Hi!
I apologize if this subject was already discussed, but I have noticed a difference between DUE board and the schematic diagram.
The difference refers to the programming port. More specifically, to the interface between 16U2 and SAM3x . There is a level buffer (IC10) which translates the 5V level of serial TX of the 16U2 to the 3.3V level serial RX of SAM3x.
Schematic diagram indicates that pin 1 of IC10 (negate "output enable" of the three-state buffer) is tied to GND. Well, it's not. It's connected along with pin 2, sharing the signal 16U2-TX.
What happens? When signal 16U2-TX is low, then the buffer's output-enable is also low, pulling down the RX signal of SAM3x.
But when the 16U2-TX is going high, then the output of the buffer goes in three-state, leaving SAM3x RX pin to manage itself through the internal pull-up resistor.
But the internal pull-up has a high value: if I'm correct, something like 100kohms or higher. So that due to the input capacitance, the RX pin is going high very slow, causing transmission from PC to DUE to fail over 115200bps.
The simplest workaround is to solder an 0603 SMD resistor of 470ohms between pins 4 and 5 of IC10 (output, respectively 3.3V). The resistor will quickly charge the input capacitance of RX pin, restoring the correct levels and timings on RX pin.
The resistor's value might be larger, but the 3-state buffer is able of +/- 24mA output current, so it's no sweat for it. As the current consumption is not an issue for my application, I let it being 470ohms.
However, care must be taken, as the 470ohm resistor is seen by the RX0 (0 pin of DUE).
Adrian