To receive radio signals you need an efficient antenna placed at a good location ideally with an unobstructed view of the horizon.
Our environment is full of E-M (electro-magnetic) noise. One very obvious source is the CPU and other digital electronics.
As signals we seek to receive are very weak, to maximize our chances we must not let the noise enter the receiver input.
The obvious way the noise can enter the receiver is when you place the antenna close to the source of this noise.
The less obvious way is when the antenna is not well balanced and all its cabling becomes part of the antenna.
To give an example: if you place a 1/4 wave stick somewhere without a proper ground, the coax the suns to the sticks becomes part of the antenna.
It will thus as well receive whatever signal or noise, for example the noise radiated by CPU or USB transmissions.
Good ground for the 1/4 wave, ferrite cores on the coax can help to minimize this type of noise.
This article may give you an idea
To receive glider signals we use cheap SDR (Software Defined Radio) receivers - actually these are hacked DVB-T USB dongles.
You tell them frequency (24-1700MHz) and bandwidth (1-3MHz) you wish to listen to and they send you back raw digitized RF data in a form of 8-bit I/Q samples.
The processing that follows is all done in software which gives us great flexibility but puts lot of demand on the CPU.
The other aspect is the dynamic range: we only have 8-bit data, thus our receiver is prone to overload by strong signals.
These can be GSM on 900MHz band and TV on the UHF band. The best way to avoid such problems is to filter them out right at the antenna.
Here the coax collinear design has the great advantage of being narrow-band: about 5-10MHz.
The frequency stability of the SDR radio is determined by its crystal. As we use cheap USB dongles the crystals are standard ones thus we should expect precision within 50-100ppm. For the most frequently used big black dongles with the R820T receiver we have seen the crystals to be 40-90ppm off.
This is quite an important factor: at 868MHz 50ppm is 43kHz and the nRF905 signals have +/-50kHz FSK deviation.
Crystal's lack of precision can be corrected, you only need to know how much it is off. Currently we measure the crystal by checking the receiver against the GSM signals which normally are precise to at least 0.1ppm level. The program which does it is gsm_scan which scans the GSM 900MHz, picks up the broadcast channels (BCH) and their frequency bursts (FB) and measures the effective frequency error of the receiver.