high side, PACK+ comes down to PACK-.  If the battery is removed from the system and the load is gone, leakage of the power FETs will tend to pull PACK+ up very weakly.  If you add a load detection circuit, it will tend to pull PACK+ either up or down depending on how it is constructed.  If it pulls it down more than the FET leakage, it won't recover.  If it pulls it up too hard, it becomes an unacceptable leakage path.  If the battery remains in the system after the protection and the system has any residual load, it is likely that overcomes the FET leakage keeping PACK+ low and a recovery circuit would not be useful.  In this case the system likely needs connection of the charger for recovery, either through a signal line or by the charger pushing the PACK voltage back to a normal level.

Another way to recover is to put a low power microcontroller or timing circuit in the battery. As a stand alone protector, the bq77908A does not provide a fault status, but if the microcontroller sees that DSG is off it can attempt to recover after some period of time for example, and could adapt the re-try interval if not successful.  This adds a lot of complexity to what started as a simple protector.