I am starting my journey into the world of Frequency Measuring Tests. I decided to use my receiver in AM mode and compare the unknown frequency to a known frequency obtained from a HP3335A (locked to a Rubidium source) that is set to produce a beat note around 300 Hz.
That AM beat-tone approach takes out any impact of the receiver in the process. However the soundcard being used to measure the audio frequency (with Spectrum Labs software) is subject to error. In my case a precise 300 Hz tone was reading low by about 39.7 ppm. Not horrible for a standard crystal source, but still a source of error. After adjusting the 48000 sample per second rate to 48001.904 sps, the audio frequency was spot on.
The source of the precise 300 Hz tone was another AM beat-tone. I used the 10 MHz output of the Rb source fed through 80 dB of attenuation to a BNC-T on my radio. The other side of the T was connected to my HP3335A set to 10.000300 000 MHz at a level of -80 dBm. Those two carriers produce a very precise 300 Hz frequency.
I turned the PC and rig off overnight, and rechecked things immediately following power on the next day. The worst case error was less than 50 microhertz (300.000 050) and it improved over time.