Several comments:
- If you don't need isolation, then (obviously) you don't need any device that includes isolation, such as a SSR. If you do need isolation, you probably don't need it for each channel. A single isolation gap, for all 16 channels, should be enough. All this could save you money.
- Snubbers don't protect the switches (the SSRs, in your case). They just reduce the probability of false triggering. False triggering is not a harm for your switches (they are there to be triggered, even continuosly). The false firings are an inconvenience (or an obstacle), if your application is such that the load should never be powered when you don't want it to (e.g., you have an electric saw, there's been an accident, and you need to switch it off right away).
- Since the 24 V are AC, if you use unidirectional switches (such as MOSFETs, or BJTs), you will need two switches per channel.
EDIT: a MOSFET is unidirectional, because it conducts in both directions, but it blocks in only one direction. For instance, a normal silicon NMOSFET cannot block current from S to D, due to the parasitic diode it has. Since that diode is there, if you want to use MOSFETs for AC, you CAN, but you need to put two in anti-series (with their sources tied together, and their gates tied together), or otherwise you won't be able to block in one of the two directions. GaAs MOSFETs don't have that parasitic diode, so one device would be enough, for AC.
- I would go for a cheap TRIAC per channel (probably, without any snubber, because 24 VAC is such a low voltage, that you probably won't hit any dV/dt limit).
- A cheap TRIAC like this one would work.
FET Type: I'm not sure what the difference is between N and P channel
The internal construction of a mosfet is different and you need different voltage levels to switch it on. Higher than source for N channel and lower than source for P channel. As you will be switching 25V load from a 5V microcontroller, choose an N channel logic level mosfet.
Drain to Source Voltate (Vdss): I'm assuming this is the max voltage it can handle going through it, so I should be finding a MOSFET that will support 25 V+?
It's the maximum voltage whitch the mosfet can withstand without letting the current to run through it.
By the rule of thumb you should double the rating to get a reliably working system. So, look for a mosfet with Vds in the range of 50V-60V. It would be OK to use a 25V mosfet but you usually don't want to operate near maximum limited values.
Current - Continuous Drain (Id): Assuming this is the max amperage going through it, so looking for one with 12.5 A+
Again - double it.
Vgs(th) (Max): I think this has something to do with the activation voltage applied to the gate that will make it activate, so I need one with less than 5 V?
Yes, mosfet dissipates least power when it's either fully on or off. Look at the graphs in the datasheet that specify Rdson depending on Vg - you want Rdson as small as possible, so you want to drive the gate above the Vgth. But note, that there is a maximum value that can be safely applied to a gate - Vgsmax. You should be safe driving it with a microcontroller, just a point to note.
Power - Max: Assuming this is the max power it can handle. I've calculated the power the solenoid would need as P = V*I = 25 V * 12.5 A = 312.5 W, so I need a MOSFET that can handle more than 312.5 W?
No, power dissipated by a mosfet would be I*I*Rdson - that's why you want as little Rdson as possible.
I don't know what Rds On (Max), Gate Charge (Qg), or Input Capacitance (Ciss) mean. Are they important for my uses?
When a mosfet is on, it's not an ideal conductor with no resistance. Rdson is the resistance of the mosfet and is dependent on different factors, datasheets usually give graphs how Rdson changes with different parameters.
You don't have to deal with gate charge and input capacitance in you application as fast (submilisecond) switching is not required. A mosfet gate presents itself as a capacitor to a driving circuitry and as it takes time for a capacitor to charge, it takes time for a mosfet to turn on that's why in high speed applications special mosfet driver ics are used that force high currents into gate to charge this capacitance as quickly as possible.
You can find cheaper mosfets with lower Rdson, just use the parametric search on digikey. Pay attention to the graph that displays Rdson against Vgth - sometimes manufacturers claim 4V Vgth and 4mOhm Rdsn, but when you look at the graph you see, that at 4V it's 20mOhm and you need to get to 9V to get the advertised 4mOhm Rdson.
Best Answer
This isn't too hard to implement. I can see the box and controller getting out of sync, but if the controller hits the zone twice and it doesn't matter what order the two sub-zones come on in, then that seems fine.
You are right in that you need some kind of memory. Since the unit will have no power to it between uses, that memory needs to be non-volatile. A microcontroller with built in EEPROM would do fine. EEPROMs are only good for a finite number of writes, but that's 100s of 1000s at least so no issue there.
When the power to the switch box turns on, all it really does is run the micro. The micro then turns on one of N relays to route the power to one of the sub-zones. It also writes the new state to its EEPROM so that it will power the next sub-zone in sequence next time.
A tiny micro running at slow clock speed can easily handle this. The 5V current will be small so a linear regulator will do well enough and be simple. Get relays that can run directly from the full wave rectified AC so there are no power conversion issues. 24V AC after full wave bridge with filter cap should be around 30-32 Volts. "24V" DC relays would work but get a little warm. Genuine 30V relays may be harder to find, so you could get 24V relays and put a resistor in series with the coil. A reverse catch diode accross the coil and a NPN transistor with base resistor to the micro is all you need per sub-zone output.
Another thing to consider is that the micro needs to see one power up each time the main controller turns on the zone. This should be as simple as putting a little low pass filtering on the micro's reset input so that it doesn't start running until a 100 ms or so after power is applied. By that time glitches and switching transients should be over.
The main controller also needs to leave some off time between powering this zone so that it toggles to the next sub-zone. It will take some time for the voltage to drop before the micro loses power or is shut down by the reset circuit. It could be a second or two depending on what values are chosen.
The more I think about it, the more I'm realizing the trickiest part of this is the reset circuit. You want to make sure the micro runs cleanly once per power up, and that it goes into reset cleanly once on power down and not too long after power down. This is all quite doable, but something that needs to be considered.