Leonardo dead slow if its transmit buffer is not emptied (by reciever)

Using Arduino Installation & Troubleshooting
#1

An given Arduino “sketch” write this once a second
$1;0;23569;0;0;0;0;0;0;0;0;0;0;0;0

  Serial.begin(115200);
.......
 Serial.print(C1 * 100, 0);
  Serial.write(59); //separator ;
  Serial.print(C2 * 100, 0);
.......

simple.
If I open serial monitor after programming (ctrl-alt-m) I see the data,
when the monitor is closed, the Leonardo is barely moving, almost stops.
ctrl-alt-m again, and it’s at full speed… , close the serial monitor, it “stops” …

seems like something related to hardware flow control.
How can I make it ignore DTR(?), and just bitbang the data out ?

#2

even millis() are almost stopped. counting seconds, stops while the terminal on the computer is disconnected.

This is a safety hazard, using Leonardo for logging purposes, may "freeze" the controller if the logging destination stops or gets disconnected.

#3

Please post the WHOLE sketch if you dont mind.

Maybe if other people can try it to see if they get the same issues or if they spot something in the sketch.

Also please describe your programming and use environment in detail as that may also help.

#4

This issue have surely been observed by others. I hope for some of the masters here to have a solution.
Anyway the always requested "whole" sketch includes/depends on data from two ADC, needs a DAC an RTC.

so I am posting something that anyone could test:

unsigned long last;

void setup() {
  Serial.begin(115200);
  pinMode(13, OUTPUT);
}

void loop() {
  if (millis() - last > 999)  {
    last = millis();
    Serial.println("one second");
    digitalWrite(13, !digitalRead(13));
  }
}

observe led blinking stops if monitor closed...

#5

On my Leonardo clone the blinking continues at the expected rate regardless of whether the Serial Monitor is open or closed. I'm using 1.6.12 on a MacBook Pro running OS X 10.12.2

#6

Could you try to power it externally, and try to run it without USB connected ? - see if it works then ?
mine does not.

#7

This is crazy, but on a Pro Micro (ATmega32U4) The attached code fail every time.
on a Leonardo, the test code above does not fail, but the full code (with more I2c devices) - does fail in the same way as the test on pro micro..

#8