• oophone will either run from battery or from USB power.
  • When on USB power, 5V have to be regulated down to below 4.2V to not blow the battery.
  • Our battery: a single-cell Li-ion accumulator
    • The nominal voltage is 3.7V
    • The charging voltage goes as high as 4.2V
    • The battery needs to be cut-off (i.e. considered empty) at a specific low voltage. The battery voltage continuously drops as power is drawn, and if it goes below the cut-off that we determine, oophone needs to switch off in order to preserve battery life. see here for reasons why we will probably want to stay above 3.0 V.
    • To be able to cut off, we need an ADC connected to the VBAT power line. The SIM900 has an integrated ADC we can query via AT commands; the AT32UC3C has something like 16 separate ADCs, so we should be fine. To measure it, the VBAT needs to be divided first, as the ADC is itself powered from VBAT and can only accurately measure lower voltages. A a simple two-resistor (high impedance, e.g. 100k) voltage divider will do.
  • The µC must also get its desired voltage
    • The µC should be able to operate on the entire range of battery and charging voltages.
    • it cannot run on more than the battery voltage (3.0 to 4.2V), so we will not use the "5V" supply mode.
    • It may be necessary to regulate the battery voltage down to below 3.6V, as the AT32UC3C specifies a maximum of 3.6V in the "3.3V" supply mode.
  • The USB bus also has power requirements, see issue #212.
  • When on live network, the SIM900 draws short bursts of 2A of electrons a few ms apart. Neither the battery nor the USB can provide this amount of current. That's why a fucking huge tantalum capacitor is needed in parallel with the battery. No one ever talks about milliFarad let alone Farad, The unit everyone uses is µF or less. The one I've picked is 1500µF. So you see, they don't say 1.5mF, no: 1500µF, that's how burstingly huge it is. My calculations were super pessimistic, so half the size would probably be enough. But more billions of pF can only be better here AFAIK. (in-between the bursts of 2A, the cap recharges at the rate the battery can provide, and during the bursts, the cap provides enough electrons to keep the 2A flowing. The 2A pulses averaged out over time amount to about 250mA; USB can provide 500mA, the battery about 1 A I think, so both have enough electrons over time, just can't provide 2A at once.)
  • See here why Li-ion batteries are so successful.
  • The battery should be charged in a Li-ion way. See image below.

LiIon battery charging chart
(source: http://batteryuniversity.com/learn/article/charging_lithium_ion_batteries )

hard limits

(These charts are drawn by git:oophone/math/battery_juice.py)


  • the dongle will only run on 5V USB power.
  • It still needs the huge cap to provide for 2A bursts.
  • Basically nothing else is needed, apart from regulating the VBUS (USB bus voltage of 5V) down to the SIM module's voltage range. I guess 3.6V, the nominal Li-ion cell voltage, is a good choice.

flat_in_days.svg (82 kB) Neels Hofmeyr, 11/17/2013 02:37 am

flat_in_hours.svg (90.5 kB) Neels Hofmeyr, 11/17/2013 02:39 am

flat_in_days.png (92.1 kB) Neels Hofmeyr, 11/17/2013 02:39 am

flat_in_hours.png (103.3 kB) Neels Hofmeyr, 11/17/2013 02:40 am

ion1.jpg - LiIon battery charging chart (48.5 kB) Neels Hofmeyr, 07/09/2014 09:11 pm