Hello JM, > In order to quantify my abilty to receive weather satellite > images using DVB-T running with GNURadio, > I have setup a very simple experiment as follows: > * tune a frequency synthesizer to 137 MHz, FM modulation, > 5 kHz deviation and 2400 Hz FM frequency > * frequency synthesizer output power setup to -91 or -101 dBm > (same setting for all receivers, 2 sets of measurements) > * connect the output of the synthesizer to the input of > the DVB-T (in all cases, a new BNC connector was solder instead > of the original 75-ohm antenna connector), So far very good:-)
> * plot the RF signal FFT and the FFT of a WBFM demodulator > block as found in GNURadio (always same settings: 30 dB IF > gain and 30 dB RF gain) Hmmm, Dongles are different. I suggest that you compare them at maximum gain which is different for different dongles. > * assess the receiver quality by the signal to noise ratio of the FM output. This is a very good method - but it is necessary to place the frequency right. I can see from your graphs that the signal is at the passband centre where you have a spur and 1/F noise. Repeat and place the FM modulated signal signal halfway between the passband center and the passband edge. Your current results are not valid because of the center spur which is within your passband. > Is this a reasonable way of assessing a receiver sensitivity ? > Is there some sort of well defined "protocol for > assessing a receiver quality for small signal detection ? What you did is much better than the standard procedures (provided you do it right with respect to the center spur (and any other spur) You will get more understandable data if you attenuate your test signal until you get the same S/N for the detected FM audio for all dongles. At some point you will find that S/N changes by something like 3 dB for a 1 dB change of the input signal. At the point where S/N vs input signal is steepest you get the best accuracy. Regards Leif
