Glove Main Loop Development

[digitaldanny]

Describe the solution you'd like Most of the main bullet points are copied from Glove PCB Features section of the PDR document. There are small modifications to requirements 1 and 3 based on changes approved by the professors. Secondary bullet points are additional context for the original requirements.

• Capture bends in 3 joints on the right hand.
  • Use the ADC and Mux interfaces to read in the 14-bit ADC values.
  • Compress the 14-bit ADC values into 8-bit ADC values to reduce the transferred message size over Bluetooth. This compression should still be enough to detect small finger movements.
• Provide sensor data that can be converted into hand orientation values on 2 axes (either on the glove or in the desktop application).
  • The algorithm will use gyroscope and accelerometer values. There will be drift on the third axis, but I do not plan on using this axis for the final demo.
  • Capture the gyro error offset while sitting still. These values will need to be subtracted from all gyro readings.
  • Create a millis() implementation to capture the time difference between sensor data captures.
  • How much can I compress these values before transferring over Bluetooth?
• The glove must operate wirelessly. I will power the glove with a rechargeable battery and transmit data to the application wirelessly using Bluetooth.
  • The HC-05 should connect/disconnect/communicate with the HC-05 on the DSP application side.
  • (Optional) Detect unexpected connection/disconnection to enable reconnection without restarting the system.
• Provide a text-based user interface on the glove using an LCD. This display will show the battery’s charge percentage and Bluetooth connection status.
  • (Optional) Create a "sprite" that shows battery state at 20-25% increments.
• Capture the battery's State of Charge. Being tracked in #51.

[digitaldanny]

It might be a good idea to have a main "kernel" loop that schedules tasks that will be completed. The LCD / State of Charge do not need to be updated as frequently as the HC-05 needs to transfer data. For example, the kernel could allow the LCD to update once every 50 iterations and teh HC-05 could update once every 2 loops.

[digitaldanny]

After experimenting with the flex sensor readings, I found out that I need to use smaller capacitor values for my low pass filter. Right now, my capacitor takes too long to discharge before taking readings for the next finger. This means that each finger's voltage can be affected by the previous finger's voltage.

I can play with 2 parameters to keep the benefits of the LPF while avoiding interference.

• Using smaller capacitor values that will discharge more quickly. This will also require changing the resistance values, so it will result in a voltage drop and lower ADC resolution.
• Software delays after switching the mux pins.

The noise is actually not very noticeable after applying the data compression. Since I am removing the bottom 6 bits of the ADC reading anyways to fit the 14-bit reading into 8 bits, I lose most of the noise there. I may end up just removing this circuit completely.