开云体育

ctrl + shift + ? for shortcuts
© 2025 开云体育
  1. Tinysa
  2. Topics

Measurement examples posted on wiki

 

To help understanding the measurement capabilities of the tinySA I added some measurement examples on the wiki

One set of wide scans and one set of high resolution scans. Both from various signal sources to show the ability of the tinySA to help gaining an understanding of the various available signal sources.
 

Erik,
? In some of the measurement examples you provided the frequency sweep starts at 0 Hz.? In peak search functions does the tinySA ignore the internal 0 Hz signal and start its search to the right of it??

? With the dc coupling (0 Hz) of the tinySA it looks like extra care has to be taken to ensure any signals with a dc component do not exceed both the ac and dc input limit of +10dBm without attenuation.


- Herb
 

The zero hertz peak is indeed ignored
The tinySA does not have DC coupling. Below 100kHz it starts to roll off and at 10kHz there is about 8dB attenuation which is not compensated. The input capacitor is 1 microF
 

To clarify more:
The zero hertz peak is not caused by DC but by leakage of the 1st LO through the 1st mixer so technically it is not a zero hertz peak but a first IF peak. but given that it is positioned at zero hertz the name is understandable
 

On Fri, May 1, 2020 at 10:22 AM, <erik@...> wrote:
The zero hertz peak is indeed ignored
The tinySA does not have DC coupling. Below 100kHz it starts to roll off and at 10kHz there is about 8dB attenuation which is not compensated. The input capacitor is 1 microF
========================================================
Thanks for the clarification Erik,
? ?The "0 MHz to 350MHz" specification on the Wiki page threw me off.? I think the HP 8566 series spectrum analyzers were specified? down to 30 Hz but using them at less than 3 times the power line frequency was a more practical lower limit.? That and a rbw that was low enough to be useful without bumping into the 1rst LO leakage peak.

? ?I'll make a mental note of the 1uF coupling capacitor.? An un-attenuated lower limit of 100 kHz should be more than sufficient for the majority of prospective tinySA buyers.

- Herb

- Herb
 

Wiki corrected for minimum input frequency
 

The left-hand Wiki menu doesn't seem to be linking correctly to "?? "

- Herb
 

Which link fails and where does it point to?
I failed to find the error.....
 

On Sat, May 2, 2020 at 02:50 AM, <erik@...> wrote:
Which link fails and where does it point to?
I failed to find the error.....
Erik,
? See attachment.? When I select the Pages-Home button I was expecting to be taken to the Wiki Page (??? ) but on my computer that doesn't happen.

Herb

Wiki link.png

 

The wiki is not hosted on groups.io but on a separate wiki server to ensure I do not run out of space.
 

One more example added.
Measuring IIP3 of the tinySA
 

Greetings Erik,
?
? ? ? ? ?IMHO the Si5351 spectral purity screen capture is a good example that illustrates the great value of the tinySA instrumentation concept for checking out the quality of a signal source.?
?
The comb of sideband spurs shown here caught my attention as I couldn't remember seeing anything similar on some Si5351 spectral purity tests I did in the past. Attached is a screen capture of the output from one port of a prototype wired Si5351 that I once did spectral purity measurements on. At the time I had an Si5351 module prototype wired to an ESP8066 development board that was remotely controlled via WiFi. Although I found that there were a few randomly spaced spurs which were about 60dB below the 57.x MHz carrier and poking up above the -80dB noise floor as shown in the attached screen capture I didn't see a comb of spurs as shown in the tinySA Si5351 spectral purity photo.?
?
It appears that the tinySA may have revealed that there was some digital noise getting into the Si5351 generator circuit, possibly riding the power rails.
?
Best Regards,
Tom, VA7TA
?

Si5351_Clk2_SpectralPurity.png

 

Yes, that is an issue with 5351 if more than one output is used.

That is the circulating current on the ground and substrate of the chip
where in both cases there is finite resistance however small and with a?
50 ohm load on a pair of outputs its easy to see the other oscillator at
its discrete frequency at 56-60db down from the other and sometimes
even the sum and difference.? IF the load resistance is higher say 1K
then the amount of current at each output goes down and so do the
undesired products.

Its one of the things that will define the maximum dynamic range
over some span.

The other area is the mixer its IP3 is better defined but never the less real.

Also any LC filters used depending on their construction and specific
components can also contribute IP3 elements.

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

Greetings Allison,

? ? ? ?Very interesting info - thanks!

? ? ? ?I also did some tests with all three outputs of the Si5351 active and terminated. It is a couple of years ago thus the details are now a bit foggy. Attached is a screen capture obtained from the CLK0 output with the three terminated outputs set to 45, 145 and 160 MHz. The span was set to 400 MHz and of course the higher amplitude spurs are the harmonics of the 45 MHz fundamental.? As you mentioned there is quite a large number of intermodulation spurs around 55 to 65 dB down which appear to be generated by the mixing of the three output signals. The crosstalk from the 145 MHz output is about 42dB down and somewhat surprisingly the 160 MHz coupling is down around -60 dB. There was a rather strong intermod spur at about 22 MHz at only 42 dB down.?

? ? ? But what I notice different in this case regarding the tinySA spectral purity screen capture is that unlike the triple output spectrum I captured the spurs look evenly spaced at similar levels across the spectrum which gives me the impression they might be from low level modulation. It makes me wonder if there was an external influence such as digital noise riding on the Si5351 power/ground rails that resulted in low level modulation sidebands.

? ? ? For interest sake back then while I had the analyzer setup I apparently also measured the signal from a single output Si570 generator set to 100 MHz. It sure looks like a nice LO signal in comparison, (:/).

Best Regards,
Tom, VA7TA

Si5351_Clk0_45_145_160_MHz.png
Si570.png

 

Reading this I got worried about the tinySA seeing ghosts so I took a SI5351 and set all outputs to 5MHz



A nice clean scan, except for the signal just above 8MHz which is probably a spur coming from the MCU in the tinySA

Then one of the unconnected outputs was set to 6MHz



Now various 1MHz spaced spurs appear

and with all outputs set to 6MHz



so the 1MHz spaced spikes seem to originate from the SI5351, or am I missing something and is there a big problem in the tinySA?
 

Then I tried to reproduce the measurement of Tom by setting my SI5351 to the same frequencies.
That failed as my SW for setting the SI5351 probably works in a different way.
Here I have two measurements that show the huge impact of changing another output

First the outputs are set at 45, 80 and 160MHz, all multiples of 5MHz.



Then I moved the 160MHz output to 161MHz



This seems to suggest you can not compare any SI5351 measurements unless you replicate the exact settings of the 2 VCO's and the 3 factional output dividers.

45MHz_160MHz.JPG

 

Hi Erik

The Si5351A is prone to crosstalk between outputs, if the outputs are heavily loaded (e.g. 50-ohm loads). There's only a single supply pin into the chip. Other suffix versions of the Si5351 in larger packages have separate?buffers with their own supply pins and this is less of a problem.?

Therefore if you want to use multiple?Si5351A outputs on different frequencies, and you don't want that level of crosstalk, you need to buffer them with a relatively higher input impedance buffer.?

Additionally phase noise performance is degraded when the outputs are heavily loaded. One partial remedy to this is to use a DC blocking capacitor, which appears to restore the phase?noise performance.?

Note?also that the output spectra of the Si5351A will depend somewhat on the internal configuration. For best low jitter performance (i.e. good phase noise performance) the datasheet?recommends even integer MultiSynth divider stage configuration where possible, using fractional feedback division on the PLL stage. There is quite a lot of bad Si5351A code out there in libraries and such, that does things the WRONG way.?

My article on phase noise measurements of the Si5351A might be of interest see?

73 Hans G0UPL
 

Hi Hans,

Thanks, excellent investigation.
The SI5351 SW used was optimized for fast sweep for my DIY VNA so it is terrible with spurs.

1 - 18 of 18