Frequency Calibration...

I have struck a puzzling thing here with regard to frequency calibration with WSPR.

I had got into the habit from my old VK2ZTO QRPpp days of calibrating frequency via WWV.

The process I would follow would be to:-

1. First check the soundcard sample rate (usually within +/-100ppm for a good quality motherboard or +/-0.1Hz @ 1000Hz - except for software using non-standard sample rates). To check the soundcard sample rate I use the audio tones from WWV while receiving in AM mode. Depending on which TaF station you are listening you get 500Hz, 600Hz, 1000Hz, 1200Hz tones at various time during the hour. Use your receiving software to display these tones and check they are with +/-100ppm to the best resolution the software can provide.

2. Next check your receiver frequency calibration in SSB mode against a WWV (say 10MHz)station as close as possible in frequency to the intended operating frequency. Adjust your receiver dial frequency below the WWV frequency (assuming USB) until you get a tone in your display (say 1000Hz). If you chose 1000Hz as your tone and if your receiver was spot-on the dial frequency would be 10Mhz - 1000Hz = 9.999000MHz. However, most likely your receiver will be off. My FT847 rig I use to monitor the FT817 WSPR signal needs to be adjusted to 9.999009MHz to get a 1000Hz tone against the 10MHz WWV, i.e., the actual FT847 frequency setting is 0.9ppm below the dial setting. For the FT817 the dial setting is 9.999010Mhz for a 2Hz error (closest I can get with a minimum freq. step of 10Hz), so the FT817 is about 0.8ppm lower than the dial setting.

So where is the problem..???

Well, because WSPR uses an audio input range of 1400Hz - 1600Hz I had no WWV audio pip tones to check it - I assumed it was spot-on on my machines with respect to display the frequency. So I went straight to step 2. above and set my FT817 dial accordingly after using an offset of 1500Hz against 10MHz WWV.

The upshot of this is that if I net using WSPR and note the dial frequency on the FT817 and then net using Argo (for example - but Spectran is the same) there is a 17Hz discrepancy. I have checked Argo and Spectran against the pip tones on WWV and they are both within a fraction of a Hz at 1000Hz, so I have to assume that WSPR on my machines is off by 17Hz @ 1500Hz.

Problem is that I have searched and no-one else has seen this, so I must assume it is peculiar to my setup.

It would be interesting if someone else can check this.

Supply a tone (nominal 1500Hz) and vary its frequency until you can see it on the WSPR display and lined up with the centre of the WSPR display. On 30M with a dial frequency of 10.138700MHz in WSPR this should be lined up with 200. Then check that same tone using Spectran or Argo to see if they agree.

Would be very interested to see any results of such a test.

Of course if you have access to a known accurate 1500Hz audio tone so much the better.