K5CM FMT May 20, 2012 (9 PM CDT)


TX Frequencies

K5CM 20 - 14122002.327 with a reference of 14120000.000

One station, WB6BNQ, now has three "zero error" readings in a row. K7HIL follows with two in a row, with WA1ABI, and KD2BD making the "zero error" list this running.  All the readings were very close considering the delta is at 2 KHz. Having looked at some of the wave files it is evident that the signals are moving around some with respect to each other, but a 4 minute average still produces an almost zero error.  Take a look at the comments from K7HIL, and WB6BNQ. Their reported average error was was less than 100 micro hertz. Impressive. Also take the time to look at the pdf at the bottom of this report.

Do we move on to 20 KHz............

K5CM 20 Meters  (9:00 CDT)

Call Section Grid TX Frequency Measured Freq Error in Hz Error in PPX
WA1ABI RI FN41jp 14122002.327 14122002.327 0.000 0.00E+00 Measured the beat note using AM detection and added that to the published reference. Good signal(s) here. Tnx for the test Connie.
K7HIL AZ DM42ph 14122002.327 14122002.327 0.000 0.00E+00 Another learning experience for me on the high value of the reference signal. Method here is a little different; I calibrate the radio's CW pitch frequency with a known RF source, allowing me to provide a vary narrow RF bandwidth (50 Hz) to lower the SNR and eliminate QRM. Actual measured frequency tonight was 14,122,002.32703 Hz.
WB6BNQ SDG DM12 14122002.327 14122002.327 0.000 0.00E+00 Using AM method my recorded value resolved to 14,122,002.236935 and also receiving the mixed signal resolved to 14,122,002.326821.  So the two processes agreed with each other rather well.
KD2BD NNJ FN20XD 14122002.327 14122002.327 0.000 0.00E+00 A solid-state Hammarlund HQ-140-X receiver with AM detector was fed into a sound card and recorded along with a 1000 Hz tone from my homebrew WWVB frequency standard on individual audio channels (44100 Hz @ 16 bits).  The audio channels were separated and bandpass filtered using SoX software, and their individual frequencies measured using an oscilloscope triggered by a DDS-based function generator.  The frequency measurement was added to the published RF reference carrier frequency after compensation for soundcard sampling error, determined through identical measurement of the 1000 Hz standard, was made.  Results for the East/West transmission were submitted.  Lots of QSB and phase modulation was observed in the recovered audio.
NF6Z NV DM26 14122002.327 14122002.326 -0.001 7.08E-11 My first FMT.  Not really set up yet.  Received on a USRP (Universal Software Radio Peripheral) with no calibrated frequency reference.  I used the known signal to make a correction for the combined effect of the error in the USRP clock and doppler (about a part in 10000).  I used my own MATLAB software to estimate the frequency.
K6IQL KS EM28MQ 14122002.327 14122002.326 -0.001 7.08E-11 Did several things, but this input is based upon measurement of the FMT signal only. NS and EW both measured and NS solution used. Technique is standard audio beat between unknown and RB clocked DDS synthesizer received by 75A4/IC246 in AM mode. Audio output processed with SL and Excel. Sorted on amplitude and phase. Also made some other measurements which may be useful. If so I will provide by email. Good fun as always. Thanks Connie  - john k6iql
KA1BQP RI FN41 14122002.327 14122002.325 -0.002 1.42E-10 Used both the AM dectection method (EC/WC) and the conventional method (N/S). Massive amounts of doppler shift obsevered on the N/S signals. My reading is based on the AM mode. Tnx Connie. 
W6IHG VA FM09TB 14122002.327 14122002.325 -0.002 1.42E-10 E-W run S9+20 with slight flutter, slight drift in tones. N-S run only S6 with two definite clusters with more than 1 Hz between. About half the time both tones shifted together between them and the other half only one tone shifted. Flex5000a, HP3336 and SpectrumLab. I used both the delta between the two tones and also my older Delta Reference against the HP3336 signal. Unfortunately they don't agree so I just picked the E-W run as the best estimate, hi  Jerry
W6OQI LAX dm04vf 14122002.327 14122002.330 0.003 2.12E-10 For this test I put my Icom 756 Pro III in the AM mode with 3 KHz filter.  I compared this receiver output to the output of an HP3336b signal generator.  I set the HP3336b to 2000 Hz and started tuning the HP signal generator for a zero beat.  I could see slight phase shift between the pilot signal and the FMT signal, leading me to believe that the doppler is not quite the same for the two frequencies.
VE2IQ ON FN15nt 14122002.327 14122002.330 0.003 2.12E-10 Condx quite stable in Eastern Ontario.  Interesting differences between EW and NS beam headings. Tnx Connie for another enjoyable test.
W8XN MI EN82dg 14122002.327 14122002.331 0.004 2.83E-10 The ionosphere was clearly unsettled -- not surprising as the path was just on the night side of the grayline. The E-W run showed a +304 mHz Doppler shift when the data was averaged over the keydown period; the N-S run showed a -233 mHz shift. The peaks were really ragged indicating rapid change. Both directions were quite strong and the two results were within 2 mHz after correction with the reference, so I averaged them to come up with my guess.
AA7OO AZ DM43fe 14122002.327 14122002.322 -0.005 3.54E-10 No problem receiving E/W and N/W runs. New technique is what this series of FMTs is about with a side benefit of developing more consistent accuracy. The primary result reported used the E/W run averaging 5 delta cursor captures in real time. The variatio
K6OQK LAX DM04ud 14122002.327 14122002.320 -0.007 4.96E-10 I took advantage of a few free minutes to give this a last minute's try.  I tuned my ICOM 746PRO to 14,120 MHz and set it to AM.  I fed the resultant beat note of the two signals into SpectrumLab runing on my laptop using my JT65-hf homebrew interface.  The last three digits were all over the place due to propagation.  The first run started out strong and over a short time faded away to nothing and then slowly faded back in.  The second, and I think there was a third run, were at or below the S8 noise level. I could see them on SpectrumLab, but they were quite feeble.  Burt, K6OQK
W3JW VA FM17TN 14122002.327 14122002.318 -0.009 6.37E-10 Using SL (900 to 3100 screen width). SSB mode--- collected data with live delta cursor and export file configured to output the approximately 2KHz difference. The data are very "noisy"--- particularly with the beam N/S.
W4UK SC EM93an 14122002.327 14122002.288 -0.039 2.76E-09 Used the AM method to derive a 2KHz tone, measured that tone and added the reference freq. I like the idea of a two-tone test so we can compensate for Doppler, but I dislike having the tones near exact KHz multiples - makes it more difficult for me to measure absolute freqs since my reference generator is limited to exact KHz mults and I like to have them all in the recording at the same time. This means I have no precise measurement of the reference freq, just the delta. Thanks again for the exercise. Jerry W4UK

This from Michael, K7HIL and Bill, WB6BNQ

After completing all the analysis, Bill gave me (K7HIL) a call and we were on the same number (again). I had him send me his data and we looked at the Reference and Test signals and saw pretty good correlation. After spending a while, was able to get the vertical axis to show that the Doppler was very close between San Diego and Tucson.


PDF FILES/FMT 5-20 Doppler.pdf


This from John K6IQL

Just for the heck of it I measured about everything but the temperature
in the shack. I made a bunch of independent measurements of both the reference
and the FMT frequency by various methods. One thing I did not do was beat the two
carriers together in a AM receiver to derive the spacing between carriers. My present
setup does not easily support that and other methods at the same time. My interest is more
in the area of perfecting single frequency techniques as these have a way to go.
So having two frequencies to measure by single frequency techniques is bonus for me.
Thanks Connie!

The attached shows various results. I generally got the best data for the NS signal as I am
north of Connie. Also my submitted result used a vertically polarized antenna, though
the signal started out horizontal. All data was "as measured" - no corrections used - other
than clairvoyance.

Hopefully this info may be of use to someone.

-73 john k6iql

PDF FILES/20may12__Summary_K6IQL.pdf





Transmitting equipment used by K5CM for this FMT:

A HP Z3801 GPS frequency source clocks a HP-3336B and a PTS-250. The outputs are combined and feed to a 10 db transistor amp which drives a 12BY7 / pair 6146. Then to the final amplifier, a Dentron 2500 running 200 watts out.  Each signal was first set to 100 watts out. There are no mixers or any device in the chain of amplifiers that might impact the accuracy of the GPS frequency source.  As a reality check,  I periodically log the frequency source against WWV and other GPS referenced sources.  I monitor each transmission with a completely separate receive system to look for any instability or short term drift. 

I typically see considerably less than 1 mHz difference at 10 Mhz when comparing  two Z3801's or a Z3801 and a Thunderbolt, over a 10 to 10,000 second period. 



160 - Vertical

 80 - New Sloping dipole. Sloping down from 140'. (favors West, North, East)

 40 - Standard Dipole hanging between two 80 foot towers. (favors East, West)

 20 - 4el SteppIR at 70'


Current W6OQI FMT transmitting equipment for this FMT:



My Transmitting setup consists of a HP Z3801 clocking a HP 3336B which drives a Johnson Viking I transmitter, the output of which is attenuated by a 6 db attenuator, to then drive an Icom PW-1 amplifier to about 500 watts. The antenna is an inverted Vee which favors north and south from the Los Angeles area.







Note: Because I put the results up quickly, there will be mistakes, so don't hesitate to point them out quickly. 






back to main page