Is a serial communication protocol which as the name suggests can be used sync (using clocks) or async(using start/stop bits).
Device 1 Device 2
+----------+ +----------+
| TX|-----------|RX |
| RX|-----------|TX |
| GND|-----+-----|GND |
+----------+ | +----------+
|
-----
---
-
For asynchronous communicationn the sender and reicever must agree on a data transfer rate, which is called the baud rate which is number of bits transferred per second.
Simplex mode is when data is transferred in one direction only. Half-duplex is when data is transferred in both directions but not at the same time. Full-duplex is when data can be transferred simultainously in both directions at the same time.
The wire protocol:
+-----------+-------------+---------------+-------------+
|1 start bit|5-9 data bits|0-1 parity bits|1-2 stop bits|
+-----------+-------------+---------------+-------------+
Normally, the sender keeps the transmission (Tx) line high, say at 5V and this is polled/sampled by the reciever (Rx). When this goes low the receiver will start sampling/reading for data packets.
idle start
| bit
↓ ↓
5V -----+ +------+ +
| | | |
| | d0 | d1 |
0V +------+ +------+
| | | |
Both sides need to agree on things like if how many data bits the packet contains, if there is a parity bit or not (and if it is odd or even), and how many stop bits are in use (1-2 bits). In addition they also need the receiver needs know how many bits the sender sent per second. Both sides need to agree on this rate as well. The receiver needs to know how many bit are being sent per second so that it can determine how long a 5V pulse is determined to be a 1 bit, and how long a 0V pulse is considered to be a 0 bit value. This is configured by the sender and receiver specifying a bits per second for sending and receiving and is called the baud rate.
Baud rate: 9600 (9600 bits/s)
idle start
| bit
↓ ↓
5V -----+ +------+------+
| | | |
| | d0 | d1 |
0V +------+
1/9600 1/9600
Remember that the receiver reads one value at a time from the wire and in the above case the first two bits are 1s. If the receiver does not know the time to sample, or is does not use the same rate as the sender, the read values may become incorrect. What the receiver does is it knows how long a single bit should be and samples with that frequency so it will determine that there were two 1 bits as the wire was held high for 2/9600.
Is a common unit in symbols per second in electronic communication. The number of symbol changes, signaling events in the transmission medium. Baud is named after Emile Baudot, who invented the 5-bit teletype code.
The difference between bit rate and baud rate is that bit rate is the number of bits sent per second, where as baud rate is the number of signal units per second. The bit rate is the number of bits to be transmitted per second, they are of two states, either 1 or 0. When this is converted into a signal and travels on a wire the signal can get distorted due to noise. So the 1s and 0s we send from the application might have a different representation on the wire and this is the reason for talking about symbols instead of bits.
So we know already that the baud rate is the number of changes to the signal per second accross the wire. Now the baud rate can be higher or lower than the bit rate.
Just to try to visualize what this means having a baud rate that can be higher
or lower than the bit rate, consider sending 1011could be sent using a baud
rate:
idle start
| bit
↓ ↓
5V -----+ +------+ +----------+
| | | | |
| | 1 | 0 | 1 1 |
0V +------+ +------+ +
And also have higher baud rate:
idle start
| bit
↓ ↓
5V -----+ +---+ +-----+
| | | | |
| | 1 | 0 | 1 1 |
0V +---+ +---+ +
So the baud rate can be higher or lower than the bit rate.
bit rate = baud rate * bits per symbol
The receiver constantly samples/reads/polls the data line and it does this more often than the baud rate. An common setting is to sample 16 times the baud rate. So our baud rate might be 9600 bits/s then the frequency that the reciever polls the data line would be 9600x16=153600 times per second. But the data rate being used is only 9600 bits per second so the receiver is reading/sampling more than that, so it is "over sampling".
The receiver will sample the data line and look for when the it goes from idle (high) to low. When this happens the reciever will start a timer:
idle start
| bit
↓ ↓
5V -----+ +------+ +
| | | |
| | d0 | d1 |
0V +------+ +------+
↑
start counter (0-15 ticks)
When the tick counter reaches 7, the middle of the start bit, it clears the tick counter and the counter restarts:
idle start
| bit
↓ ↓
5V -----+ +------+ +
| | | |
| | d0 | d1 |
0V +------+ +------+
↑
7 ticks
This is done to estimate the middle of each bit. So it can then count from 0 to 15 and then we know that is the first data bit (data bit 0) and that value can be read into a register:
idle start
| bit
↓ ↓
5V -----+ +------+ +
| | | |
| | d0 | d1 |
0V +------+ +------+
↑ ↑
+-------+
0 15
This process repeats for the number of bits that make up the data (this is agreed upon to be 6, 7, or 8 bits). And this will optionally happen for a parity bit if that is in use.
If the stop bit is configured to be 1 bit then we again count to 16:
stop bit
↓
+------+ +
| | | |
| | d7 | |
+------+ +------+
↑ ↑
+-------+
0 15
If the stop bit is configured to be 0.5 bits that we count to 8:
stop bit
↓
+------+ +
| | | |
| | d7 | |
+------+ +------+
↑ ↑
+---+
0 7
If the stop bit is configured to be 1.5 bits that we count to 24:
stop bit
↓
+------+ +------+
| | | | |
| | d7 | | |
+------+ +------+ +
↑ ↑
+--------------+
0 24
If the stop bit is configured to be 2 bits that we count to 32:
stop bit
↓
+------+ +------+
| | | | |
| | d7 | | |
+------+ +------+ +
↑ ↑
+-----------------+
0 32
From the previous section about oversampling we can follow the following formula to calculate the baud rate for a system.
OVER8 = 1, then 8 will be used as the oversampling value
OVER8 = 0, then 16 will be used as the oversampling value
f_ck
Tx/Rx baud = --------------------
8 * (2 - OVER8) * USARTDIV
f_ck = peripheral clock
USARTDIV = divide factor to generate different baud rates.
The peripheral clock should be specified in the reference manual:
6.2 Clocks
...
Several prescalers can be used to configure the frequency of the AHB and the APB
domains. The AHB and the APB domains maximum frequency is 48 MHz.
So if we want to have a baud rate of 9600 we can plug these values into the forumla:
48Mhz
9600 = -------------------------- =
(8 * 2 * USARTDIV)
48Mhz (8 * 2 * USARTDIV)
9600 (8 * 2 * USARTDIV) = -------------------------- * ------------------
(8 * 2 * USARTDIV) 1
9600 (8 * 2 * USARTDIV) = 48Mhz
9600 * 16 * USARTDIV = 48Mhz
153600 * USARTDIV = 48Mhz
153600 * USARTDIV 48Mhz
----------------- = -------
153600 153600
48Mhz
USARTDIV = -------
153600
48000000
USARTDIV = --------- = 312
153600
So this is the value that should be placed in UART_BRR:
27.8.4 Baud rate register (USART_BRR)
This register can only be written when the USART is disabled (UE=0).
Bits 31:16 Reserved, must be kept at reset value
Bits 15:4 BRR[15:4]
BRR[15:4] = USARTDIV[15:4]
Bits 3:0 BRR[3:0]
When OVER8 = 0, BRR[3:0] = USARTDIV[3:0].
When OVER8 = 1:
BRR[2:0] = USARTDIV[3:0] shifted 1 bit to the right.
BRR[3] must be kept cleared.
So we have 15 bite available in this register for our usage. Notice that the
bits 4-15 are specified as USARTDIV[15:4] which is like an array slice.
If we take the value we calculated above, 312 and turn it into binary form we
get:
USARTDIV = 0000 0001 0011 1000 (0x138)
USARTDIV[15:4] = 0000 0001 0011 (0x13)
OVER8 = 0:
USARTDIV[3:0] = 1000 (0x8)
BRR = 0x138
OVER8 = 1:
1000 << 1 = 0000
USARTDIV[3:0] = 0000 (0x8)
BRR = 0x130
So where is OVER8 specified?
27.8.1 Control register 1 (USART_CR1)
...
Bit 15 OVER8: Oversampling mode
0: Oversampling by 16
1: Oversampling by 8
This bit can only be written when the USART is disabled (UE=0).