Thread is a little wide to , pick up all the points
amazed , no live psk on the bands , must admit , it was never too robust , worked nicely on the higher bands , where not too much 'turbulence , especially 10 meters
/OFDM systems are good at dealing with where where interference at one very precise specific frequency./
Possibly not so, this is a ofdm signal ,[was a image ?] showing selective fade , a high % of the signal escapes the attenuation , the out come being, the system continues to function , ie its very capable of dealing with noise distributed in the channel , in terms of time and spectrum , only needs to capture ' bits / patches of the transmission , to recover the pay-load ..
But for that, it must have a rough idea when the bits start.
That's why systems like very narrow-band modes (WSPR and similar) use an external time-reference (i.e. the clock of the PC) to determine when it must start receiving.
Not so , opera is a free running system, with no time locking required , once the data bits are recovered , the sync is then extracted ..noting opera is packet like system where the data frame is fixed length and the payload is also fixed length ,
the coding encapsulates the payload , and not having to provide a open ended , non time limited decode , there is no start or finish ,
only the need to capture 40 + % this requires a high redundancy level , noting the deep search [ correlation] second chance detector provided for 477&136 KHz , retains the 40+% min capture criteria , as well as the AFC , making it very robust and independent of equipment stability
Other low s/n modes , require to establish ''lock' before the decoding can commence , the two tone start to olivia , or the ROS pre-amble , the delay to print is the FEC time , The ROS-MF modes have a much longer FEC time to over come the QSB found on MF , ROS-HF much shorter , as Opera op8 bridges the slow qsb MF , Op32 30 mins , works well on 136 where fade times are very long , each retains the 40 + % requirement
Opera could be categorised as a data burst system , perhaps not so [much ?] packet , as that requires the conventional start - data- stop format , but still a single burst , could be received ..
wspr uses time frame , as the coding is very low level , make use of correlation , to provide low s/n , correlation is normally used to recover the data bits from the RX stream , going for large blocks , requires accurate time matching ,
this couple with multiple passes , where the signal is repeated , verbatim, slowly recovers the signal [ may be 2 dB per pass] , not having handled the detection in DSP , this limited in terms of AFC - has none , It responds to tones +/- of the selected qrg, to provide drift accommodation , the down side being strong carriers , desensitise the system , also requires high % recovery to function ..
Its a detection system , it detects signals , that was its original function, the issue's occur when , comparisons are made with free running data modes , one repeating issue is the 50% duty cycle of opera , to deliver the same energy as wspr-2 ,, opera needs to run for x2 the time , however as the coding efficiency is very high op2 and wspr2 have similar s/n ratings , with only half the supplied energy and single pass ,
APSK and QAM , I don't have the proof , only point to the concept being used , I assume provides the maximum pay load for the minimum bandwidth , the orthogonal aspect , is linked to carrier stacking in the B/W , as not to interfere with each other .. being FM , the side bands are exact replicas of the main and extend -/+ , where in mfsk , the tones are placed by the modem pattern generator , each only occupying a single frequency , being left to the dsp to determine where and what fills the pass band