SDRplay RSP2 spectra

[rhb]

A small experiment using and RSP2 with SDRuno, HDSDR & SDRconsole v3 beta.

Setup is a BG7TBL noise generator, a B&K 2025 function generator, Rigol DS1102E & the RSP2.

I've got a tee on the front of the RSP2 with a tap to the scope and feed to the RSP2 with a -40dB attenuator on the other end of the tee.

I've tried to duplicate the SDR settings as closely as I can figure out for the 3 packages.

Test cases are:

No source, noise source, function generator with 625KHz sine wave output.

[rhb]

Here's the result with the noise generator running at about 12.1V from a linear supply.

[rhb]

Here's the function generator displays.

I have no way to tell which of these is better.  They are far more different than I expected.  With the signal source disconnected, the Pi attenuator is effectively a 60 ohm input termination.

I'm not very experienced with any of the software packages, so finding what settings are available and setting them is a good bit of a challenge.  Is anyone familiar enough with this setup to suggest causes for the differences?

[hendorog]

Can you draw us a picture mate, this not making sense to me:

Setup is a BG7TBL noise generator, a B&K 2025 function generator, Rigol DS1102E & the RSP2.
I've got a tee on the front of the RSP2 with a tap to the scope and feed to the RSP2 with a -40dB attenuator on the other end of the tee.

[rhb]

Unconnected cables are the function and noise generators.  Scope is connected to the Tee.  Attenuator is in the Pomona box.

[hendorog]

So what exactly are you wanting to test? All you said was that it is an experiment.

I think you want to compare the three SDR's against each other?
In which case, why not just test them one at a time with only the function generator connected directly to the input?
Then repeat with the noise source.

One setting you haven't matched up is the RBW.

I have to say I don't get the logic of plugging everything in together, you haven't explained why you are doing that.

[rhb]

The scope is there to make sure I don't set the function generator too high and blow out the front end of the RSP2.

Finding a way to match RBW is pretty much impossible so far as I can see.  There does not seem to be a way to control the FFT length directly.  A million point FFT will have a very different RBW from 8192 points.  In fact, trying to get anything set the same is difficult.  If there is a way I've not figured it out yet.  The programs all seem to take a different approach to reducing the number of pixels in the spectral display to fit the screen.  There are no options for setting VBW that I have found yet.

The entire reason for setting up the program and taking screenshots of the 3 cases was to limit differences between the program settings for each source.  Much easier to change inputs 3 times than to start up each program 9 times.

This all got started when I noticed some very strange artifacts in spectral displays from one the programs that I could not readily identify.  So I started fooling around with the tools at hand comparing the results from different programs. The programs were written with hams in mind.  So lots of features for using an SDR as a panadaptor with transceivers.  No real support for using an SDR as a test instrument.

My interest is using the RSP2 for spectrum analysis.  From my experience so far it looks as if I shall have to write my own software to get satisfactory results.  The PLL architecture guarantees spurs so to be usable for spectral analysis the input will need to be switched from a 50 ohm termination  to the source so that internally generated artifacts can be measured and corrected.

[Ebel0410]

Two weeks ago I've made few tests with my SDR RS2, connected on a SMG 0-1GHz R&S.
Unfortunately I've not saved the results, maybe I will proceed to another tests in the coming days.
However I remember that RSP2 is pretty good in level accuracy and can be used as a not expensive level metter.
I used SDRUno to get the results.
I totally agree with you, SDR programs are not really suitable as "Spectrum analyzer".
Fortunately, I own a Spectrum (HP8594), I'm a Lucky man.

[rhb]

It would be very helpful if you made some plots.  I'm using an eBay BG7TBL  noise generator with a -40dB attenuator followed by a 5 element 1 MHz high pass filter.  I rebuilt the original 3 element filter yesterday and the results are now pretty good.  I observed one remaining spurious output around 6.4 MHz I *think* is coming from the noise generator, but have not yet tracked the source down.

I scrolled through the spectrum using SDRuno and things were pretty flat except in the HF region and a few really strong spurs (260, 360, 380 & 408 MHz).  I could see the level shift as the chipset reconfigured to work around the band gaps for fixed IF operation.  Unfortunately, I have no way to determine whether the variation in output level is the RSP2 or the noise generator.

I plan to get an SA, but not for a while.  I need to start looking into writing my own code for the RSP2 to operate it in a manner better suited to test & measurement tasks.