Project Logs

Collapse

• V1.1 - Battery powered

  max • 05/22/2024 at 01:30 • 0 comments

  Got to test out v1.1 of the device on set this week, and it once again worked beyond all expectation. This iteration includes a battery, charger board, and power switch, in addition to some other basic improvements and groundwork for future stuff.

  
  First and foremost, the device is smaller than v1 in every dimension. I've rounded off some more edges to aid with the feeling of a small & portable device, and shaved down any walls that I felt could stand to be thinner. I inset the screws that mount the display to the framework, to make a little more room inside, as well.

  The battery I put in there is an Adafruit 2500 mAh LiPo pack, and the management board is a TinyUPS3.0. I stripped off the battery connector & pins and soldered my connections directly to the board to save space, with good ole hot glue globs to provide strain relief. Haven't had the chance to test runtime on the battery, but it easily lasted half an hour during lunch on set, and I don't really intend to run it on battery power for long stretches of time, anyway.
  

  I designed a push-button for the top of the case, the intention being that a short press will lock/unlock the touchscreen, and a long press will initiate shutdown/boot. Currently the button doesn't do anything - I need a reliable way to tell when the rpi is in its powered-down state, but haven't found one yet. My plan had been a sort of latching power switch using a GPIO pin, resistor, and NPN hooked up to the enable pin on the battery management board, but every pin exposed by the rerouter carrier board seems to pull high even when the cm4 is in shutdown. For now I just have a simple on/off switch connected to that pin instead. The Pi is using overlayFS, so unsafe shutdowns should be fine.

  
  I think future iterations will have to use a separate controller (possibly a QT Py, as they're small and I seem to have an abundance of them) to handle that button, so I can have it do whatever I need without relying on the pi GPIO etc.

  Speaking of, I had to find an 8-pin fpc and breakout board in order to access gpio & i2c on the pi. No problem finding the breakout, but I only had a 15-pin cable at my disposal. Some careful trimming later, and it's working beautifully. Don't trust the pin labeling on the breakout board, however - I did it backwards once before I got i2c communication with the battery board working. But hey, it is working! I can monitor battery voltage etc from a Python script now. Still looking for an elegant solution for graphical battery monitoring, though.
  

  Not strictly related to the device itself, but I also came up with a nice little enclosure for the camera-side transmitter.

  The only real issue with this iteration of the device is that I seem to have fried something on the battery management board - handoff from DC power to battery power doesn't work anymore. The device works fine in both states, but it seems that something on the board gets overloaded and shuts off when it tries to transition from one to the other. Frustrating, as it was working fine until the first test with the case fully closed. I'll have to open it up and take a look soon, though that probably means just buying a new board and being more careful.

• V1 - We have liftoff

  max • 05/10/2024 at 00:33 • 0 comments

  First entry.

  The goal of this project is to create a small (ish) tablet-style computer for the remote control of cinema camera systems on set. All modern cinema cameras include a web interface of some sort, the main reason we don't rely on these on set is because WiFi systems are not reliable enough. There is just too much interference in the 2.4ghz (and even 5ghz) spectrum on set to get a reliable connection more than a few feet away from the camera. Many attempts have been made to rectify this - stronger antennas, external routers, etc., but none of them have worked any better for me.

  But then I discovered the BitBox, which operates in the 900mhz spectrum instead. While I couldn't afford to pick up one of those lovely systems, it sent me down a rabbit hole of research about 900mhz networking, ultimately landing on the WiFi HaLow protocol. This is a standardized protocol for bridging internet connections over a 900mhz bridge, making them pretty much impervious to interference and increasing range by a ton. The one drawback is lower bandwidth, but camera control is practically nothing, data-wise. It should be perfect.

  I found another camera person making DIY camera control boxes, and from there found the Loocam Bridge kit. This is a $60 plug and play kit for bridging Ethernet connections over 900mhz. It works completely seamlessly, without any setup needed. Kind of mind-blowing.

  My first test was with a SmallHD monitor with a camera control license, and it worked so much better than over wifi that I almost just stopped there - but I really wanted a dedicated device for camera control, rather than running it through my monitor, so I set to work building this little guy. I got it up and running in time for my most recent job, albeit just barely. Work absolutely continued while on set, but it was functioning! The DP called to flip the camera sensor for low-mode, and I was able to do it from my box without needing to reattach the viewfinder.
  

  Some photos of the current state of things:

  I was reluctant to permanently modify the 900mhz board (a Netbridge 2.1 board from Loocam) until I was sure this would work, so at the moment I've just soldered the power and ground lines onto the sides of the barrel jack. It should be able to pop back into its original housing without issue, though now that it *is* working here, I doubt I will ever do that. Perhaps I will remove the barrel jack entirely now for space savings.

  Challenges that I had to overcome to get to this point:

  • Designing the case. I didn't want to use the adhesive backing on the LCD, in case I needed to redesign the case later on down the road. I also wanted to essentially overmold the edges of the LCD to protect it from bumps. I landed on a 3-layer design, which is probably over complicated, but oh well. The LCD is screwed into the middle "frame" part, then covered with the front panel. The pi and 900mhz board go into the rear case, then the ribbon connects the two halves. I added some fins and slots on all four sides to help align the parts, and there's an M3x40 screw in each corner to bind the whole thing shut...but at the moment it's just gaff taped closed so I can easily access the insides for further tinkering.

  • Getting the touchscreen working. I selected a Waveshare LCD module (5in DSI LCD B), but first attempted this with a BTT Pi TFT which I just could not get working. The Waveshare worked fine once I figured out that I needed...

  Read more »

View all 2 project logs