Adding power path to TP4056 solar battery charger and limit output voltage to 5.5 V or below

Question

I have a 6 V solar panel and a bunch of TP4056 modules with the battery protection circuits lying around and I thought I'd use them to make a solar powered BLE circuit with the nRF52840 dongle for some future projects as also have a couple of those lying around.

The project was inspired by a solar charging circuit from one of Andreas Speiss's videos (https://www.youtube.com/watch?v=37kGva3NW8w) where he added a "power path" so that the load (in my case the nRF52840) runs directly on solar when there is sunlight instead of discharging the battery. However, I don't quite understand how the circuit works and I was hoping that someone here could provide an explanation.

If I have understood correctly P-channel MOSFET will only conduct when the voltage difference between the gate and the source is negative. So when there is full sun the solar panel will output 6 V and the source voltage will then be 6 V minus the drop across the Schottky diode. Vgate - Vsource will be positive and the P-channel FET will act as an open circuit.

However, I don't understand how the circuit works when there is no sun, i.e. Vgate = 0 V. Won't the source voltage also be 0 in that case, i.e. Vgate - Vsource = 0 V and still act as an open circuit? The source voltage would have to be 2 V in order to reach the gate threshold voltage. Will Vsource be the same as the battery voltage? If yes, can someone explain why? I am probably missing something fundamental here so please bear with me.

There is also a second issue that I would like to get some advice on. The nRF52840 has an internal regulator that has a minimum input voltage of 2.5 V and max 5.5 V. Now the 6 V solar panel will actually output ~6.25-6.3 V in full sun (based on measurements) so I was wondering what the best way to limit the voltage to max 5.5 V, while still being able to power the nRF52840 with solar down to 2.5 V.

I came a cross this What is the best way to limit voltage of a solar panel? question, which initially suggests using a shunt voltage clamp. Going through the comments it would seem that a LDO wont have to dissipate as much energy as the shunt voltage clamp. So do you think its better to go with a LDO with a minimum input voltage of 2.5 V, max 6.5 V and a 5 V output voltage e.g. the MAX8867 (https://eu.mouser.com/datasheet/2/256/MAX8868-1389206.pdf)

I made a quick schematic in Kicad, but the circuit/component to limit the voltage to 5.5 V or below is missing as you can see.

Answer 1

When there is no sun the gate of the FET will be at zero volts allowing power from the battery to drive the load. When there is sun the gate will be at the same voltage as the source disabling the battery connection, the diode will provide the power to the load.

The source will initially be at the battery voltage minus the voltage drop across the body-diode (~0.7V) so it will be ~3-4V depending upon the state of charge of the battery. With the gate biased like this the FET will be conducting and bring the source and load voltage to within a few millivolts of the battery voltage and supply the load from the battery.

A relatively high value resistor across the solar panel may be necessary to ensure the gate is at zero volts with no sun and get clean switching.

All discrete MOSFETs have an internal diode between drain and source because of the way they are constructed you must not forget about it when analyzing circuits using MOSFETs. Also, MOSFETs will conduct in the reverse direction with the drain positive relative to the source (in the case of P-channel devices) although the voltage will never get more than a volt or so even at high currents due to the body-diode. The gate still needs to be negative to the source for conduction to occur even if the drain is positive.

These characteristics are often exploited in battery switching and circuits such as this.

Answer 2

I built and tested the circuit for a while in the past. It works fine in standard cases - lets say the sun is shining (-> power path) or at night (-> battery). But... I viewed a strange behavior especially in the early morning and late evening - inexplicable brownout-errors. After a while of head scratching it becames clear to me. The problems appear in the "twilight-time" - then the current from the solar panel is not enough to power the load and also the MOS-FET is in a state where you not know, if it is open or closed or something inbetween (depends on the voltage of the solar panel ) which leads to less power than expected on the load side. This leads to i.e. brownout-errors in those times. This issue is not new and some guys worked on it (https://github.com/gbhug5a/Solar-Power-Load-Sharing/tree/main) In my case I gonna go the standard way without power path and going to test the op-amplifier way.