In the previous blog I explained the creation of complex RF digital waveforms on the 33622A generator. I thought that it would be nice to use these signals in a sort of real world scenario, and maybe push it a little bit further.


So, I devised a small experiment. 33622A will be the source of the signal, QAM64 modulated carrier at 49 MHz (and three others at frequencies below that), 5120 ksymbols/s symbol rate and channel bandwidth od 6,4 MHz. That signal is fed into a fiber optic node, specifically into the external return path input of the node. That signal is then fed trough a simulated 17 km of single mode fiber optic cable, using the Fabry Perot laser at 1310 nm wavelength.

At the other end signal is received using the exactly same fiber optic node, and RF signal is fed into a Promax Explorer HD analyzer.



From previous measurements expected CNR of this signal straight from the generator is somewhere around 40-45 dB.


Input to the fiber optic TX is set to -10 dBm.



And on the receiving side I got this:



If we take 40 dB of CNR as a reference, entire transmission path added only 1,3 dB of noise to the signal.


Now lets add some aditional impairment to the signal in a form of white noise. This noise is internally generated in the 33622A, modulation type sum, bandwidth 55 MHz, 5% modulation level.




Here are the measurements...



Here is the same setup, only with 10% noise level:




And with 20% noise level:




I have also taken shots of QAM constellation diagrams and added them together to quickly visualize noise effect on the signal integrity.