This is a report of an experiment, a circuit board design using the BQ25504 , an energy harvesting chip. The chip is intended for tiny, free-standing devices that harvest power from their environment, so they can run indefinitely without new batteries.
It is programmable using resistor nets. Unfortunately, some of the programming is for the battery management functions:
- the undervoltage condtion, UV, when the chip should stop drawing from a battery
- the overvoltage condition, OV, when the chip should stop charging a battery
If you are not using a battery, but say a super cap, you might not care about UV and OV. But the chip’s pins that do the programming must still be connected.
That datasheet example shows three resistor nets, one resistor net for each of the above functions, and one resistor net for the battery OK condition. (And another resistor net for MPPT.)
My question is: can the three functions be programmed with one resistor net? The answer is yes.
How I did it:
- keep only the resistor net for OK programming (ROK3, ROK2, and ROK1)
- connect pin 6 (VBAT_OV) to pin 10 (OK_PROG)
- connect pin 8 (VBAT_UV) to pin 9 (OK_HYST)
On the evaluation board for the chip, with:
- ROK3 = 1.4Mohms
- ROK2 = 4.2Mohms
- ROK1 = 4.4Mohms
This yields programmed voltages of:
- UV = 2.2V
- OK_PROG = 2.4V
- OK_HYST = 2.8V
- OV = 3.1V
(Which meets the requirements for ordering, specified in the datasheet.)
To actually modify the evaluation board requires unsoldering three resistors, cutting a trace, and soldering in two jumpers.
In my test, it worked. ( But I didn’t actually test how the chip behaved in the OV conditions. I didn’t use an oscilloscope to verify the waveforms.)
The three resistor nets have seven resistors total. Four extra SMD resistors is no big deal, but if the intended use is in tiny devices, why not eliminate them? (If your application doesn’t need battery management.)