Gain, overload and ghost signals: getting real dynamic range from RTL-SDR, SDRplay, Airspy and HackRF
- Replies 2
- Views 63
- Created
- Last Reply
Top Posters In This Topic
-
Mod_Mike 1 post
-
Mod_Chrissy 1 post
A better way to browse. Learn more.
A full-screen app on your home screen with push notifications, badges and more.
A common complaint with wideband SDRs goes like this: the waterfall looks crowded the moment a decent antenna is connected, the noise floor jumps, and signals appear in places where nothing should be. People often blame the software, but SDR#, SDRuno, GQRX, HDSDR and CubicSDR all display the same samples. Most of the time the cause is the front end running out of dynamic range, or local noise getting in through the antenna and the USB cable.
Here is a framework for discussion. Sensitivity and dynamic range are different things. A dongle can hear a very weak signal on a quiet band and still fall apart when a strong broadcast transmitter sits a few megahertz away. Dynamic range is limited by the ADC resolution, the analog front end (LNA, mixer, filtering) and how you set the gain. Roughly, each ADC bit adds about 6 dB of theoretical range. That is why an 8-bit device such as the RTL-SDR (RTL2832U) or HackRF One has much less headroom than a 12-bit Airspy R2 or a 14-bit SDRplay RSP-series receiver. Real performance also depends on the tuner and filters, so the bit count is only a rough guide.
The simplest test for overload is to change the gain and watch the signals. Reduce the RF or LNA gain by 10 dB. A real signal should drop by about 10 dB, and the noise floor should move with it. Third-order intermodulation products fall about 3 dB for every 1 dB of reduction in the offending signals, so they vanish much faster than real signals. If a "signal" appears or disappears out of proportion to the gain change, or mirrors a strong signal on the opposite side of the tuned frequency, suspect intermod or an image. An FM broadcast notch or a bandpass filter ahead of the dongle will often clean things up more than any software setting.
HF needs its own discussion. The RTL-SDR can do direct sampling on some models, such as the RTL-SDR Blog V3, which is simple but limited by 8-bit quantization and weak filtering. Without good preselection, strong broadcast stations overload the ADC quickly. An upconverter shifts HF into the range the tuner handles well. Its LO accuracy, filtering and spurs all matter, and you have to apply the correct offset in the software, otherwise every frequency readout is wrong by the LO value. Purpose-built HF receivers like the Airspy HF+ family or SDRplay units with HF input and notch filters generally cope better with crowded bands than a dongle with a converter.
Antenna and noise questions matter just as much. A broadband whip or discone attached to a dongle sees everything from local FM to pagers, and that can saturate the front end. An active loop or a tuned antenna for the band you care about brings its own filtering. Common-mode noise from the USB cable, switching power supplies, LED lighting and the PC itself can raise the floor by many dB. Ferrite chokes on the coax and USB cable, a short USB extension to move the dongle away from the computer, a powered hub, and a feedline run away from household wiring are all worth trying one at a time while watching the noise floor.
For discussion: how do you set up gain staging on your own receiver? Do you start with AGC and then switch to manual, or start low and raise the gain until the noise floor just lifts? Which filters (FM notch, bandpass, LPF for HF) have made the biggest difference, and for decoding modes like FT8, ADS-B or weather satellites, have you found a gain setting where decode rates drop at both ends?
Link to comment
https://www.hamradiobase.com/forums/topic/10892-gain-overload-and-ghost-signals-getting-real-dynamic-range-from-rtl-sdr-sdrplay-airspy-and-hackrf/Share on other sites