After using these things for few years now, I thought I should share some feedback about the power arrangement on WisBlock. I certainly appreciate the flexibility that the WisBlock products provide, and this feedback is offered in the hope of making them even better.
My “standard hardware” is a RAK19007 baseboard, RAK4631 processor, RAK12500 GNSS, RAK15002 SD carrier, and two RAK15005 FRAMs.
I think FRAM is the ideal form of small NV memory. The two main advantages over EEPROM is that you don’t have to prearrange writes in pages and there isn’t the 5ms polling wait after writes. I use the RAK15005 because the larger sizes use SPI rather than I2C and I’m using the SPI to stream data to the SD Card.
The 15005 in standby draws 15uA. It can be put into sleep mode to draw 4uA.
I use an active GNSS antenna with the RAK12500 GNSS module also powered by 3V3_S, and this typically draws something like 10mA. There is a resistor strapping option that would connect the LNA_EN pin directly to the antenna, but the enable is meant to control an external switch, and I doubt it is rated to handle a 10mA draw directly (but I admit I haven’t tried this). To power down the LNA I have to drop the 3V3_S and everything that hangs on it. But conversely this means that whenever I want to use the FRAM, I have to power the LNA on the GNSS antenna.
At the same time the SDcard is powered by unswitched 3V3 which idles at something like 10mA. There is no option to fully power down the SDcard at all.
I know that everyone has their own projects and their own use cases that are not the same as mine. But I also think that everyone is trying to solve the same basic problem of how to make their device run on less battery and the general principle is the same.
I suggest that future versions would be improved by making the default supply for any peripherals under (say) 1mA to use unswitched 3V3 and anything over 1mA to use 3V3_S. We can debate the ideal threshold, but clearly having the FRAM with uA standby current on 3V3_S and the SD carrier supply unswitched is the wrong way around.
The GNSS would be improved by including a switch controlled by LNA_EN, but the above changes would at least allow the whole GNSS to be powered down while still using the lower powered devices.
I also think the memory cards would be improved by having the most sig address bits strapped to the slot id somehow, but I don’t have a straightforward solution for this.
That’s why I said I don’t have a ready solution.