This from Jeff, W3JW
For the radio only FMT, I used both receivers within the IC-7800. All oscillators, for both receivers, are phase locked to the 10 MHz internal reference. Since this reference is an OCXO, I choose to do my basic system calibration in the middle of the daylight hours on the day of the FMT when doppler spread, shift and multipath effects are minimal. I use the lowest useable frequency transmission of both WWV and CHU. You could make a case for performing the calibration just before/after the test over a path ""similar"" to the actual test path, as an attempt to mitigate doppler shift errors. I enable the internal calibrator and set the sub receiver to produce a 900 Hz audio output tone. The operating mode of this receiver can be CW, USB or LSB. The choice of 900 Hz is my personal preference, but is selectable by the operator. In my case 900 Hz works well for my ears when comparing two tones. It is also the highest CW pitch tone (phase locked to the reference oscillator) that can be selected in the CW mode. The main receiver is set to operate in the CW mode with the CW pitch set to 900 Hz.The main receiver is operated with frequency step size equal 1 Hz selected. I operate the main receiver BW at 1.2 KHz when initially acquiring the test signal then reduce the BW to 250 Hz for measurement. Tune the main receiver dial so that you hear the lowest beat frequency between the audio out tones of the two receivers. I use a set of headphones with main receiver to the left phone and sub receiver to right phone. You can use one (or two) speakers also but the beat seems (to my ears) to be more pronounced with headphones. The main receiver dial frequency (plus or minus the fraction determined by the beats) is the frequency of the test signal (no doppler or multipath correction). Be sure to save this frequency to a receiver memory or write it down (or both!). Count the number of beats between the two audio tones at the dial frequency determined above. I recommend counting the number of beats in a 10 second interval; then each beat represents a 0.1 Hz frequency increment. Then move the main dial frequency 1 Hz higher in freq and record the beats in 10 sec. Repeat for two Hz above the ""best"" dial frequency. Then move to 1 Hz below the ""best"" dial frequency and record the beats in 10 seconds. Repeat for dial set 2 Hz below the ""best"" frequency. You now have enough information to determine whether the actual freq is above or below the main receiver dial indication and the beat values tell you by how much (to the nearest 0.1 Hz) plus some redundant ""check data"". Keep in mind that the more beats per ten seconds that you hear, the further away (either up or down) you are from the actual frequency. If the test signal is available for longer periods of time, you could count beats over longer periods (anybody for 100 sec?) for improved precision.","01/11/2013","01:59:28 PM"