Working principle.
PSMN5R4-25YLD MOSFET, LTC6702 Comparators, 1N4733A 5.1V Zener, MCP1810 3V Regulator, 1N5819 Schottky.
Performance without MCP1810 and 1N5819. It will only get better with these two parts. Oscilloscope screenshots below:
https://imgur.com/KMR9tmJ
https://imgur.com/G0RLPNS
https://imgur.com/IBNHV89
https://imgur.com/PErN3Ic
Two 18650 lithium cells with internal resistance set to 100mOhm on the left. The capacitor in parallel with the 18650s is only for simulation convergence.
10nF capacitor at the output of the pot is in place to hold the voltage at this node during power-mosfet ON-OFF cycle.
2.2pF capacitor at the oscillator is in place because voltage spikes were observed during mosfet ON-OFF cycle in Spice simulation. Capacitance at this node removed the voltage spikes.
The switch has been moved to the output of the driving comparator to avoid DC (non PWM) signal after the button is released (opened).
The capacitor at the gate of the mosfet is in place to model gate capacitance of the mosfet.
The 330ohm resistor at the gate of the mosfet is in place to reduce the current drawn from the driving comparator. In spice the current drawn from the driving comparator is approximately 12mA. Without the 330ohm resistor current drawn from the comparator is approximately 80mA. This is far beyond the capability of the comparator.
Thanks to Hurz, this device has more driving power over 18650 lithium cell voltage range. On the far right a voltage doubler is in place configured as a bootstrap. This voltage doubler charges a 630nF capacitor seen near the 5.1V zener diode during mosfet ON-OFF cycle. This regulated voltage is then fed to the VCC supply of the driving comparator. This increases the available voltage envelope of the driving comparator despite the voltage of the 18650 cell.
This circuit is still a WIP. It is currently configured with a 555 timer voltage doubler in real life. It has been tested using the bootstrap voltage doubler but had performance problems at duty cycles < 50%. Note that these measurements were made using a multimeter because I do not have an oscilloscope. I suspect this issue is related to the resistors below the pot. I.e. being slightly too small causing PWM to malfunction on the lower end.
I included the bootstrap voltage doubler in this circuit because I plan on moving forward with this design. It is more elegant than the 555 version and I'm (perhaps foolishly) confident that I've identified the reason for the problem.
If anyone attempts to build this be warned the current running through this device is very high. It is a dangerous circuit if improperly used!
