Second, let's not lose sight of the fact that your campaign setting is very small. You're limiting yourself to one city. Not that there's anything wrong with that. It sounds interesting, but, it does cut down on the work load.
The relative scale of a setting has very little to do with the amount of work it requires to prepare. I've seen entire worlds described in 34-page gazetteers and I've seen single cities described in 600+ page tomes. (And gotten use out of both.)
But that leads me to a more general point, which is...
What changed, for me anyway, was a dawning realization of a couple of things. One, very few players ever invest even a fraction of the interest in the setting that the DM does. They invest in the story, they invest in the campaign, but, as far as the setting goes, most of the players I've played with aren't terribly concerned with it beyond how it affects their character.
What I realized was that when I was designing campaigns, I'd have all these ideas for the setting - background, history, geography, etc. I'd do research into whatever elements I thought would help, I'd spend hours and hours trying to build my next world. And, inevitably, the stresses of trying to do that AND come up with next week's material for the session burned me out.
I've been talking with a friend about these issues lately, and he mentioned something that I thought was quite useful: There is
broad prep and there's
deep prep.
What constitutes breadth and depth depends on the scale of what you're prepping, but in general you need to broad prep so that you have a framework to build on, but you should only do deep prep into those areas that the players are eminently going to see.
He recommends thinking of prep in three tiers: The broad strokes, the details, and the development.
For example, if you're designing a village for sandbox play your broad prep might be the name of the village; a map; maybe some quick notes on the town's general history, purpose, location, etc. Your detailed prep might be a dozen or so NPCs and a few locations that the PCs are likely to interact with.
Then, at that point, you wait and see what happens when the PCs actually interact with the village. What inn do they decide to stay at? Which NPCs do they seem to really like interacting with? Those are the things that you should come back in and develop with even greater depth. Everything else (the locations and people they're less interested in) you can leave sketchy.
My current campaign world now comprises several hundred pages of notes and dozens of maps. But it got it's start as a 15-page player handout; 2 pages of "secret pre-history"; a world map with the borders of the major kingdoms marked; a 2-page campaign outline; and a single adventure.
But within that material I had the broad strokes of the entire world laid out: The major countries, cultures, gods, religions, languages, historical events, and the like. It was a broad -- but simple -- foundation.
Everything else that has been designed for that world has been designed as I needed it. And, in general, I follow a pretty simple maxim: If the players aren't going to see it, then it's as if it never existed at all.
Not every setting book does this mind you. Ptolus for one does not. Freeport also includes adventure hooks for its locations. Ravens Bluff had their Points of Interest. Fantastic stuff and I wish more publishers went in that direction.
I'm going to go ahead and pimp the sig.