So it turns out I had the LCD pinouts back to front
My first attempt at reflow soldering went well. Little bit too much paste - bridge on a pin that didn't matter.
but the LCD backlight driver didn't work. Progress was hampered by the fact that I've only ever done Arduino. Between Keil and TrueStudio, I had a hard time getting the discovery board example programs to compile. I needed them to compile so that I could add in an 'HAL_GPIOB pin1 enable' thingy.. i.e. turn on the LCD backlight, since the examples are all meant for the amoled screen with no backlight.
I did some playing. In fact I burned out one of my LCD flat flex by sticking a li-po on it and getting the polarity back to front. I guess this is why I need a current limiting power supply. DP711 or DP832 though.. ? DP832 is more money but crucially its kind of huge
Anyway then I constructed a better test.. 5 plain LEDs wired in series on some breadboard, to be driven by my little circuit. and it worked! 3.3v in, drives 5x LEDs nicely.
Then I looked closely at the LCD datasheet for the FFC pinouts and it looks like I had it the wrong way up. Well, what I mean is that I assumed the FPC to be going in the connector at top-contact rather than bottom contact. So pins 1-17 should be pins 17-1 instead. Left to right-flip.
I don't think this is fixable. So I have 49 keychain PCBs
In all honesty, it quickly became apparent that I have made myself a PCB here that's very difficult to debug (since it's face down when fitted to the ST Discovery board). Perhaps I'll spend a bit more time on rev 2 and put some test points, as well as maybe more than one type of LCD connector.