- Continued simplification of power supply - now primary switcher followed by LDO
- Inclusion of night-time 'dark' blocking FET
- Slight adjustments to FET driving traces, adding back in debug components.
I moved to a 13v switcher followed by a 5v LDO vs. the dual switcher in the last design. Primary this gives a more stable voltage source into the ATmegaxxM1 for its ADCs - as well as providing a stronger VDD for the fet boost source. In looking at the extra loss associated with the LDO, when fully active I calculate 175mW loss vs. perhaps 10 for a switching PS. However this reduces down to the 5mW range when the uC is 'idle' or asleep - likely the in the same area as switcher (or even a little lower). Given the uC can likely be placed into sleep for a large portion of time (most the peripherals work independently), am looking for low final loss.
Debugging has progresses enough that I thought it was time to add in a night-time blocking diode (using a FET for less loss). The drive for this is pulled off of the boost ckt for bank-A, anytime that is working the blocking diode FET will get enabled. It uses the intrinsic capacitance of the FET to keep enabled between pulses, while the 1M resistor provides a turn off after a period of inactivity.
I increased the trace widths of the FET drivers, and added in DC blocking capacitors, as well as 10K resistors between the drivers and the uC. Idea hear is to provide SOME level of protection to the driver and uC while debugging code. The final version of the PCB will have these removed. And I likely will go to a 4-layer PCB for any final PCB as well.