LiFePO4 replacement for Ride On Mower Lead Acid Battery - Update

I fitted the silicon diode in parallel with the mosfets and ran into trouble. Once the mosfets turned off the alternator had trouble regulating with the reduced load. The alternator voltage rose and the mower shut down. I found that a resistor to ground before the diode eliminated the issue. With some of the proof of concept stuff behind me I'm going to build version 1. It will have the following characteristics:

1. A number of parallel mosfets which when turned on allow the mower to be started and the alternator to directly charge the batteries.
2. Once the battery voltage rises above the first set point the resistor bank will be switched across the alternator to help stabilize the alternator output
3. As the battery voltage rises to the second set point the mosfets will be turned off and the battery will continue to be charge through a series diode with perhaps 0.6V of loss.

At this point the maximum charging voltage should be around 13.4volts. Combined with a charge balancer it should ensure no damage is ever done by overcharging the LiFePO4 cells. The state of charge should approach 90% at most.

I'm leaning towards using a micro controller instead of all analogue circuitry because it offers flexibility to alter parameters or add additional control mechanisms should I need them.

Results with Flat AA and AAA cells - Further Update.

Finally, the exhausted AA cell stopped working after 120 days. It would have run longer but I changed the on time to 10ms to accelerate it's demise back on day 51. The trials with exhausted cells has led to the following conclusions:

1. The larger the battery physical size the longer it can run below 0.8volts. 
2. The "on" time is critical with regard to energy consumption but is also important in relation to the electrolyte migrating/refreshing between flashes.

More support for the estimated 10 years life from a fresh AA cell.

10W Class AB with parallel transistors - Follow-up to Initial SOT223 Package Results

Eventually I found time to measure the input impedance of the amplifier. With 3 turns on the input transformer the input impedance measured 340 ohms on 80m. This is the right most curve of the three shown below - the labeling of the markers might cause confusion. With 2 turns it dropped to 159 ohms and with just a single turn it was 37 ohms. 

Input impedance to 10W "JBOT" Amplifier

 

Gain for the 10W JBOT stage has risen to 20dB with the 1 turn winding. However, I'm not completely satisfied with the "shark fin" distortion that is present on the output. Perhaps I am too picky since I have never previously looked at the output of an amplifier before any low pass filtering. 

Next steps:

Add a low pass filter and measure the level of the harmonics.

If the harmonics are acceptable then I will have met my goal of a robust 10W amplifier.