First FMT with "new" 6700 is a success

George O'Brien

I just read the results of the ARRL November FMT. I used my "new for me" Flex 6700 and got within 0.04 Hertz of the official frequency on the two transmissions that I could receive. The GPS disciplined oscillator gets most of the credit, however, FLDigi's Freq Analysis mode + using two other slice receivers to monitor WWV on 2.5 MHz and 5 MHz to get an estimate of the Doppler shift (in Spectrum Lab) made the difference. This rig is a technical "tour de force". Thanks to the engineers and software developers that made it possible.

Alex_KM5YT

I have urged Flex to do 2 things to extend the capabilities GPSDO:  use a chipset that handles all current GNSSs and add a rubidium clock on-board.  Tiny packages exist for both.

Army Curtis

Congratulations George. Well done. I also used my 6700 for the first time in the November FMT after having used my 5000 in several previous runs. Can you share more details of your setup that allowed determining the carrier frequency to more than 1 decimal place please? And a really dumb question if I may... where are the results posted of the November FMT? Many thanks.

Paul Christensen, W9AC

George, I would sure like to hear more about the test setup and how you accurately resolved to less than 1 Hz (the smallest Flex tuning step). 

Using a GPS-DO connected to the Flex, I had thought about using the CW sidetone and running it into one scope channel.  Then, take the receiver audio and route it to another channel.  Set the scope for X-Y Lissajous mode and get to within the closest 1 Hz in SSDR.  Look at the remaining phase difference between channels and compute the remaining frequency shift.  Of course, that requires a highly accurate CW sidetone and I'm not sure if Flex divides it down in any way from the internal or external time base. 

Sounds like FLDigi does much of this?  Since each PC is subject to frequency error, I'm guessing that there's a calibrate routine in FLDigi.  I'll need to study the document files more carefully.

Paul, W9AC

George O'Brien

Hi Alex,
I could see the theoretical advantages of having other GPS systems to further increase the accuracy and to ensure improved reliability through redundancy, however, I'm not clear on what the rubidium clock would achieve, except if the
radio were in a mobile environment.  Could you explain further?  I like learning, especially from others.

George O'Brien

Thank you, although my part was just putting together a logical chain of measure as we physicists call it.  Here is a brief explanation:  1. Used slice A set up in USB with the receive filter set at 500 Hz offset and 50 Hz wide to receive the unknown carrier  2. Used FLDigi in Frequency Analysis op mode to detect the peak.  Note: It interpolates many measurements received from the DAX stream to give a fitted frequency value to 0.001 Hz (1 mHz).
3. Furthermore, captured the values in default "analysis" file using FLDigi 
4. Used another slice set up the same way but this DAX stream feeding Spectrum Lab [free software and very powerful] and monitored known WWV at 5 MHz and 2.5 MHz to measure "errors" of Doppler shift and sound card drift [also 1 mHz resolution]
5. Because I didn't have time to configure Spectrum Lab for automated capture, I manually recorded the peak value frequency from Spectrum Lab.
6. Analysis - Used Excel worksheet to calculate mean value of errors and applied to the mean frequency value of FLDigi "analysis" file.

As they say in Paris, voila!  It was fun and a pleasant surprise to be so close.

George O'Brien

Hi Paul,
I've enjoyed reading your posts about CWX and QSK.  I'm also a CW buff.

Rather than repost about the setup, please read my reply to Army Curtis.

Now, about the errors, the basic frequency measurement of the GPSDO had an uncertainty of 1.19e-10 at the time.  At the highest frequency for this FMT of 7 MHz, that contributed only ~ 8.4e-4 Hz or 0.0008 Hz.  Now, I've read that Doppler shifting in the HF range contributes about 0.13 Hz of uncertainty, so I had to find a way of estimating this.  Furthermore, as you rightly noted, the sound card would be another, unknown source of error.  So, I was lazy and lumped all these errors together by monitoring a known frequencies from WWV.  I chose the 2.5 MHz and 5 MHz signals assuming that the Doppler shift would be close to those of the FMT carriers at 3.5 MHz and 7 MHz.  This was my first ARRL FMT.

I hope this helps.  BTW, I'm just a run-of-the-mill guy who's been a ham since 1969 when I was a freshman in college. Got my novice call of WN4PDQ back then. :-)

George O'Brien

Go to b4h.net and click on November FMT Results

Alex_KM5YT

Rubidium standard always handy if satellite view patchy; cities (or mobile) - but might help if there were regional outages, jamming/EW due to conflicts (they are essentially military systems), or satellites damaged by space debris/solar flares.  

George O'Brien

Okay, gotcha now. Thanks for the explanation. Still, I don't know that John Q Ham would want to pay the added cost for these features. Even as options, the development costs might never be recouped by FRS.

Alex_KM5YT

My thoughts were that even my (Samsung) tablet supports GLONASS, and adding
GNSSs Galileo and BAIDU will be (are?) standard fare in the receiver chipsets. I
bought a rubidium standard module on ebay for ~$150 to play with, and there are
very small packages now.  The key to low pricing is their inclusion - as I understand it - in cellular systems.  I don't see a big expense - and it logically 'completes' the xxxDO. John Q Ham won't bother with the xxxDO anyway.