Signed and notarised. Download, unzip, drag to Applications.
Fixed
- The FM noise blanker was blanking 8 samples in every 9 on a clean signal. A run limit that reset on its own escape sample, and a level reference that could never recover from a step change. It was on by default and affected every radio, so a strong local station arrived distorted with no pilot lock and no RDS. Confirmed fixed on air.
- CEQ was amplifying multipath rather than removing it. The equaliser drove the composite to a fixed level instead of flattening it; at real receiver amplitudes that made a reflection roughly three times worse than leaving it alone. It is now level-independent and behaves identically on any radio.
- Typing a frequency could leave a station distorted where stepping to it did not — the same blanker fault, armed by the larger level step a typed tune produces.
- RadioDNS station logos that would not resolve now do (a redirect was not being followed).
- A frozen SDRplay gain readout now offers the reset instead of holding silently, and the advanced RDS panel no longer shows two "initialising" chips.
New
- Composite eye in the Advanced RDS panel — the FM composite in time, triggered on the recovered 19 kHz pilot rather than an edge detector, with the audio filtered out so the pilot, stereo sidebands and RDS show as a braid. It reveals things nothing else on the panel does: composite peak behaviour, and whether a marginal signal is nearly locking rather than simply absent.
- Automatic Direct Sampling for HF — an owner setting beside the converter, off by default, for dongles that reach HF by direct sampling. It warns (without blocking) on an RTL-SDR Blog V4, which has its own up-converter and does not want it, and tells listeners when the mode changes.