de W4AVO 2/28/2008

[ Note- What follows is a beginner's approach to the rather challenging task of off-the-air frequency measurement. The techniques are neither novel nor state-of-the-art. They are, however, succesuful in providing a measurement accuracy in the sub-hertz range with ordinary station equipment. ]

Modern transceivers have ten or one hz readouts and have or can be fitted with OCXOs or TCXOs with typical 1 hour stabilities of 0.1 hz or better. On the air stations such as WWV, WWVB, CHU, and broadcast staions with GPS lock give high accuracy calibration points at several points across the spectrum. Even low end laptops have sound cards that can be configured to do frequency tracking and measurement using DSP and FFT techniques.

With these tools, it is possible to calibrate your own station receiver and measure the frequency of off the air signals. How accurately? Well, not in the same class as laboratory grade setups with local Cesium or Rubidium standards or GPS disciplined XOs but a goal of within 0.25 hz most of the time and within 1.0 hz almost always seemed obtainable.

Can this be done with gear found in most ham stations? Here is my equipment and approach-

• Equipment-
  • IC-706Mk2g with the cr282 high stability crystal option
  • 1/8"stereo < > 1/8" mono audio patch cord ($0.25 at a hamfest)
  • 90's era IBM ThinkPad with Windows 95
  • Digipan 1.7 -a free PSK31 sound card program
• One time DSP cal-
  • Warm up computer for several hours
  • Feed receiver audio to notebook with patch cord
  • Set up Digipan to AFC and show received audio freq to 0.1 hz
  • Lock on and track WWV AM audio on the "waterfall"
  • Cal notebook DSP @ 440, 500, 600 hz. Tweak DSP dead on
• Radio cal-
  • Warm up radio for several hours
  • Set radio to 500 hz LSB above WWV on 2.5, 5, 10
  • Measure audio freq of indicated WWV carrier (500 +/-)
  • Determine radio error as a function of freq. Do this just before the measurement
• Measurement-
  • Place radio to a frequency LSB approx 500 hz above unknown
  • Measure audio freq- an average of readings over 30s or so is better than a spot measurement
  • Using dial freq, measured audio frequency, and receiver freq error - calculate unknown
• Example-
  • Just before the test the 706 reads high (i.e. is low) by 1.9hz @2.5, 3.8hz @5 and 7.7hz @10
    A straight line eyeballed thru a plot of the cal points estimates receiver error as 3.0 hz at 3.8 mhz
    (A linear regression by calculator works too. In this case it yields the same result: 0.05+0.7743*3.8)
  • dial frequency 3807500
  • audio 456.2
  • unknown = 3807500 - (456.2+3.0) = 3807040.8
    (the plus in 456.2+3.0 would be minus if the radio read low (i.e. was high))

The DSP/FFT program of choice of the FMT big guns is DL4YHF's Spectrum Lab. It is flexible and powerful but has a steep learning curve. So far I can't coax it to run reliably on my '90s vintage laptop. Here is a link to a good tutorial on using this program for frequency measurement by VE2AZX. I have downloaded the program and the tutorial. SL is high on my list of things to master. For now, I use Digipan because I have it and it runs on my computer.

Below is a screen shot that shows the mechanics of audio frequency measurement with Digipan 1.7. Frequency in Hz is across (...1000....2000....) and the display moves downward with time ("waterfall"). Noise from the receiver is blue/yellow and the receiver beat note is red. The diamond is placed by the cursor and when clicked locks on and tracks. The frequency readout is in a lower box. In this example, the tracked frequency is 1008.9 hz.

How did it do? In general it met my goals. A lot depended on the quality of the off the air signals that are mostly received via sky wave. The reference is off the air; noise, signal strength, and doppler affect the receiver calibration. These also affect the unknown measurement.

Here are the results of four Frequency Measurement Tests for which I submitted measurements. Not too shabby for a total expenditure of 25 cents.

FMT	Transmitted	Measured	Difference	Full Results
3/2008 K5CM **	1846169.245 Hz	1846169.330 Hz	+0.085 Hz	K5CM
	3569003.235 Hz	3569003.260 Hz	+0.025 Hz
	3567513.115 Hz	3567513.080 Hz	-0.035 Hz
	7077111.575 Hz	7077111.530 Hz	-0.045 Hz
	10115111.284 Hz	10115111.380 Hz	+0.097 Hz
	14067233.267 Hz	14067233.250 Hz	-0.017 Hz
	18077634.574 Hz	18077634.960 Hz	+0.386 Hz
1/2008 K5CM	3577521.200 Hz	3577521.400 Hz	+0.200 Hz	K5CM
	3579832.523 Hz	3579832.400 Hz	-0.123 Hz
	7077075.247 Hz	7077075.300 Hz	+0.053 Hz
	14057126.252 Hz	14057126.600 Hz	+0.348 Hz
4/2007 K5CM	3550451.587 Hz	3550451.420 Hz	-0.167 Hz	K5CM
	7055802.836 Hz	7055802.950 Hz	+0.114 Hz
11/2006 ARRL	1854317.500 Hz	1854317.700 Hz	+0.200 Hz	ARRL
	3587117.500 Hz	3587117.800 Hz	-0.300 Hz
	7038804.900 Hz *	7038806.400 Hz	-1.500 Hz

If you want to pursue this topic, look into these links:

• To read about and download Digipan
• To read about and download Spectrum Lab
• To examine the ARRL FMTs from W1AW
• Connie Marshall is the FMT guru. K5CM's frequent FMTs are accurate and well run.
• You are invited to join a Yahoo group of FMT affectionados (a.k.a. FMT-nuts)