Frequency measuring question

WA2DVU
Posts: 47
Joined: Wed Nov 01, 2017 3:27 pm
Location: Cape May, NJ

Frequency measuring question

Postby WA2DVU » Mon Dec 16, 2019 7:33 pm

I have been noticing some errors when measuring frequencies when using my 7000, phase display, and DSB function.

My 7000, and HP 3336 sig gen time bases are synced to a lucent ks 24361 gpsdo unit.

When I input a sig (5100 khz) from the generator the trace dot should not rotate as the sg and 7000 time bases are synced with the gps The dot does not move. Good!

Inputting 5000 khz the dot moves - changing the sig gen fx to 4,999,999.980 (-20 milliHz) the dot does not move.

Just wondering why some frequencies do not move and others do. To stop the rotation on the moving ones lowering the input fx by -20 milli hz works. I have reset database and reinstalled existing db and tried three other sig gens and they all play the same.

It is no big deal. I was just playing around with the gear and saw this problem. Again wondering why it was happening.

Part of my business is measuring frequencies for clients. I started doing this 40 years ago when broadcast stations had to have their fx measured each month to keep FCC happy. That is not needed any more but still have some clients that need it done.

My fx measuring rig is an HP 3336 level gen, 3586B selective level meter, Lucent GPSDO, and a tek 475 scope.

Also a member of the FMT user group. Gotta get a life.

Check out my call WA2DVU on qrz.com for some pics.

I would imagine the error is somewhere in the 7000 synth but again no problem with normal operation of the rig.

73, Bill WA2DVU
Cape May
WA2DVU
Posts: 47
Joined: Wed Nov 01, 2017 3:27 pm
Location: Cape May, NJ

Re: Frequency measuring question

Postby WA2DVU » Mon Dec 16, 2019 7:57 pm

If you want to play with fx measuring with your rig - look at a carrier on the phase scope and dsb function. If the trace is rotating slow enough to count how long it takes to do one revolution. Then compute the reciprocal of that time in seconds. That will give you the fx error in hz.

To compute the fx offset divide that reciprocal by the frequency have input to the rx.

Check out 5, 10, 15 mhz of WWV. We all thought we were receiving those exact fx at our antenna. At my QTH in Cape May WWV can be off by a HZ or so. No big deal for most of our ham radio activity but if we are putting a bus on the moon it could make a difference!

73,

Bill, WA2DVU.
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Frequency measuring question

Postby w-u-2-o » Mon Dec 16, 2019 8:11 pm

WA2DVU wrote:Just wondering why some frequencies do not move and others do. To stop the rotation on the moving ones lowering the input fx by -20 milli hz works. I have reset database and reinstalled existing db and tried three other sig gens and they all play the same.
I am betting this is an artifact of some rounding or quantization in the DSP code. Something like that would cause these discrete jumps in the results.

73,

Scott
User avatar
w-u-2-o
Posts: 5540
Joined: Fri Mar 10, 2017 1:47 pm

Re: Frequency measuring question

Postby w-u-2-o » Tue Dec 17, 2019 1:49 am

I corresponded with Warren Pratt about this. The software sends the frequency to the firmware in 1Hz increments. However, the complex oscillator in the firmware is derived from the 122.88MHz clock and cannot hit all those 1Hz frequency increments exactly. So that is probably what is accounting for the deterministic frequency errors.

Since the errors are deterministic, it should be possible to account for them in the client side software. However it's probably not really necessary. 20 milliHertz is not going to make a difference to any waveform I can think of, not even something like JT9.

73,

Scott
WA2DVU
Posts: 47
Joined: Wed Nov 01, 2017 3:27 pm
Location: Cape May, NJ

Re: Frequency measuring question

Postby WA2DVU » Tue Dec 17, 2019 1:34 pm

Hi Scott,

That makes sense - it is still amazing that the system can process 1 millisec stuff. The error does not affect any operations. I just wondered why it happens and tossed it to the gurus! Thanks again.

73,

Bill, WA2DVU
User avatar
DL2XY
Posts: 116
Joined: Sun Jul 30, 2017 9:47 am

Re: Frequency measuring question

Postby DL2XY » Tue Dec 17, 2019 6:23 pm

The NCO has a resolution of 32bit.
122.88 MHz / 2^32 = 28.610 mHz quantization.

73 Walter

Return to “Thetis”