RTL-SDR FIR filters

FlightAware ADS-B Flight Tracking
22

I ran another comparison between a new RTL-SDR module (the one in the black case) and an old one (silver case), both v3 versions:

Black case:
  -35.5 dBFS noise power
  -13.9 dBFS mean signal power
   -1.7 dBFS peak signal power
     69 messages with signal power above -3dBFS
Decoder:
      8690 total usable messages
        2243 DF0 messages
         930 DF4 messages
          59 DF5 messages
        1761 DF11 messages
         148 DF16 messages
        3524 DF17 messages
          20 DF20 messages
           5 DF21 messages

Silver case:
  -37.1 dBFS noise power
  -15.4 dBFS mean signal power
   -2.0 dBFS peak signal power
     32 messages with signal power above -3dBFS
Decoder:
      7949 total usable messages
        2080 DF0 messages
         856 DF4 messages
          49 DF5 messages
        1564 DF11 messages
         134 DF16 messages
        3245 DF17 messages
          16 DF20 messages
           5 DF21 messages

Again about a 9% improvement for the newer RTL-SDR module compared to the old one. Also in this test the newer module has higher signal levels.

Conclusion: If you want to improve performance in an RTL-SDR based system, don’t bother trying to fiddle with FIR filter coefficients, just buy a new RTL-SDR dongle! This kind of goes along with another observation – I have the distinct impression that my system worked better a year or so ago, Could it be that these dongles deteriorate with time? If so, I would suspect the analog part – the tuner chip and surrounding components.

1 Like
23

No, it is just about that they are not all equal. My better dongle is older, has been more used, and spent most of that time outside on the balcony.

24

Ok, that makes sense. If that’s the case it also points to the analog part being suspect. Digital doesn’t have that property. I think the analog filter characteristics are in part determined by components external to the tuner chip. Could be the variability in those components.

Another possibility for the differences in dongles is the frequency stability. Ever notice how when you run rtl_test, it says PLL not locked? Every one says ignore that, it doesn’t matter. However, what is the PLL doing – is the PLL frequency wandering about? I would think that could be a problem.