Yes you are right! The time is between 3 and 4 milliseconds it is not 0,75 millisec! To discover this j looked the TXD and RXD signals on my board using a digital memory scope.J readed the timings using cursors on the scope screen so the time maybe it is not 100% accurate but j think is more close to 4 milliseconds.

The FIFO buffer is enabled and is 1024 bytes for both RX e TX.

Discussing with my custumer about the problem j proposed the idea of changing the approach in this way :
- The PC sends a "Start Sampling" command.
- The board start sampling and sends the data at regular intervals until a "Stop Command" is issued.

In this way we are sure all the data is equally spaced in time, and the PC don't need to send continuosly commands with critical timings.
Obviously the program on PC side must be faster enought to collect and elaborate the data, but j think this can be solved using a well dimensioned buffer.
Do you think it is a good idea??

Thank you!