Hi guys,
Just to let you know that I found a problem with my board design. It the way I am using the resistor dividers to shift 5V from arduino to 3.3V for the RFM22. Because I have modified the RF22 library to run at very high speed, the resistors and the capacitance of the traces begin to act like a RC filter. This can corrupt the SPI data, causing lockups. This means that the SPI bus is operating very close to the limit, voltage problems with power supply can easily cause a lockup, especially if you run a lot of servos. If you only use the board for spec scan or telemetry you may not have this problem at all.
As you can see here, the SPI clock out from arduino (blue) is distorted when it gets to the RF22 module (yellow). This screen shot is a bit exaggerated because I was using X1 probe, it's better with X10.
Solutions:
- Software solution: reduce SPI speed, change the SPI speed down to 2MHz in the RF22.cpp file. Change the line in RF22::init() to SPI.setClockDivider(SPI_CLOCK_DIV8);
- Hardware solution: reduce resistor values. I tried halving the resistor divider values (R2 - R7), and it works much better. Replace 1K with 470 Ohm, and 1.5k with 680 Ohm.
- Both. <-- I recommend this for best reliability