Nick:

You can do it this way, but it requires you to totally understand the
mathematics and granularity of ALL of the frequency sources and
synthesizers in the superhet receiver.
And if there are any audio soundcards or sampling devices involved, the
specifications and origin of the sampling clocks.  If it is the USB clock
in your computer, you can be pretty much screwed right there.

These are generally not published by the manufacturers, so you will need a
friend at the manufacturer to provide you this information. Or be ready to
do a lot of reverse engineering.

The typical RF frequency measurement system for off-the-air measurement
only uses the superhet for signal reception, then adds a precision (GPSDO
derived) local oscillator/signal source for which you totally understand
the mathematics and granularity, because you built it yourself, and an
audio counter that allows you to measure the audio difference between the
signal to be measured, and the local precision reference that you inject.
In this system, the absolute frequency of the superhet receiver is a "don't
care".  You just want to measure the difference between the signal to be
measured and your local precision reference.

--- Graham

==

On Mon, Jun 20, 2016 at 9:51 AM, Nick Sayer via time-nuts <
time-nuts@febo.com> wrote:

Never tried it but a Selective Level Meter aka HP 3586A/B/C ?

On Mon, Jun 20, 2016 at 7:51 AM, Nick Sayer via time-nuts
time-nuts@febo.com wrote:

You need to be able to measure frequency accurately in the milli-Hertz
range to be competitive in the frequency measuring contests.

I doubt the Selective Voltmeters have that level of resolution. I think
they 'only' read to 0.1 Hz.

--- Graham

On Mon, Jun 20, 2016 at 11:52 AM, Pete Lancashire pete@petelancashire.com
wrote:

On 6/20/16 7:51 AM, Nick Sayer via time-nuts wrote:

Most of the folks doing the FMT these days use some sort of audio
spectrum analyzer program and estimate the frequency using that.

The signal isn't very high SNR (unless you're in Newington and they are
radiating from W1AW<grin>) - I'm not sure a narrow band filter followed
by a counter would be the best way to go.

You need a good reference to calibrate the frequency of your receiver -
depending on its design, you might just be able to measure your 10MHz
reference to figure out the offset for your receiver, and if it's stable
over the time required, you're good to go.

The ionospheric Doppler will spread the signal a few tenths of a Hz, so
getting millihertz is more random luck of the draw.

I note also that the last ARRL FMT ran at 10PM EDT with a transmitter in
California (where it was 7PM).. this is a particularly BAD time of day
to do the test, because the ionosphere is changing effective height and
attenuation so it greatly penalizes folks who are relying on skywave
propagation.

You can practice using WWV/WWVH, by the way.. nice AM signal with a good
carrier.

To echo the point that Jim Lux made in another post...

I'm not sure how those of us who live an appreciable distance from the transmitters can expect to get mili hertz accuracy  in light of the HF Doppler shift over long distance paths with restorting to tactics such as simultaneously monitoring another transmitter that has a precisely known frequency and is located near the target transmitter to determine what the Doppler shift is at a given point of time.  (I'm not 100 percent sure this approach would work but it seems viable to me at first glance.)

I've spent some quality time monitoring WWV, and have read various papers and I am quite convinced that HF Doppler shift is a real issue at times.

All the best
Mark S
VE7AFZ

Sent from my iPhone

On 6/20/16 10:10 AM, Graham / KE9H wrote:

And realistically, since the millihertz digits are pretty random from
ionospheric propagation, it's sort of a lottery at that point.

Hi

The cheaters way is to simply use a fully synthesized radio tied to a known
reference frequency. Feed the output into one channel of a sound card. Feed
the other channel of the sound card with a known frequency tone. Post process
it to death with your choice of FFT programs.

Another approach is to use a synthesized generator. Feed an appropriate level tone
into the antenna input of the radio through a combiner. Set it so that both it and
the FMT signal are inside your passband. Some care should be taken to see that
you do not radiate your test tone.

If you are going to rig up something special … just do a SDR of some sort. The post processing
is all the same once it gets to audio. The noise and fading are such that there is
essentially no way to directly read the frequency. That was true back when I was within
ground wave for the transmissions. It’s even more true if you don’t live next door to them
RF wise.

Bob

Yeah, that’s what I’ve got in mind. Both the synthesized tuning LO and the second LO would be derived from an external 10 MHz standard. My thought is to make a 10 kHz passband tunable in 5 kHz steps. Now, designing a receiver may be biting off more than I can chew, but that’s what I was going to try.

Well, using the same 10 MHz standard to make a 1 or 5 kHz tone (or both) should be easy. From what you and others have said, it sounds like trying to use a TIA to read the audio frequency is a non-starter.

Most of the folks doing the FMT these days use some sort of audio
spectrum analyzer program and estimate the frequency using that.

Or use the audio spectrum analyzer to measure the difference between
the frequency being measured and the precision reference. You are
correct, it is usually not a classic counter.

You need a good reference to calibrate the frequency of your receiver -
depending on its design, you might just be able to measure your 10MHz
reference
to figure out the offset for your receiver, and if it's stable over the
time required, you're good to go.

Well, even if the synthesizer is GPSDO referenced and locked, a digital
synthesizer
will have a minimum step size or resolution that it can generate.
You might not be able to hear it with your ear, but it is there.
For instance a DDS based synthesizer with a 32 bit tuning word and a 200 MHz
sampling clock will have a step size of 46 milliHertz.
So when you enter a decimal frequency into the synthesizer, you get the
closest frequency it can generate, which can have an error of up to +/- 23
milliHertz. And the actual error versus frequency entered breaks into sort
of
a Moire pattern if you plot it.  The errors are deterministic, but a user
normally does not have the
information to figure them out.

The ionospheric Doppler will spread the signal a few tenths of a Hz,  YES.

so getting millihertz is more random luck of the draw. NO,
it is all about how good your averaging method/strategy is, over the period
of the test measurement.
A lot of the Doppler error can be averaged out.
Some of it is an actual net vertical movement of the reflection point, as
you said, worst at sunrise/sunset.

Look at the scores/accuracy for some of the recent frequency measurement
tests.

--- Graham

==

On Mon, Jun 20, 2016 at 12:16 PM, jimlux jimlux@earthlink.net wrote: