开云体育

Your expectation on harmonics are dependent on the cause.

For example input overload can do that, often you get other products like 3/2s
as you have two signals generally the source and then the internal
signal with some offset.? Yes, third harmonic divided by 2, higher order is
possible as well.

it is also critical to know the generators harmonic level, hence the reason
for the setup I use with filters.? Most of the HP generators are very clean but
that is not absolute, there is always some.

Allison
-----------------------------------------
Please post here, direct emails go to a dead letter box.

It slowly occurred to me that I am doing this still too complicated and there is an even simpler measurement approach. (if this is too much details for this group feel free to shut me up)

Using only one clean signal (harmonics really low) followed by a step attenuator it should be possible to observe the harmonics generated inside the tinySA and be able to predict everything what happens and calculate the IIP3 and SFDR
So I took a regular tinySA (no "special" this time) and started with a powerlevel calibrated signal at about -4dBm at about 9MHz.

This is how it looks with 0dB external attenuation



2nd harmonic at -54dBc and 3rd harmonic at -63dBc.
Calculating the IIP3 gives (-4 + (-4 - -63)/2) = +25dBm

Reducing the input with 10dB (using the external step attenuator) implies the 2nd harmonic should move to -64dBc and the 3rd to -83dBc and the IIP3 should stay the same



2nd order moves to -66dBc (seems ok) and 3rd order moves down 30dB so into the noise floor.?Let's pessimistically assume it is at -98dB so -84dBc. All looks good.
Calculating the IIP3 gives (-14 + ( -14 - -84)/2 ) = +21dBm which is a bit less so my assumption about the level of the 3rd harmonic may be too pessimistic.

Further reducing the input to -24dBm gives:



The RBW was reduced to 10kHz otherwise the 2nd harmonic would be invisible. Its now at -100dBm thus -76dBc which is indeed again 10dB better as it should be.
With the RBW at 30kHz this would have been invisible so the SFDR with an RBW of 30kHz seems to be 76dB.
In other words:?The internally generated intermodulation products of on or more tones at -24dBm or lower will no longer be visible at an RBW of 30kHz

So much to learn.

RBW means Resolution Bandwidth,e.g. the bandwidth of the narrowest filter before the power detector
This implies you can not see anything inside a good SSB transmission but you are able to see if there is a lot of over modulation

Thanks Allison for the advice.

As preparation for detailed performance measurements of a tinySA special using an high IIP3 mixer I?added two LC filters after the generators to remove the harmonics and did two measurements
One with both 9MH (f1) and 10MHZ(f2) at about 0dBm and another where I reduced f1 to -10dBm
If all is well all changes in harmonics should be completely predictable.

First with both at 0dBm



and the second with f1 at -10dBm



With the added filters all changes are now indeed as predicted
With f1 changing -10dB we get:
2 f1 changed -20dB (disappeared)
f1+f2 changed -10dB
3 f1 did not change
2f1+f2 changed -10dB
3f1 and 2f1+f2 boths disappeared.
2f1 - f2 and f1 - 3f2 also disappear
The calculated tinySA "special" IIP3 is about +27dBm

Now I think I have sufficient understanding and predictability to do some more detailed performance measurements.

AS the minimum RBW is 2.7kHz you can not "See" you modulation?
For that you best use a receiver and a PC with line in.

How close to a carrier I can get? Many Hams, educators and me would like to see and check their modulators. That is, can I see what does spectrum of? SSB with 1kHz modulation look like. Or 300Hz and closer.

Perfect, thanks.?

Minimum RBW is 2.7kHz

Clear, for the todo list if I have a spare moment....

On Thu, May 7, 2020 at 08:31 AM, <erik@...> wrote:

Would this line also induce some behavior?
Like a pass/fail message?
Or just a line?

==================================================
Eric, Just a line that you can set the value for and turn on and off.? Primarily used for quick visual indication of of how a measured signal compares against a set reference line.? I believe the line is usually labeled "ref" or "ref line" when it is displayed.? "limit line" might be another choice for a label to prevent confusion with reference level.

Some advanced analyzers do provide some type of pass/no pass or go/no go indication when the line is exceeded, but that seems a waste of memory space on the tinySA and might be better implemented in the software application if there was a user demand.

- Herb

Would this line also induce some behavior?
Like a pass/fail message?
Or just a line?

Me too.
Not now

I understand there needs to be a frequency limit somewhere, but just wondering why 960mHz.? I would love to see it go to maybe 1.5gHz.
Keith

Erik,
? Another feature I hope you can add to a future wish list is a menu selectable reference line.? It would be a line drawn across the length of the display at a level chosen by the user that could be turned on or off.

? It's a feature I have used frequently on other spectrum analyzers for quick visual pass/fail indication (limit line).

- Herb

As the internal flash can endure above 10k write cycles I assume using the internal flash with a warning not to store too often is OK.

Saving settings to one of the internal memory slots now also saves the stored trace and loading a saved setting will restore the stored trace.

This provides 4 memory slots to save traces?

If storage is a very important function, consider adding a WSON8-packaged SPI FLASH to the current PCB. some SPI connections need to be modified, and consider whether the current MCU has enough resources to drive the SPI FLASH.?A 1Gb (128MB) SPI NAND Flash costs about $1. Attached is the datasheet.

Hi Erik,

? ? ? ? ? ? Digital scope functionality sounds good and thanks for your trace storage addition to the list!

Tom, VA7TA

For testing the procedure I'e used is:

Both signals are low pass though a 7 or 9 section filter.? They must be
clean and the usual source had? HP part numbers.

Each signal has its own step attenuator.

I then combine them with a resistive mixer or other RLB (makes a good combinter).

Then with know quality gear measure levels.? its more to calibrate what
I'm putting in vs what I think I'm putting in.? Then verify the SFDR of
the test setup.? Its very easy at high level for one source to push
another source into distorted or mixed output.? Isolation and filters
are helpful in keeping that from being an issue.

With that SFDR, Compression and IP3 are determined.

Allison
-----------------------------------------
Please post here, direct emails go to a dead letter box.

An important parameter of a spectrum analyzer is the spur free dynamic range (SFDR).
This is the range in dB between the noise floor and the strongest signals that can be used without generating visible mixing products inside the spectrum analyzer.
If the IIP3 is known you can calculate the SFDR
As I have looked previously?at IIP3 I thought it would be interesting to see a practical example what staying within the SFDR implies.

As a test I'm using two somewhat clean input signals, one of 15MHz



and one of 9MHz (not so clean, see the small spur at 30.5MHz)



If you add these signal in a way they do not interact and feed them to the input of the spectrum analyzer you should only see the pure sum of the two spectra above.?

So first the combined spectrum with 0dB attenuation.




There are two signals that should not exist. The first at 24MHz (the sum of 9MHz and 15MHz) at -80dB and a second at 39MHz (2 times 15MHz + 9MHz) at -86dB . One would expect something at 33MHz (15MHz + two times 9MHz) but this one is not visible.?
In order to investigate if these mixing products are generated in the spectrum analyzer or in the generators (due to lack of isolation) I did two extra measurement.

First with 5dB increase in attenuation in the spectrum analyzer?



The signals that should not change did not change (within the +/1dB measurement accuracy) . The 24MHz mixing product went down with about 1dB and the 39MHz mixing product disappeared.
This seems to suggest the generator delivering the 9MHz signal is contributing to the mixing because with a 5dB increase in attenuation the mixing products both should have disappeared.
But the positive conclusion is that the tinySA will give you a SFDR of at least 70dB at a RBW of 30kHz.

Increasing the attenuation to 10dB gave this picture.



?The signal at 24MHz went down with another 1dB so this again suggests I may have to find a better way to mix the output of the two generators.

I welcome a critical review of my conclusions by the esteemed audience of the group.