Frequency testing with GPSDO

I use quite an old Yaesu FT897, getting on for 16 years old. Not old enough to be venerated as vintage, but old enough to be getting on for decrepit and occasionally in need of work, and on yesterday’s 10m net we struggled with the nominal net frequency of 28.405MHz, requiring frequent use of the clarifier. I was under the impression that the FT897 was within +/-50Hz of frequency, but the experience made me wonder. It is the task of net control to stay on frequency and not shift to track callers, but it’s also worth asking if it is the right frequency.

I have a 10MHz GPS disciplined oscillator, a Chinese product from ebay. Many frequency meters can take a 10MHz input to use a more accurate reference. NASA have an article on the stability of GPS. Long story short, it is good enough¹ for amateur radio frequency calibration. More on GPSDOs and ham radio here.

Initial test on Weston SDR

1kHz Tuning tone received at Weston SDR

For an intial test I looked for where a 1kHz tuning tone appeared on Weston SDR. The graduations are 100Hz apart at this scale, I would say I was within 50Hz of the 28.406 nominal frequency (28.405 suppressed carrier plus 1kHz audio tone).

Sources of potential error are, of course, Weston SDR’s frequency scale, and to a much lesser extent the crystal in the sound card. I have no idea what the frequency accuracy of Weston’s KiwiSDR is, the sound card crystal should be within 100ppm. Even a really nasty crystal should be good to 1000ppm, so 1kHz ahould be between 999 Hz and 1001 Hz even on a rubbish sound card. I am using a G4ZLP USB audio interface into a Raspberry Pi, the ZLP sound card hopefully aims higher than that. So I will ignore the sound card crystal, since 1Hz is low relative to the 50Hz nominal requirement for acceptable intelligibility.

Listening to the 10MHz output of the GPSDO

I receive 10MHz at 10,000,015 Hz

Adding together my dial frequency of 9.999 MHz with a USB audio frequency of 0.001015 MHz (1kHz) I receive 10MHz at 10,000,015 Hz, so I am tuned about 15Hz low at 10MHz.

Listening to the second harmonic of the GPSDO at 20MHz

I receive 20MHz at 20,000,028 Hz

Adding together my dial frequency of 19.999 MHz with a USB audio frequency of 0.001028 MHz (1kHz) I receive 20MHz at 20,000,028 Hz, so I am tuned about 28Hz low at 20MHz.

Listening to the third harmonic of the GPSDO at 30MHz

I receive 30MHz at 20,000,028 Hz

Adding together my dial frequency of 29.999 MHz with a USB audio frequency of 0.001044 MHz (1kHz) I receive 30MHz at 30,000,044 Hz, so I am tuned about 44Hz low at 30MHz.

In conclusion the frequency was within 50Hz of nominal spec which is acceptable. I haven’t tracked the error over temperature or time, but I would say that Yaesu did well, particularly as this is the old version of this rig which does not have a TCXO. It is good enough for voice and I can at least claim that net control was on frequency in the 10m net ;)

Weston SDR are vindicated, and they correctly showed that I am transmitting about 40Hz low, so I would say their frequency scale is spot on.

1. I was able to hear the frequency synthesiser PLL locking as a slight tremble on changing frequency step, after which the tone sounded stable to my ears, and looked stable on the waterfall, eliminating significant instability in the radio or GPSDO. ↩