An economical way to measure phase noise using the PN2060A cross-correlation Phase Noise Analyzer
An economical way to measure phase noise using the PN2060A cross-correlation Phase Noise Analyzer
Six weeks ago, I had heard of the phase noise of an oscillator but didn't really know what it was, much less how to measure it. In a short space of time, courtesy of Google and Youtube, I now know what it is, how to measure it and can measure it with an inexpensive phase noise analyser.
In this post:
1 A brief summary of phase noise and how it is measured.
2 An overview of the PN2060A Phase Noise Analyzer, a low-cost instrument (US$480) available directly from Yanjun Ma BG6KHC in China, The instrument is a work in progress with new models available soon, but at a higher cost.
3 Using the PN2060A Phase Noise Analyzer to measure the phase noise of two modern spectrum analysers' reference 10 MHz and a CTI OCXO. All very easy after a couple of small hiccups as noted.
In conclusion, the PN2060A Phase Noise Analyzer is an excellent, low-cost and easy-to-use instrument.
What is phase noise and how is it measured?
Phase noise is small variations forward and backwards in the time domain, as seen on an oscilloscope. In the frequency domain, as on a spectrum analyser, phase noise appears as a spectral spread or slight short-term variations in frequency around the carrier.
A very good explanation of phase noise and how it is measured. Understanding Phase Noise Fundamentals, by Rohde Schwarz https://www.youtube.com/watch?v=hfgaEjf1154
A spectrum analyser can measure phase noise, but only for devices under test, DUT, with phase noise significantly worse than the instrument itself. All spectrum analysers use local oscillators to move signals to a measurement IF and have inherent phase noise.
A phase noise analyser uses the cross-correlation method to remove the phase noise of the instrument, leaving just the phase noise of the DUT. Modern phase noise analysers use a pair of reference oscillators that feed into analogue to digital converters, along with the signal from the DUT and all processing is done in the digital domain, either in the instrument or with an attached computer.
PN2060A Phase Noise Analyzer
The PN2060A Phase Noise Analyzer is a low-cost instrument from China, currently US$480. It is available from its maker, Yanjun Ma BG6KHC. Yanjun responds to emails very quickly and is very helpful.
It works in a similar way to the Microsemi 53100 phase noise analyser (US$24000), itself a modern lower-cost instrument compared to those from Keysight, R&S and others.
PN2060A Phase Noise Analyzer https://qsl.net/b/bg6khc// The location of the instructions and programs. https://qsl.net/b/bg6khc//pn2060a/
The Microsemi 53100 phase noise analyser, https://www.microsemi.com/product-directory/phase-noise-and-allan-deviation-testers/5565-53100a#overview Very good application notes on using a phase noise analyser. https://www.microsemi.com/product-directory/phase-noise-and-allan-deviation-testers/5565-53100a#resources
The analyser is a work in progress by the maker. Currently, there are two models, PN2060A and a later PN2060B. I have a PN2060A. I am not sure of the availability or pricing for the PN2060B, other than more expensive. The PN2060B has a lower noise floor compared to the PN2060A. There is a PN2060C in development, according to the maker's website. The PN2060C "Eliminate the “phase compensation”, and minimize the coupling among ADC channels".
The PN2060A Phase Noise Analyzer, its architecture and block diagram. Two external reference oscillators and a split signal from the DUT go into separate ADCs, with the mixing and filtering done digitally before it all passes by a USB3 link to a computer for final processing.
Operation of the PN2060A Phase Noise Analyzer
The PN2060A Phase Noise Analyzer driver needs to be installed. It is very sensitive to the type of USB3 port used and won't work with USBC or USB2, which can be a problem with a laptop. See the operating instructions.
Two good-quality 10 MHz reference oscillators need to be attached. I used two from Aliexpress at AU$22 each. They have a CTI OCXO oven-controlled crystal oscillator with a sine wave 10 MHz output and low phase noise.
Again from Aliexpress, an RF Microwave Resistor Power Splitter DC‑5G at AU$11 for 5. I should get one in an enclosure instead, but it works fine.
Just ordered this one from Aliexpress. (DC- 4GHz Resistive Power Splitter, Power Splitter SMA Interface)
Initial proof-of-concept set-up. Yes, that is the kitchen table. It is recommended to run everything from separate battery supplies to avoid correlated noise from power supplies. Well-shielded coax should be used.
The program can then be run. The window has all the settings and buttons. Click stop then set then start and the measurement will commence. The speed of the test is determined by the power of the computer (fast button) and the sample (4096). I use a 2023 MSI i7 laptop which is very fast, and it completes to 1 Hz in about 10 minutes.
A screenshot of the program measuring the 10 MHz reference output from a Siglent SSA 3021X spectrum analyser. The grey line is the PN2060A phase noise that is significantly lower than the DUT. The program can be stopped as desired. It generates a .csv file that can then be analysed by Matlab to remove spurs and construct the table of phase noise results by frequency. Just change the name of the .csv file either in file explorer or in the Matlab script as needed.
It is not necessary to use Matlab if you don't have it, as the values can be read from the computer screen to the necessary accuracy. If you are poor and can't afford Matlab, search for it at https://softarchive.is/windows/. You will need a paid file server account.
The phase noise for the Siglent SSA 3021X is ok but not great.
The phase noise of a Rigol RSA3015N is better.
Lastly, the phase noise of a CTI OSC5A2B02 board, the same as my reference oscillators, is even better. Not bad for US$22. The phase noise figures are in good agreement with the specifications of the CTI OCXO.
I am a radio amateur and am unable to check this instrument against a professional phase noise analyser. However, as above, I did check the phase noise of an OCXO with a specification sheet and the instrument seems reasonably accurate, especially for my purposes. All the results I have indicate it is reasonably accurate for different quality oscillators. To a point, a cross-correlation phase noise analyzer is self-calibrating due to the algorithm. The main issue is getting a low-noise floor.
This comment has been removed by a blog administrator.
ReplyDeleteYou may want to look at the taitien vlcu and crystek cvss-945x oscillators, very low phase noise and at 100mhz you get a 20*log10(fDUT/fREF) dB phase noise improvement when measuring a 10mhz dut with a 100Mhz reference
ReplyDelete