In case you missed it, Part 1 is listed here

Aesthetic Function

Next one is a bit of a personal pet peeve. I really enjoy auto dimming when it is implemented correctly. I'm a photographer and I specialize in low light photography. The next thing I need is to have a glaring cell phone screen light up and ruining no less than some night vision. For detecting the amount of light, most use an LDR in a voltage divider network and pass it to an ADC. It works for those simple things, but I am dead set against using any ADC lines for just a backlight control. I needed something more precise and with a logarithmic output so it'd adjust relative to how the human would. This eventually led me to the APDS-9007. It outputs a small amount of current dependent upon how much light hits it, from 3 to 70K lux. Wikipedia, as usual, has a decent article on lux.

A detail about the SEPS525-based displays stuck out to me: They have a resistor valued in the high 60K's to low 70K's tagged onto 2 pins. It's even labeled as a control for internal brightness. Information is lacking about how much current is flowing through the resistor, if it's a constant current source on the other end, if it's a part of a voltage divider, if a greater value increases brightness or decreases it, etc etc. The only way to figure this out would be to test it. Unfortunately I'm dealing with small pitch SMD components so I either make breakout boards and check everything and then make up the prototype, or I build the test circuit into the boards. I chose the latter especially considering I'd need 1 of the 2 resulting circuits, albeit modified, in the final design to have this function.

I'm sure you've figured out that I'm talking about dimming control on the displays by way of the APDS-9007. Using a transistor as gain and coupled with a pair of resistors and switching between a PNP and NPN transistor, I can figure out how this resistor's value affect the OLEDs display brightness and use it to my advantage. Full details are outlined here. In short, I'll have independent and tailored response of the OLED displays without needing user interaction.

Mechanical Design

So, my target size was to get the display to fit inside a 2" gauge pod. Unfortunately this turned out to not exactly be possible. Well, maybe. I need to see if I can have a 2" body and is a 2.25" board dia. is an issue. Now, how did I get to this size? First it was a simple function of the display size. Though the viewable area is only ~1.7", there are conductors and extra glass that add extra size to the OLED display's structure. I have to also fit 3 buttons within proximity of each other on there somewhere too. Since the display is wider than it is tall, the top of bottom would need to be utilized for the buttons.

Buttons. I kinda hate them because they take up so much space for such a mundane task. For a spell, I researched capacitive buttons that way the space behind the screen had some function. The added component list and complexity isn't worth it for this.

I then realized that I had a little problem, the display FPC. There was simply no way to have it plugged in on the same side of the PCB as the display resided. I had to break out the calculator and let it deal with converting mm to mil and pixels to each and back and for etc etc, while I was having to think in 3D space to know where that connector would be on one side of the board and where the display would end of residing and where the graphics would visually end up residing.

From the graphic way up there at the top, you'll notice I have a slot in the board. I came to the conclusion that it is the most elegant solution of running the FPC through the board instead of chopping off the bottom of a circle. I think I have it shuffled appropriately to minimize chaffing and to allow some manufacturer tolerances. It was a few hours of sitting around with a calculator, calipers, a display. and photoshop that I kind wish I could forget.

You'll notice too that the buttons are up on top. Due to the display's FPC I didn't have much of a choice in placement of the buttons. I almost rotated the board upside down for this so your fingers or hands wouldn't cover the display when pushing a button, but it proved to not be desirable. Why? If you had a gauge that was inside a pod and it wasn't centered, how much would it bug you? Well, It'd bug me a bit. With the board rotate 180˚, the placement of the gauge face(s) would be such that their centerline would be around 0.2" lower than centered. The lack of concentricity would easily be seen. Having it right side up, buttons on top, allow the centerline to be less than 0.05" below center of the board.

I have the buttons up and the top, and the center one, the selection button, raised along the perimeter of the board. Below it is the ideally placed light sensor. Now, the other part I've had to tackle is figuring out how to tack the display to the board. I've strategically placed a view vias around to fit into holes of the FPC, act as a shelf for the display, be corner tie downs so it won't fall forward, and to allow some material to be roped though (wire, thin silicone tube) the board and be a cushion behind the display to help protect against vibration. Overall, a little more tedious, but less permanent than glue.

I have the LEDs aligned with the display's viewing area, not the center of the board.I may shuffle them down latter, but if the whole screen has something displayed on it by chance, then I prefer to have the accent and warning lights aligned with that region. Who knows, I may split the difference in the end.

Adapter Boards

Lastly is the 6x2, well two 3x2, female pin header on the board. A mini DisplayPort cable is like most others where they are flexible, but tight spaces are not their friends. Yes, the cable can be bent to fit in such a space if a right angle connector is used on the PCB, but then I'd have less board space to clearance the cable. As perpendicular adaptors are hard to come by, making a simple adapter board proves to be advantageous. Ironically, this also simplifies the board layout in case a different receptacle has to be used so the display baord's pin out stays constant.

This provides me with another unique solution. If my chance my assumption about being able to drive the displays while doing everythign else on the board is wrong, I have a "simple" solution. I can turn the adapter boards into an intermediary. I've had the concern expressed that the displays I'm using might be too small. At the same time, there isn't enough bandwidth to drive a larger display at 60 fps, let alone 3 of them. If I change the design scheme over to sending drawing commands instead of pixel data, then I just need a MCU on the other end of the cable to interpret the drawing commands and draw the display for me. Either way, D-DAQ will meet my design specs even if I have to wait until version 2 to do so.