Hi,
I want to implement a function to disable the usart tx and at the same moment enable rx and vice versa.
But now when I do it like this:
void switch_to_rx()
{
UCSR0B |= (1 << RXEN0);
UCSR0B &= ~(1 << TXEN0);
}
it only turns off the register but doesn't enable the other register. Do I have to be careful with timings here?
You talk of multiple "registers"? But these are just two bits in the single UCSR0B register? How are you observing their state anyway? Your code looks right.
Oh my bad. I'm debugging with the Atmel ICE and right before sending I set a breakpoint and can observe that the rx bit is set, but the tx bit is not.
My transmit function:
void USART_0_write(const uint8_t data)
{
uint8_t tmphead;
switch_to_tx();
/* Calculate buffer index */
tmphead = (USART_0_tx_head + 1) & USART_0_TX_BUFFER_MASK;
/* Wait for free space in buffer */
while (USART_0_tx_elements == USART_0_TX_BUFFER_SIZE)
;
/* Store data in buffer */
USART_0_txbuf[tmphead] = data;
/* Store new index */
USART_0_tx_head = tmphead;
ENTER_CRITICAL(W);
USART_0_tx_elements++;
EXIT_CRITICAL(W);
/* Enable UDRE interrupt */
UCSR0B |= (1 << UDRIE0);
}
Here is a screenshot while debugging
Omg I'm so stupid. I forgot to call the function. I'm sorry, thanks, the code works.
I generally always work with a colleague (though most of the time he sits 120 miles from me) because it helps to bounce ideas off someone else with this kind of thing.
.
Happy to have been a help on this occasion.
Sounds like you need a "Rubber Duck" :-)
We don't know which AVR model. Nor the purpose of this safari. [RS485? We usually tie them together on the driver ship, and I just leave both tx and rx enabled]
Anyway, the code sequence in question is two RMW in a row. Is there any danger there, Will Robinson?
Let's see; if a Mega88 or so we have
So if 9th bit then there may be some fussing. What I'm suggesting is reg = TXMASK; and reg=RXMASK;
