State of Charge Drivers + Calculation

[digitaldanny]

Describe the solution you'd like

• Capture Open Circuit Voltage of the battery using the voltage divider circuit.
  • During MSP432 initial power on, disable external hardware using the Pch power switch. This eliminates a majority of the power draw from the battery.
  • Enter low power mode. This will limit the MSP432's current drop from the battery.
  • Capture ~OCV of the battery using P5.2's ADC. Note that this input voltage is divided by 2 to handle the case where the battery voltage is greater than 3.3V (the MSP432 pins can only handle 3.3V inputs).
• Capture current draw per tick using in-amp circuit and P5.4's ADC.
• The state of charge at any given time can be calculated using the formula below. S(0) can be estimated using the OCV voltage and a LUT in the firmware matching the voltage level to a battery capacity in Amp-hours.
• (Optional) Create a "sprite" that shows battery state at 20-25% increments. Otherwise, use brackets and pound symbols to represent this -> [#####] = 100%

Additional context

• 42 - Prototype

• 26 - Research

OCV and Coulomb Counting circuits below:

Pch MOSFET Power Switch:

[digitaldanny]

It's important to add a current offset to the calculated current draw due to the MSP432 and TPS63020 loads. As shown below, the MSP_3V3 power rail has some additional current that does not go through the shunt resistor. This would affect the battery charge estimation.

The datasheets for both of these devices show the typical current draw for both of these devices. I just need to add these currents to the calculated current draw.

Current MSP432 (VCORE1) - 80 uA / MHz * 16 MHz = 1.28 mA Current TPS63020 - Negligible

MSP_3V3 Rail Current Draw

MSP432 Datasheet - Active Mode Current Draw

This means the current draw conversion in the software will be Vshunt / (0.1 * G_amp) + I_msp + I_tms

[digitaldanny]

This diagram clarifies why separate OCV readings are required. If a load is attached, there is current going through the battery's internal resistance. This means the voltage measured across the battery will have a variable amount of drop as the internal resistance changes. The OCV mode disconnects the load and has minimal current drawn by the low-power mode MSP432 to read the battery's actual voltage levels.

[digitaldanny]

Doing a continuity check on the 0.1 ohm resistor makes the buzzer go off. I think there might not be enough voltage drop across the resistor to detect with the 0.1 ohm resistor. I'll try replacing this with a 1 ohm resistor today.

[digitaldanny]

Going to estimate the power consumption based on the datasheet values. If I have extra time, I will look into the in-amp issues and try to get the dynamic SOC estimation working with Coulomb Counting.

[digitaldanny]

The main current variation I will see is from the HC-05 and the flex sensor positions. For the static SOC estimation, I need to find the average current draws of these devices.

This Arduino forum says that the HC-05 takes about 34 mA running and <1mA while sleeping. From my uC testing, the main glove loop iterated 10-12 times before receiving an ACK from the slave device. I am going to guess from this testing that the MSP432 can process the data to be sent much quicker than the HC-05 can send / receive over a baud rate of 9600 bps, so the HC-05 is likely in active mode most of the time. I will use the value of 34 mA for this device.

Based on my flex sensor measurements, the resistance range is 25 - 100 kOhms. This range should be added to the 22 kOhm resistors in series to determine the current consumption for 1 flex sensor. The current being fed to the microcontroller for the ADC reading is negligible, so I will 0 this out.

• Min resistance range (max current draw): 22k + 25k + 22k = 69k, i_max = 3.3/69k = 47.8 uA.
• Max resistance range (min current draw): 22k + 100k + 22k = 144k, i_min = 3.3/144k = 22.9 uA.
• 3 flex sensors means that there is an i_max = 143 uA and an i_min = 68.8 uA.
• I will average these values for the flex sensor static current draw value = 90.9 uA.

[digitaldanny]

Copying over the power consumption estimations from the #30 issue. Also adding the new flex sensor estimations.

• MSP432
  • Active mode = 80uA/MHz. I believe this means that there will be 80*16uA = 1.28 mA.
  • There are a few low power modes that offer between 25 - 660 nA for all clock speeds. If I use one of these low power modes, I might be able to drop down the current draw average.
• 64x48 OLED LCD
  • Typical 15 mA
  • High 30 mA
• HC-05 Serial to Bluetooth Module
  • Active 35 mA
  • Sleeping <1mA
  • Typical ~= 35 mA.
• Flex sensor circuitry
  • Flex sensor voltage divider: Average values calculated previously (90.9 uA).
  • CD4051BE Analog Demux: Input current is negligible for each input no matter the voltage level.
  • LM324 Amplifier: Max output current around 40 mA, but the actual current draw is minimal since the ADC pin from the MSP432 does not pull much current.
• MPU-9250 Accel/Gyro/Mag
  • Gyro - Active 3.2 mA, sleeping 8 uA
  • Accelerometer - Active 450 uA, sleeping 8 uA
  • Overall ~3.3 mA average with all 6 axes and DMP disabled
• TPS63020 Buck/Boost Converter
  • Quiescent current = 25 - 30 uA
  • 100 mA efficiency ~= 93% will act as a multiplier for all components powered by this IC

[digitaldanny]

Static current consumption = (Mpu6050 + Flex sensor divider + hc05 + oled + msp432) / Buck-Boost Efficiency = (3.3 mA + 90.9 uA + 35 mA + 15 mA + 1.28 mA) / 0.933 = 58.49 mA.

[digitaldanny]

Typical relationship between LiPo SOC and OCV. Notice that it is nearly linear at the beginning and becomes exponential towards the bottom 40-50% of the battery life (3 - 3.4/3.5 V).