Ok thanks guys - I wasn’t aware of the distinction between SCR and MOSFET devices, but I know what to look for now.
I see I can buy a
50 pack of mosfets for €15 or so, maybe just putting some on a Veroboard with some extra wire to thicken the copper traces and suitable heatsink could yield a DIY mosfet switch cheaply enough? These FETs can be paralleled, right? so putting 50A through three 49A Mosfets in parallel would keep them all well within spec and cool enough with a heat sink for good measure? I suppose I can’t assume the current splits equally, as the devices may have variations in DC resistance, but adding enough in parallel and ensuring the total load current doesn’t exceed the rating of each device seems safe enough (with heatsinking in addition).
I need to research to see if such devices need any circuitry to bias the FETs to switch them on, or whether just putting some plus volts on the gate is enough. Can it be that simple?
And what about switching my PV panels directly to my LiFePO4 battery through such a switch, under the control of the BMS such that the switch disconnects when any individual cell reaches, say, 3.6V (or lower) while leaving my Renogy 20A MPPT SCC (which would be connected to a different string of panels) to finish the charge? That way I could add a further 50A of charging capacity to my system very cheaply, avoiding having to buy more SCCs.
Making an AC SSR would be more difficult. That would require four MOSFETs configured like a full-wave rectifier, high and low side switching, and synchronization of the gate with the power. Unlike a diode, MOSFET doesn't have a diode drop of 0.4 to 1.5V, just resistance, so loss can be lower.
For DC switching a single MOSFET can work. It can be either high-side (for positive rail) or low side.
Low side is a bit easier; n-channel MOSFET has lower resistance (think it is that electrons vs. holes thing - when you play musical chairs, do the people move or do the unoccupied chairs move?) You can get a MOSFET good for 100A, but gate can't be more than 20V above source. Resistance between source and drain varies with voltage gate to source, so that has to be high enough, typically 5V to 10V, to get source-drain resistance down around 0.001 ohm. If no current flow during switching, little power dissipation, but if current is flowing there can be considerable dissipation while voltage drop between source and drain is multiple volts. Low-side switching can be a problem if circuits on both sides use "ground" as a reference, for instance for RS-485 or other communication. The shifted voltage can burn out drivers.
High-side switching is most easily done with p-channel, driving gate no more than 20V below source. But p-channel uses holes for carries, is higher resistance. n-channel can be used, but that requires driving a voltage higher than positive supply. This can be done with a charge-pump (which is slow) or a boot circuit (which can't maintain DC, OK for PWM but not on continuously.) Drive ICs are available.
If current is flowing during switching, considerable power is dissipated in the MOSFET during the time (millisecond? Second?) it takes to drive the gate, and to change voltage of load. This can melt the MOSFET. If switching a capacitive load (input to an inverter with many electrolytic capacitors), that looks like driving a short. Must switch gate gradually, let the 1/2 C V^2 energy of capacity be dissipated in MOSFET over a period of time. In this case, if the load (inverter) starts drawing DC current before MOSFET has been driven on hard, it will burn up MOSFET. Either need a delay before load turns on, or a "power good" signal from the drive circuit to enable it.
MOSFETs can be paralleled. Because they have an "on" resistance which is linear, not an exponential curve like a diode or BJT, they share current fairly well.
MOSFETs may look like they would be good analog switches, like for linear regulation, but they can't perform to the specs shown on their data sheet. Power MOSFETS have been optimized for use in switching power supplies. The "Safe operating area" curves of current and voltage vs pulse duty ratio in the data sheets are incorrect, and portions of the MOSFET will heat up and go into thermal runaway. They need to be derated 10x or 100x from what data sheets say if used for analog.
I list all these issues because I've seen them in PCB designs with MOSFETs and power-good/inrush circuits. The analog and safe operating area issue is written up in NASA papers.