To question about if there's a circuit and components for an 118a frequency generator, I've never seen one myself, but perhaps there's some out there somewhere. Since it seems to be a potted unit, it was never intended to be repaired, so I'm guessing there isn't a schematic provided in the BSP's (Bell System Practices). However, there might be an internal document at Western Electric.. such a document might be harder to come across. Or perhaps there's a patent that describes its design?

My guess is it's a simple ferroresonant transformer circuit, perhaps much like in the TelLabs 8101 ring generator (see below). These are well suited to divide the input AC frequency from 60hz -> 30hz used for 1A2 ringing.

In a 551 KSU like my 551a, these 118a's are indeed "powered on" at all times, but I've noticed they don't consume much power at all when idle, which leads me to think there's some additional current sensing smarts built in to power down the transformer when there's no load from ringing.

I can tell you the TelLabs 8101 does NOT have such a load sensing circuit, as it does consume some power when idle -- about 8 watts. The idle AC hum of the 8101's transformer is loud enough to be heard across a large room.

Here's a photo of a TelLabs 8101, and an excerpt from a technical manual for it showing the ferroresonant circuit and a brief circuit description:

[Linked Image from seriss.com]

Thing is I think such transformers are custom wound for this purpose, and not just simple isolation transformers with a center tap on the secondary with a capacitor tank circuit or such. But I've not really investigated this; exotic transformer designs are not my thing. But there's quite a bit of info out there on ferroresonant transformers.

There's certainly many electronic circuits for ring generators from e.g. 12VDC, 24VDC, etc. IIRC, the small "Black Magic" and "PowerDSINE" ring generator modules have fully documented circuits (that are much more complicated) and do some clever stuff to handle dead shorts and drop voltage as load increases, and work equally well in place of an 118a.

Anyway, I'd be certainly interested too in seeing a schematic circuit for the 118a. But it might be hard to replicate if it's using a custom wound transformer.