Can track and record two signals simultaneously.  One can be the unknown remote signal: "Unk:".  The second can be a local signal generated from a precision signal generator: "Ref:".  If the reference signal is much more stable than the receiver, then drift and thermal effects observed on both the unknown and reference signals should be identical.  Post measurement processing will allow the user to remove those from the unknown frequency.

Setting the unknown and reference tracking points using

• waterfall cursor: SHIFT left mouse click for unknown track
• waterfall cursor: CONTROL left mouse click for reference track
  Above screen shot shows the unknown (blue), reference (yellow), and cursor tracks (3 lines) on the waterfall.
  
• Fine tune both using the respective frequency controls:

        

• Outer buttons: +/- 10.  Hertz
• Middle butons: +/- 1.0  Hertz
• Inner buttons:  +/- 0.1  Hertz
  

Each signal track is uniquely identified by color.  The track color is displayed to the right of the frequency controls.  The plot color and waterfall track bars are keyed to one another.  The signal plots show error from the assigned tracking points.

Signal tracking for each can be turned on and off using the respective Track button.

The averaged computed frequency is shown in the readout control.

The averaged signal amplitude in dBvp is displayed to the right of the frequency.

Each track can be independently cleared.

The controls "Interval", "T-scale", "Scale", and the button "Record" are common to both tracked signals.

• Interval: select the time interval between records in the data comma-separated-value file (format shown below).
• T-scale: select the number of minutes displayed in the plot panel.  This control will not effect the data history, only it's visibility.
• Scale: select the vertical scale of the plot panel.  This control does not effect the actual data, only it's display.
• Record: toggle the csv recording on / off.  Each new ON starts a new file that is date-time stamped to the nearest minute.  Filename aging is applied to the filename if more than a single on/off event occurs within the same minute.  Either or both Unknown and Reference "Track" must be enabled for any recording to take place. 
  
  Recording will synchronize the data lines with UTC.  Synchronization should occur with a second.  The indicator to the left of the recording button will change from white to dark red when data recording has begun.
  
  

The recording files are saved in the folder ~/.fldigi/fmt; C:\User\<login>\fldigi.files\fmt\

Example of recorded csv file opened using LibreCalc:

• Recorded file format as shown .  Recording is not initiated by waterfall selections as occurs in the analysis modem.
  
• The corrected column is computed: C5 - I5 etc.
  
• The average entry is computed: average(M:M)
• The standard deviation entry is computed: stdev(M:M)
  

A plot of the elapsed time versus unknown frequency produced:

FMT configuration:

Right click the  left most control on the status line FMT to open the configuration dialog to the FMT settings panel:

Adjust the various colors to suit your visual acuity.  The decoder converts the tracked signals to base band, much like a direct conversion receiver might do.

Enable "Thick plot lines" for a bolder plot grid and data lines.

Filters:

• The base band signal is then filtered using a digital FIR filter.  The LPF BW controls sets the bandwidth of the FIR filter in units of Hertz.
• The decoded signal base band signal is then filtered using a moving average filter in which a number of data points are averaged over a block defined by the time duration of the block.  The units of the Data Filter are in Seconds.

Receiver/Audio:

• Your receiver may not be adjustable to less than 1 Hz.  The Rx Offset adds a fine tuning adjustment to both tracked signals.  It does not physically change the transceiver tuning.  The Rx Offset can be adjusted in 1, 10 and 100 millihertz increments.
  
• You may already have used the WWV time tick modem to compensate for the audio codec (sound card) master oscillator error.  The Rx Codec PPM control mirrors the control on the sound card panel.  Audio codec compensation insures that the frequency measurement will be correct over the received audio bandwidth.  This is discussed further under audio code sample rate errors.
  

Waterfall:

• fldigi waterfall controls include a x1, x2 and x4 scale.  When changing between scales the program will try to always center at the current signal tracking point.  Select whether that will be at
  
• the unknown signal track,
• the reference signal track,
• the median point between the two tracks, or
• the default of 1500 Hz (no selection)

Wave file recording

• Record Audio - turn on and off wav file recording.  wav file is located in same fmt folder as data recordings.  New file is opened with a date-time stamped file name each time this function is enabled.  Current recording file name is displayed in the context control.  When enabled the "Sync to record" is deactivated.
• Sync to record - wav file recording turns on/off in sync with the data record button.  wav file and csv file will have identical date-time stamped filenames, one with a .csv and the other a .wav file name extension.  When enabled the "Record Audio" is deactivated.

Experimental controls:

• Pre-filter - a first order low pass filter applied to the base band audio output from the digital mixer.
• Cull Level - remove computed frequency values which fall outside of +/- the cull level from the averaged value.  The frequency discriminator used to track the signals is sensitive to noise bursts and deep QSB.  Culling large transients before averaging will improve the track line.
• Sample rate - the frequency discriminator / decoder can be set to a number of sample rates.  Suggest using the native sample rate of the audio codec.  For most transceiver built-in audio codecs the rate will be 44100.
  

Audio Codec sample rate errors

A sample rate clock error measurable in parts per million, ppm,  is a factor that needs to be considered for several modems, including FMT.  The error is visible when receiving an MFSK, IFKP or THOR image.  The error shows up as a slant in the received image.  For images it is usually the differential ppm between the sending and receiving codec.

On FMT measurements, the ppm error effects the error across the receiver passband.  Owners of a GPSDO can verify my measurement.  The transceiver is fixed at a known frequency; for the test it was 5.003.000 Hz.  The GPSDO is then slowly stepped from 5.003.400 to 5.005.800 Hz.

At each step the tracking error is measured.  The absolute error is interesting but not relevant to this measurement.  We are interested in the stability of the error across the passband.  Here are my results for three ppm settings, 0, +10, and +20 for the FT991A / transceiver internal codec.  The WWV time tick adjustment for ppm results in +10 ppm.

The WWV time tick ppm measurement and the passband error correlate almost exactly.  This tells me that I can compensate for the audio tracking points for any calibrated ppm setting, preferably the one that results in a zero error slope across the passband.