Only receiving "<0><0><0><0><0><0><0>" on USART signal - microcontroller

Question

When I try to receive the USART signal with my Silicon Labs CP210x USB to UART Bridge. The only thing I receive is:
<0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0><0> etc
I got the right settings selected: baud rate: 9600 data bits: 8 parity: none stop bits: 1
I use a ATmega128A3U. The test_LED (see code below) on port E pin 0 is working. I used a oscilloscope to check the uart signal. See picture here: http://imgur.com/dPxkdZ6
Does someone know a solution for this?
Does anyone know how to fix a framing error? (My UART software is giving this error)
CODE:
#define F_CPU (32000000UL) // CPU clock speed
#include <avr/io.h>
#include <util/delay.h>
void USARTF0_init() {
USARTF0_BAUDCTRLB = 0; //BSCALE is 0
USARTF0_BAUDCTRLA = 0xCF; //BSEL is 207
USARTF0_CTRLA = 0; //Disable interrupts
USARTF0_CTRLC = USART_CHSIZE_8BIT_gc; //8 data bits, no parity and 1 stop bit
USARTF0_CTRLB = USART_TXEN_bm | USART_RXEN_bm; // //Enable receive,transmit and high speed mode
PORTF_OUTSET = PIN3_bm; // Port F pin 3 as TX
PORTF_DIRSET = PIN3_bm; // TX pin as output
PORTF_OUTCLR = PIN2_bm; // Port F pin 2 as RX
PORTF_DIRCLR = PIN2_bm; // RX pin as input
}
void sendChar(char c) {
while( !(USARTF0_STATUS & USART_DREIF_bm) ); //Wait until DATA buffer is empty
USARTF0_DATA = c;
}
void sendString(char *text) {
while(*text) {
sendChar(*text++);
}
}
int main(void) {
USARTF0_init();
PORTE.DIRSET = PIN0_bm; // make port E pin 0 output (test_LED)
while(1)
{
_delay_ms(10);
sendString("Hello World!\n\r");
PORTE.OUTTGL = PIN0_bm; // test_LED Toggle
}
}
Code source: http://morf.lv/modules.php?name=tutorials&lasit=29

Answer 1

To recap the discussion, so it can useful to someone later.
If you have an oscilloscope, capture a trace for a simple test case, as #DrOctorooi has done with "#~00":
He has marked start and stop bits, and also the data bits (for serial/UART they're least-significant first)
This confirms that UART somehow works.
One should also confirm the voltage levels. 0-3.2V looks about right.
And the time base. On the captured trace it appears 1 bit takes 1.625ms, which means a baudrate of around 615. That's quite far from expected 9600. In fact, that's around 16 times slower.
Since the error has to do with time, the immediate questions are the MCU clock and UART clock dividers (and on more complex devices, and intermediate dividers and clocks).
It turns out the dividers were meant for the 32MHz clock, but this MCU has a clock of 2MHz when it comes out of reset (16 times slower, as we have seen above). Solution was to recalculate the dividers.