First try, and we've got a shaky display. Very shaky. I can make out the Beagle dog logo, see screens changing but the display isn't stable at all.
Soldering problem? Ran the soldering iron across the FPC connector to make sure all joints are tight and good. Did a continuity check, all connections look good, no bridges when viewed under a magnifier.
Continuity loss? Used a multimeter to check connections to the FPC. All looks good, nothing strange. Somewhere in between this testing the LED backlight wasn't lighting up when I connected the LCD. Whoops! I burn 4 LED drivers in the process (but they seem to have open LED "protection"?). Add a 16V Zener across the LED lines to prevent re-occurrence. The LEDs are in a 3-series (and currently unknown parallel lines) configuration and draw 9.5V.
Maybe some signal integrity issues? It's Scope time.
- Put cape on the BBB, LCD disconnected. Probe signal pins. The highest frequency signal, PCLK (@30 MHz) appears clean enough. Data lines look good as well.
- Put cape on the BBB, connect LCD and power on. On probing, data lines look noisy, but, they were clean before putting on the LCD?
- With the LCD and BBB powered, probe VDD_3V3 rail. Lots of high frequency spikes. Power down the LCD. Spikes disappear.
- For the next test, power on the LCD but power down the backlight. No noticeable spikes. Probe the LCD data lines. Signals look good.
- Look at the LCD panel under strong light. Was able to make out the Beagle on the screen, once again. The screen looks stable (a direct consequence of the fact that the signals look good on the scope).
- Power on the LED backlight and see again the noise on the data lines and 3V3 rail.
On testing with a multimeter, the LCD (with backlight) draws ~400mA from the power rail.
Okay, now that we know the issue, we should need to just add some bypass caps to fix it? Sprinkle a few 10uF 0805 ceramic caps, one before the LED driver and the other across the LED driver output still the display appears to be as unstable as it was before.
At this point I asked in the HaD channel, and with feedback that maybe the onboard 3.3V of the BBB is not supplying enough current to power everything, try to power the entire cape with a 1117-3.3 fed from the 5V of the BBB. An unstable display, still.
I look at existing cape reference designs (CircuitCo BB-VIEW7 and 4DSystems 70T) and find that they power their LED drivers off the 5V rail, and not the 3V3 VDD rail. Maybe they knew something about ripple from LED drivers affecting the LCD supply?
(To be continued)