I wrote: > - the wheel also seems to be fine. Debouncing needs some looking > into.
Solved. The problem wasn't debouncing but that I forgot to set a pull-up on CLK / SCLK / WHEEL_B. CLK is the only signal that doesn't have a hard-wired pull-up in the Ben. By not setting the pull-up, the B signal (half of the rotary encoder) floated. So the state the Ben saw depended not only on the switch but also on flux, humidity, whether I touched that contact, and so on. Setting the pull-up removed all "ghost movements". > - voltage levels, especially the level of the high voltage (the OLED > has a DC-DC converter to generate ~12 V), Fun discovery: it's only ~7.4 V, which is actually what the controller chip is designed to provide. The display panel is rated at 12.0 +/- 0.5 V, so I'm quite below that. It does look good, though :) > - behaviour when turning on more than 50% of the pixels. Nothing untoward happens. I see a little bit of "shadowing" (lines with more pixels getting dimmer) but that's only when turning on a lot of pixels and even then it's barely visible. Seems that the total current is still within what the controller can supply. The controller's charge pump is rated at 6 mA while the display could draw up to 31 mA at 12 V. The 7.4 V operation seems to help :) I've then made some more systematic measurements: OLED voltage Default image All white Display off --------------- --------------- --------------- ----------- 3.3 V + pump 9.98 mA 22.8 mA (1) 114 uA 7.5 V external 3.54 mA 11.7 mA 398 uA 12.0 V external 7.57 mA (2) 667 uA (1) decays from 22.9 mA (2) overcurrent event The first set is with all the power coming from the Ben, measuring the total current delivered by the Ben through VDD. I did drive all IO lines but WHELL_COM low for the "display off" measurement (since I'd otherwise have a negative current on VDD) but left them in their normal idle state otherwise. I then turned off the charge pump and supplied 7.5 V externally. The current measured is now what goes into the OLED's VCC and thus doesn't include the (usually negligible) consumption of the OLED's logic. The two data sets suggest that the charge pump will be able to drive up to about 50% of the pixels without any problems. Then I increased the external supply to 12 V. Brightness increased noticeably and so did the current. Alas, something went wrong when trying the all white image: the display briefly flashed on, there was an angry beep from the overcurrent detection of my power supply (set to 50 mA), and the display went dark. After that, it wouldn't accept external power or output the charge pump voltage, so it seems the trace/wire got fried somwhere. It still runs fine on the charge pump, though, despite the connection to the external 2.2 uF buffer capacitor now being severed. I'll have to see how the display performs in bright daylight. If it's still readable when only powered by the charge pump (or if even running it at 12 V wouldn't help), then I'd prefer to stick with ~7.5 V. Otherwise, we'll need another and more complex boost converter. - Werner _______________________________________________ Qi Hardware Discussion List Mail to list (members only): [email protected] Subscribe or Unsubscribe: http://lists.en.qi-hardware.com/mailman/listinfo/discussion
