Neutral wire and high frequency arc compartment

I think you are getting earth loop problems a bit mixed, but as far as running the neutral correctly is concerned you should bring it to a central block and make all other connections direct to that point in a star configuration. This is also correct for earth and fase actually.

NOT MUCH :-)

Do you mean there is no transformer to the uC? i.e. you are just using a voltage dropper?

Yes.

This could be made to work if the single return wire is routed with the supply conductors daisy-chain to all 3 loads so that all 3 supply conductors are tightly bundled with the return along their full length, and attention is paid to details at the controller entry, but I don't recommend it, it will let in more noise than the 3 twisted pairs if all other factors are equal.

If your controller is using relays for switching power to the water and gas solenoids and the HF relay, then no part of these circuits is particulary sensitive to HF interference and the issue is keeping the noisy lines from coupling HF energy into the box with your controller. Using twisted pair wiring from the control relays to the load solenoids and relay will help reduce differential noise pickup by largely cancelling capacitive coupling to the supply and return wires (which alternate being closer to the noise source) and minimizes magnetic field coupling by keeping the loop area between the supply and return conductors small. Common mode noise (noise voltage the same direction in both conductors, think of the pair of wires acting together like a single antenna) will not be reduced by twisted pair wiring however and you should route the wiring as far from the noise source as you can, keeping it close to the metal walls of the enclosure (the capacitive coupling to the case provides a leak for HF common mode from wiring to ground), and keep the length of these wires inside your controller enclosure to a minimum, while still routing close to the metal walls, (which also minimizes common mode loop area). This is analagous to an antenna near the ground being less effective than one in the air, and small antennas being less effective than big ones.

You could also consider a separate filter for control power which leaves the controller enclosure, on the supply to and preferably as close to the control relay contacts as possible, to reduce noise radiation and conduction to the controller power supply.

Thank you.

OK. I will just use twisted pairs.

Yes, and those relays are mechanical. (the control has built in relays too, I use those to switch mechanical relays)

The way I did it (I cannot take new pictures until Monday, since I will be away for eBay developers conference), is that the control board has the microcontroller, mechanical relays connected to the microcontroller, wires going from these relays to DIN rail mounted terminals, and then wires would go outside from the DIN terminals to devices like gas solenoid etc.

An older picture of my panel is here:

it has no DIN mounted terminals yet. Relays are glued to the board with 3M sticky double sided tape.

The one possible issue is that even though I wil have wires going from the DIN terminals to solenoids and HF, twisted, as you say, the wired going from DIN terminals to mechanical relays are not twisted.

Is that a concern?

That's good to know, yes, I will try to keep them close to the walls.

Inside (on) the control board, can I wrap them with foil that I would ground? Should I even consider it?

Should I use a zener diode (like Panasonic 151 zener) on the 120V wires? Just to cover my butt?

My microcontroller, when it reads inputs, can average them over a dozen measurements, so noise is not a huge problem when it comes to reading input.

Never knew that.

that would mean having many filters, one on each line, right?

Should I maybe just install some AC capacitors between every 120V line going into the HF compartment, and the neutral?

Or is that all overkill?

Glen, I forgot to say thank you for your sage advice.

Probably not since they are presumably short and fairly close together.

I wouldn't bother with foil, but I would pay attention to the hole through which the wires enter the controller enclosure, minimizing the size of the opening with a metal romex clamp or routing the wire through a piece of small conduit going part way to the solenoids etc.

Couldn't hurt, probably won't help with HF noise but should limit inductive kick on relay opening assuming it is a bi-directional surge suppressor type zener rated for your coil energy, which will otherwise be dumped into relay contact arcing as the relay opens. It is better to dump this energy close to the solenoid and these protection parts are often soldered directly to the solenoid terminals.

Still worth trying to minimize ...

I had in mind one filter on the power to all 3 relay contacts, as close as possible to the relays, the object being to stop the HF noise from getting to controller power supply. This position also filters HF noise from relay contact arcing.

Small line filter rated capacitors (X and Y rated) at the terminal boards of all the 120 V wiring would reduce noise some, but your existing line filter already has these, as well as a common-mode choke. The combination is a lot more effective than just the capacitors.