OFDM. COFDM: Rohde&Schwarz GN2130 digital voice modem

Rohde&Schwarz GN2130 digital voice modem

OFDM, Ch - 48, Sh - 56.25, Br - 44.4(4), QAM-16/PSK-4, k = 17/64

Author: SergUA6 

Band Width ~ 2700 Hz

Low Range ~ 360 Hz

Baud Rate 44.4(4) Hz

n-Ary (PSK/MPSK) Mixed QAM-16/PSK-4

Count of Carriers 48

Step between carriers 56.25 Hz

ACF Typical ~ 45 ms

RX mode  SSB

The record of GN2130 is here.

Sonorgams:

pic.1  General view

Diagrams:

pic.2 General quantity of the channels

pic.3 The constellation in the channel

R&S GN2130 - is the The voice modem, optionaly delivered to HF transceiver XK200 of R&S corporation. It is used for a speech transmission after conversions into a digital form, closing of the data by means of internal cryptoprocessor is possible. In this example, 48 channel OFDM signal is orginized, with standard enough parametres. 


Modulation in the channels is mixed QAM-16/PSK-4.  Features: it is necessary to mark presence of the auxiliary/pilot-symbols , in every second symbol through every one channel for synchronisation and customisation/fine tuning of the adaptive corrector the special signals are transmitted. Also there is a presence of the preamble with duration about 600 msec. Maximum common technical speed for QAM-16 is ~ 8533 bps.

Update SA to version 6.2.2.0/1

Update SA to version 6.2.2.0/1

In this version added the ability to copy the various fragments of the signal from WF module.

Copy fragments signal from the sonograms, can not operate sufficiently exactly. This is not a problems of SA, it is a fundamental limitation FFT.

The fact that the FFT, in general case , makes a transition from time domain into the frequency domain, and such a thing as time loses its meaning. Within the FFT block does not have time, and to say exactly where in that block is one or another component of the spectrum is impossible. This leads to the fact that selecting/separating out a segment of the signal at the sonograms, there is always uncertainty within the block FFT. Sometimes this quantity is large enough. Hundreds of milliseconds or few seconds and more, depending on size of block FFT.

Copying a fragments of the signals as waveform is much more accurate, but has own problems. Impossible to say exactly, from the image of waveform, exist/included a some frequency components in the selected segment or no.

Although the selection of segments from by waveform and is much more accurate, nevertheless, also not an accurate method. Accuracy is limited to one or two samples. In general, the problem is really accurate measurements and actions, is surprisingly/unexpectedly deep and multifaceted task.

Also introduce the possibility of transferring a fragment of the signal directly in the FSK-2 dem, and solves the problem of obtaining "clean" bit stream, in many difficult cases.

A small example of obtaining a "pure" bit stream from the rather complicated to demodulate the signal.






As usual, apart from updates, fixed bugs found by users. And including minor corrections to improve the work of the various modules and the whole program.


Good Luck!