Other Parts Discussed in Thread: MSP430F5359
Folks:
As I mentioned in the earlier "related" post, we build a board that contains a BQ27510-G3 Battery Gauge (accessed by an MSP430F5359 microprocessor). As I understand it, Battery Gauging is primarily based on Coulomb Counting and that, in turn, is based on reading, once-per-second every second, the current being drawn through the external shunt resistor. (The differential voltage across that shunt resistor is measured once per second every second). In addition, I believe that the specified window time for the measurement is 125 ms (but I don't know exactly what that parameter means to you, e.g., "integration time", "conversion time", etc.).
During routine operation of our module, our current consumption profile is challenging for the Battery Gauge to evaluate because we don't draw anything like a steady amount of current from the battery. Instead, we have the following sort of current consumption:
- Three days of drawing 300 µA most of the time,
- Intermittent operation at about 10mA draw for a period of microseconds to a second or so, and
- Intermittent, high-frequency operation (via a ~24 KHz PWM) at up to 750 mA draw for periods of milliseconds to a second or so.
Our experience with the Battery Gauge is that it doesn't provide us with a very accurate estimation of the charge remaining in the battery and we suspect that this is because the current measurement by the Battery Gauge don't correctly reflect the influence of our high-current pulses, either because the waveshape of the current pulses confuses the gauge or the once-per-second Coulomb Counter measurement doesn't adequately describe the actual integral of the consumed current. Routinely, we see the Battery Gauge running down to a claimed 0% State-of-Charge while the battery continues to successfully support the device for an additional day or more. (That is, we think the Battery Gauge is giving us quite-pessimistic State-of-Charge values and we alert our users far too soon about an impending dead battery.)
Note:
- We do think that we're giving the Battery Gauge accurate programming data as to the chemistry and capacity (etc.) of the battery. We're going to stage an experiment where we data log at high frequency (using external DMMs) the Battery Voltage and Current using batteries of tested (known) capacity so that we can confirm or deny the values reported by the Battery Gauge.
Do you have any suggestions as to how we could make the Battery Gauge better-accommodate our current consumption? For example, can the sampling rate of the Coulomb Counter be changed?
Atlant
