AN149
Rev 1.0 | 9/12
www.cmostek.com
the error between them. The system can withstand up to 15.6%. Other similar products in the industry cannot reach this
level. On the other hand, the effect of this system depends on the accuracy of the local symbol rate adjustment. Because
the crystal frequency is fixed at 26MHz, the higher the symbol rate, the lower the accuracy of the adjustment, Generally
speaking, the accuracy will be higher if the symbol rate is below 50 kHz. It can receive more continuous no flipping data
without error. That is, the lower the symbol rate, the more the correct sampling of data, the actual symbol depends on the
actual measurement. If the symbol rate is above 50 kHz, the amount of no flipping data that can be sampled correctly will
become less and less. If the symbol rate is 250 kHz, it is possible to only correctly sample 2-3 bytes without flipping data
because of the low accuracy.
3.
MANCHESTER system
–This system is changed by the COUNTER system. The basic feature is the same. The only
difference is that the system is specially designed for Manchester codec, Special processing can be done when the TX
symbol rate has unexpected changes. A common mutation is that TX data sometimes has two consecutive 0 or 1 patterns
after being encoded by Mann Chester, If one of the 0 or 1 symbol length suddenly grows longer than 50%, then the RX may
sample three 0 or 1. If the symbol length suddenly becomes shorter more than 50%, RX will only sample one 0 or 1. In
order to answer such a special situation, in this clock recovery system, the sampling will stop if 3 or more symbols without
flipping are detected in the received data. This avoids the error when the symbol suddenly grows. What happens if the
symbol length suddenly becomes shorter? Because this situation is often common in the current application, users can
uniformly speed up the symbol rate of RX. The amount of symbol can be correctly sampled when the symbol length is
shorter than 50%.
How to select the suitable CDR system:
RFPDK will select the system according to the configuration of the symbol rate. When the symbol rate is
less than or equal to 50
kHz, the TRACING system is selected. The system can not only realize the tracking,
but also correctly sample the long data
without flipping. When the symbol rate is greater than 50 kHz, less
than 200 kHz, it can be set according to the actual need of
users. If users tend to accept the large symbol rate
error, users can use the TRACING system. If users tend to accept the long
data without flipping, users can
use the COUNTING system. When the symbol rate is greater than 200 kHz, TRACING performs
better than
COUNTING. When the entire packet conforms to the Manchester coding rule (i.e., no more than 2
consecutive
symbols without flipping), and finds the symbol rate mutation in the application, MANCHESTER
system can be considered.
In addition, no matter which type of CDR system is adopted, it will improve the sensitivity of the 1-2dB
than receiving the RAW
data after the system is turned on.
How to debug the TRACING parameter:
Tracing's tracking function is done every time when the data is flipped. So the ideal situation is that each
symbol will make an
adjustment when the packet has preamble. This will speed up the tracking speed.
The following introduces the debugging methods for several parameters.
CDR_RANGE_SEL<1:0>
This parameter selects the symbol rate deviation that can be tracked. The percentage given in the
register content is not the
percentage of the symbol rate, but
it’s the percentage of the symbol length, For
example, the symbol rate is 10 kHz, a symbol
length is 100 ns, then +/-15.6% is refers to the TX symbol
length is in the range of 84.4 ns to 115.6 ns, RX can detect the symbol
and adjust its length to close to TX, so as to eliminate the deviation of the symbol rate. If it is converted to a percentage of the
symbol rate, then +15.6% length percentage can be converted to -13.5% symbol rate percentage, and -15.6% can be converted
to
+18.5%.
The user might ask why not directly select the maximum range of tracking. This is because the larger the
tracking range, the
harder it is to track, and the more time it takes. Therefore, it is recommended that the user
select the most appropriate range
according to the actual application requirement.
