What's FEC?
Satellite transponders are rather noisy communications channels are are therefore subject to a large number of errors when a signal is sent through them. Because satellite transmissions are broadcast, the receiver cannot send a message to the transmitter to say "I didn't get that last piece of information, please re-transmit it". As a result, Forward Error Correction is used, where the transmitter sends error correction information along with the actual signal so that should errors occur, the receiver can re-generate the bit stream.
FEC when used with QPSK modulation uses two forms of error correction. The first, called convolutional coding with the Viterbi algorithm code is quoted as a fraction, for example, 2/3. The fraction defines the amount of the symbol rate that's used for real data, with the remainder used error correction purposes.
After the convolutional error correction code has been removed and used as needed, a second error form of error correction is used called the Reed-Solomon code. This correction results in 188 bytes out for every 204 bytes coming in with the remainder used as parity bits to help correct any remaining errors. Additionally, the FEC scheme also uses interleaving of the data stream to prevent noise bursts from interrupting the flow of data in much the same way that CDs use it to prevent scratches from causing drop-outs.
Consider the following message:
This is a sample message
If interleaved, it might look like:
eTs haais mgi smeaesp l
Should an error occur and say wipe out the 'mgi' part of the message, the de-interleaved message will now read
This *s a sa*ple messsa*e
As a result, only single characters are missing from the message (shown here as asterix), rather than an entire word missing in the case of non-interleaved data.
As a final step, the QPSK symbols are scrambled to ensure that long runs of the same symbol value don't cause a lack of change in phase of the carrier. Since the QPSK demodulator obtains its signal clock from directly from the signal, there must be a large number of phase changes in order to re-generate the clock and of course scrambling results in this. Note: this form of scrambling is not the same as scrambling of the decoded signal.
Why use different SR/FEC values?
When people purchase time on a satellite, in effect they are primarily paying for the bandwidth. Therefore if a programmer wanted to transmit three video channels via a transponder, he would use less bandwidth than a service that transmitted six. However, the bandwidth of a transponder is finite and therefore an upper limit is placed on the SR (typically between 28 and 29 MS/s). By reducing the amount of FEC information sent along with the actual data, the number of channels can be increased. However, this then means that errors are harder to correct and that the down link stations must be able to receive a certain signal strength (i.e. use a certain size dish) in order to receive quality programming via the transponder.