Chimera: 2.
a thing that is hoped or wished for but in fact is illusory or impossible to achieve.
(from the Oxford English Language Dictionary).
Phase Noise. We know what it is, but how do you measure it? Pete N6QW and I went through this back when people were casting phase noise aspersions at (Pete's!) beloved Si5351. More recently phase noise hate has been focused on (my?) beloved Franklin oscillator. When I asked a very technically proficient and guy at the VWS club if he could measure phase noise, I was surprised when he honestly said that he could not.
And now we have the IMSAI guy saying, essentially, the same thing. Wow, if the VWS guy and the IMSAI guy -- with all the spectrum analyzers at their disposal -- have trouble measuring phase noise, what hope do we ordinary hams have? I mean, at best most of us have just an oscilloscope, a NanoVNA, and a TinySA.
Look, I know that phase noise is real and in certain circumstances, it is important. But sometimes I suspect that its measurement is also a bit of a technical chimera: If, for whatever reason, there is a circuit that you don't like, you can claim that the phase noise of that circuit is bad. Or horrible. I think we see this sometimes with the Franklin oscillator. Very few hams will be able to measure it and dispute the assertion that the phase noise is bad.
For a perhaps painful walk down SoldeSmoke's "Phase Noise Memory Lane" go here:
Thanks to the IMSAI guy for a great video. But let me ask: Is there a better, simpler way to measure phase noise? One that will avoid chimerical results and that could be used by hams with sort of standard ('scope, sig gen, NanoVNA, TinySA) test gear?
Posts
Hi, PN measurement is a topic I am interested since a long time and I tried different methods.
ReplyDeleteFirst I wonder why focus on phase noise only?
Transmitting high amplitude noise on SSB is as bad as high phase noise (except for FM reception). So, in many cases composite noise (An + PN) matters.
Sure, sometimes it may important to find out the nature of the noise, for instance to achieve improvements.
Measuring composit noise straight forward with a good SDR like the Perseus and other good 14 or 16 bit DDS RX. If the dynamic range of your SDR (or spectrum analyzer) is insufficient, a self made crystal notch filter may be used to supress the carrier thus increasing the dynamic range. Finally with a DDS RX like a 14 or 16 bit Redpitaya and a gnuradio application even PN and AN may be measured. 73 yves, de hb9ewy
Hello Bill, after looking at the video of IMSA guy agin, it lokks to me that he is measuring composite sidebandnoise, not phase noise only. To measure phase noise only, a phase detector like an RF mixer in quadrature could be used. Or you strip the amplitude noise first with a diode limiter.
DeleteSM5BSZ explains a sideband noise measurement with amplitude limiter in one of his (many) videos: https://www.youtube.com/watch?v=ChkCCp68yRo
My experimental gnuradio Phase Noise utility is on Github: https://github.com/hb9ewy/grc_phasenoise
The noise of the ADC clock limits the lowest possible noise levels. With correlation measurement techniques it is possible to overcome this limitation.
Probably I will continue to work on the gnuradio PN utility this winter.
vy 73 de yves, hb9ewy
there is a report about TX composite noise measurements for amateurs with a crystal filter by DC4KU, unfortunately in german but with many illustrations: https://dc4ku.darc.de/Transmitter-Sideband-Noise_DC4KU.pdf
Delete73, Yves
Hello Bill, A very good read is some work by Hans Summers. He was looking at the Phase Noise of the Si5351 and the QCX rig. He found how you treat the impedances has a big effect on that part and a few other tricks also. He also used a crystal filter method and was able to get reasonable correlation to tests by others including ARRL lab tests of the QCX. You will find story in his QRP-Labs web page in the Si5351A synthesizer section with a subsection on Phase Noise. Enjoy 73, Dave VE3EAC
ReplyDeleteHi Bill:
ReplyDeleteMy $0.02:
Hans Summers Si5351work proves that the implementation is just as important (if not more so) than just the choice of the method. Any of these methods whether PLL, DDS free-running, etc can be serviceable , depending upon application. And any of them can yield poor results depending on implementation.
A technique can become like a fad for a while, until its real merits are understood. Often, they are replaced by something else. Remember Tunnel Diodes? Oscillator/Amplifier/Mixer, even a Flip-Flop !
Phase-Noise Measurement? Cue the next rabbit-hole, please.
73!
With regard to the "tiny" devices, indeed there is one for phase noise! Called "tinyPFA", using one specific nanoVNA hardware (nanoVNA-H4) with a different firmware, this has quite good performance and relative ease of use. One caveat still - it is only a comparison instrument, one needs a "better" oscillator as reference to check against, and exactly at the frequency of interest! Still, for a VFO or even Si5351, one could use a fixed crystal frequency that is inside the tuning range as reference, the "spot check" should be well good enough. But how good is the comparison crystal oscillator, and which topology; power level; crystal cut; etc. is best? "Build both" and compare, endless fun! https://www.tinydevices.org/wiki/pmwiki.php?n=TinyPFA.Homepage
ReplyDelete