RF frequency measurement, measuring techniques and comparison to an SA. Accurate?

Hello,
I wonder how frequency measurements accuracy is actually supposed to be achieved on these microcontrollers or general purpose frequency counters, compared to a modern spectrum analyzer frequency measuring.

I mainly consider the close-in signal frequency measurement.
What I mean is this:

A modern spectrum analyzer, tuned in an unmodulated carrier with maximum resolution, can measure the frequency of the peak of the signal as well as the frequency at each specific point of the signal trace (using markers), where the signal does not peak (left and right signal sides).
But a microcontroller or other digital counter measures only the frequency of the peak of the signal (am I right?) and it cannot measure it's sides.

So how does the SA perform frequency measurement at any point of the signal trace, whereas the digital counter does not?

So how does the SA perform frequency measurement at any point of the signal trace, whereas the digital counter does not?

Click to expand...

The spectrum analyzer doesn't display a signal trace. It shows (possibly manipulated by averaging or peak hold filters) a spectral scan of the input signal. The frequency counter measures an average frequency by counting signal cycles over a specified gate time. Both measurement methods show different properties of the input signal. In case of a modulated or time-varying signal, both can give misleading or "wrong" results, also a SA if the settings aren't well considered.

A SA shows more information than a frequency counter. It can have a high precision reference frequency and respective accurate marker readings.

FvM said:

The spectrum analyzer doesn't display a signal trace. It shows (possibly manipulated by averaging or peak hold filters) a spectral scan of the input signal. The frequency counter measures an average frequency by counting signal cycles over a specified gate time. Both measurement methods show different properties of the input signal. In case of a modulated or time-varying signal, both can give misleading or "wrong" results, also a SA if the settings aren't well considered.

A SA shows more information than a frequency counter. It can have a high precision reference frequency and respective accurate marker readings.

Click to expand...

Yes, what I mean by "signal trace" is the top line of the amplitude of the signal, as it is scanned by the sweeper in a real rime SA, essentially the line you see on the screen. It is not the right term I used.

Ok, that is what I was thinking, a frequency counter measures the average of the signal, which should be near the peak of the signal for a symetrical signal (assumming an unmodulated carrier for the example, not SSB). In DSB it should give the carrier frequency, or repeatelly fluctuated between LSB and USB?

In comparison, the modern SA can measure the frequency of the markers precisely and these markers can exist on the slope of the carrier and not necessarily on the peak. How is this achieved?

You can analyze a modulated signal in time domain to know what the frequency counter result will be. Generally speaking, you can expect that it's pretty near the carrier frequency for sufficient gate time and the assumed symmetrical modulation schemes.

A SA is essentially a (dual- or triple mixing) superhet receiver, newer types may use zero-IF in last stage. If all involved LO and IF frequencies are crystal-accurate calibrated, you'll know the exact frequency of each x-position in spectral display.

FvM said:

You can analyze a modulated signal in time domain to know what the frequency counter result will be. Generally speaking, you can expect that it's pretty near the carrier frequency for sufficient gate time and the assumed symmetrical modulation schemes.

A SA is essentially a (dual- or triple mixing) superhet receiver, newer types may use zero-IF in last stage. If all involved LO and IF frequencies are crystal-accurate calibrated, you'll know the exact frequency of each x-position in spectral display.

Click to expand...

So the SA does not actually perform actual frequency measurement, it just measures the difference of a very accurate LO and IF to determine the marker frequency. Am I getting this right?

Perhaps, we should simply say, the SA is performing (spectrally resolved) level measurements rather than frequency measurements.

To "measure" a signal component frequency, you perform a peak search, either manually or automated.

worth adding that most frequency counters 'slice' the signal at a pre-determined threshold (trigger level) so only the larger signals get counted. Essentially they are digital instruments counting part of an analog signal, they don't count every 'twitch' of the waveform. A spectum analyzer is an analog instrument so it can display all levels from their noise floor to overload point and the sweeping of the spectrum along with the IF filtering allows investigation within a narrow frequency window. All the SA I have used do not actually measure frequency, they tell you the center frequency of their bandwidth at the marker point as calculated from their own local oscillator/mixer circuits.

Brian.

FvM said:

To "measure" a signal component frequency, you perform a peak search, either manually or automated.

Click to expand...

And this actual signal component frequency measurement is done by measuring the difference of this "peak search" (LO frequency at this peak) and the IF?

- - - Updated - - -

betwixt said:

All the SA I have used do not actually measure frequency, they tell you the center frequency of their bandwidth at the marker point as calculated from their own local oscillator/mixer circuits.
Brian.

Click to expand...

That perfectly makes sense to me.
They calculate their center frequency by the LO and IF difference and they "move" the RF waveform (actually the LO freq) as you change frequency. Hence the marker is only moved up/down, depended on the current signal level and does not ACTUALLY change frequency, regardless of what you see on the screen. It is the LO that changes frequency, so recalculation of the marker frequency is done.
I hope I am getting this right.

Yes, that's right. The peak marker may be fixed in the display ('sliding' the frequency below it) or may be adjusted manually or automatically to the position the software determines to be largest amplitude but the SA's own LO and filter offsets are used to show the frequency at that point.

Brian.

betwixt said:

Yes, that's right. The peak marker may be fixed in the display ('sliding' the frequency below it) or may be adjusted manually or automatically to the position the software determines to be largest amplitude but the SA's own LO and filter offsets are used to show the frequency at that point.
Brian.

Click to expand...

Thanks a lot!

Having worked on transmission equipment, counters have accuracies in the order of 1 part in 10^9, as accurate as SAs are the scan width against marker thickness, puts them in the ball park of "that component is 2.3 KHZ above the carrier" (for location and identification purposes). To get to a 10 HZ accuracy, with .3mm width display cursor, would give about 30 HZ/mm, 300 HZ/cm or division, so if your SA has a scan of 300 HZ/div and can stablely lock a carrier, you are home. This would be a much better SA then In have ever seen, because the 10 HZ refers to a 10 GHZ carrier, most 300 HZ/dv sweeps are only available on low frequency machines, where the frequency tolerance becomes .1 HZ, so you need 3 HZ/div sweep.
Frank