Benchwork
The benchwork for the prior layout was constructed by ripping 3" joists from OSB. Subroadbed was masonite spline. As a result, the benchwork for the entire layout was about $40.

The initial construction for New Britain was using the same approach for the benchwork itself. Subroadbed is just OSB due to the large flat areas and lots of track. As I was working on the lower deck and trying to determine the specifics of the benchwork, I found that the furring strips available were of decent quality, and much cheaper than shopping in the dimensional lumber section. So I switched to using 1x2 and 1x3 furring strips for the benchwork.
 
Berlin Line
The initial Berlin Line was about a 3% grade on masonite spline. This worked great, until I tested it with the 44-tonners. They could only pull 2-3 cars up the (curved) grade. Now that I've been able to add 3 oz of weight to the locomotives it might have been OK. But I had no idea I'd be able to do that then.

So I had to raise the grade, and to ensure it was even I opted to build a level deck and use Woodland Scenics foam risers instead. This works very well (although it's a bit more complicated on multitrack lines and sidings). 

The problem is that it meant I also had to raise the entire east side of town. This puts a grade on the long siding where cars are dropped off, which is also the yard lead. Operationally works OK, and even better after I ended up lowering the deck as much as I possibly could (which also involved rotating the helix).
Turnouts
My initial intent was to handlay all of the turnouts using Central Valley turnout strips and Proto87 Stores frogs and points. Part of the reason for this was to use a technique I developed with a z-scale PC board tie as the throwbar and spacer bars. While not terribly difficult to build, I found I wasn't all that fast either.

But the big issue came with trying to actuate the turnouts. I had several approaches that seemed to work in testing, but they were harder to implement in practice. I also found that I had a very small margin of error. Even under-table actuation was more complicated than I expected.

Enter MicroEngineering turnouts. I hadn't actually purchased or tested these, but once I did I found I really liked them. Not quite the level of detail that I was looking for, but very good and I figured I could add the missing details myself.

My primary goal is to get the layout up, scenicked and operating. So I decided that with ME turnouts  I could get the layout operational much faster, and at a similar cost. 

I also really like the idea of flipping the points of the turnout directly, because you always know exactly which turnout you are throwing.