When I got back from work I sat down and sketched out the diagram from my the night before.
Holy hell, it actually seemed to mesh pretty well with reality.
Most of the time you try to get stuff done way too tired — or even sleeping — you wind up with all the things that only make sense in the altered state of the slumbering mind. I know I’ve solved big problems in the past and took notes. Sitting at a Denny’s with Gabe solving all the problems at NetForce. The next morning (or, honestly, the afternoon) would bring the rude realization that I may as well have been on crack.
Not this time.
The chip on the top left is the TI TPS2561a, the one under it is TI TPS2513A. To the right is a couple of USB ports (or the pins for them). To the right of those are the low-value current sense resistors and relatively high-value decoupling capacitors.
The solid lines are on the chip side of the board. Dotted are on the opposite side. The top side is actually the bottom in a physical reality, but who’s counting.
The only thing I left off is the TI TPD2E001 ESD protection. I did wind up drawing a few extra traces for the enable and fault lines I didn’t really need to draw.
The lines on the bottom go off to a dual current-sense amp (not a challenge in layout) and then to an ADC. The digital lines go off to a I2C multiplexor. All of that part can be autorouted. In the case of the power traces, I think I have to manually route them (in my sleep I guess) to ensure that they don’t wind up acting strange since I’ll be passing 2.1A over them.
Oh, and I want to keep it a two-lawer board. :-)