PN2060A GHz phase noise measurement using only reference OCXOs

PN2060A GHz phase noise measurement using only reference OCXOs.

In this post, I extend my earlier posts on using a PN2060A phase noise analyser to measure GHz signals with downconverters but using the same reference OCXOs for both the PN2060A and the local oscillators. This avoids using an additional pair of oscillators for the LO signal generator and the consequential addition to spurious phase noise in the measurement.

The results seem plausible for the RF sources I have on hand at 1 GHz and are consistent with the instruments' specifications.

https://qsl.net/bg6khc/pn2060c_phase_noise_analyzer.htm

Microchip 53100A phase noise analyser for GHz measurement

The Microchip application note AN3899 investigates using dual channel downconverters to measure GHz phase noise. A simple modification is to use the local oscillator reference oscillators as the reference oscillators for the PN2060A phase noise analyser. This allows just one reference oscillator for each channel. I don't know why AN3899 uses a single reference oscillator for the analyser as two would seem better for cross-correlation. (I read somewhere that it makes little difference as the noise from the downconverter oscillator is larger and swamps the reference oscillator.)

https://ww1.microchip.com/downloads/en/Appnotes/AN3899-UHF-and-Microwave-Measurements-with-the-53100A-Phase-Noise-Analyzer-DS00003899A.pdf

I use a pair of BG7TBL WB-SG1 wide-band signal generators, 35 MHz to 8 GHz (~US$160). It uses an ADF4351 a wideband phase lock loop (PLL) synthesizer with a voltage-controlled oscillator (VCO). The reference is a used, commercial grade, 10 MHz OCXO with a phase noise floor around -145 dBc/Hz. The signal generator has provision for input or output of a 10 MHz reference.

https://reeve.com/Documents/Articles%20Papers/Reeve_WB-SG1_SigGen.pdf

In my case, the 10 MHz reference output from the signal generator is used as a reference 10 MHz for the PN2060A phase noise analyser. This avoids needing a second set of local oscillators that add extra phase noise to the measurements, although the VCO phase noise will probably swamp it.

 The VCO in the synthesizer adds phase noise but is unavoidable. The specified phase noise for the ADF4351 is −86  dBc/Hz 10 kHz offset from a 3.3 GHz carrier. This will swamp the phase noise of the signal generator's OCXO at -145 dBc at 10 kHz. The fundamental output of the ADF4351 is 2.2 to 4.4 GHz. A wider frequency range is achieved by doubling or dividing. As such, measuring phase noise at 3.3 GHz should indicate the measurement's accuracy compared to specifications.

"moRFeus" frequency converter and signal generator from 30 MHz to 6 GHz

My first test device is a "moRFeus" field-configurable, portable, and budget-friendly frequency converter and signal generator from 30 MHz to 6 GHz (~US$150 but out of stock). The signal generator chip is not specified and claims phase noise of -98 dBc/Hz at 3 GHz.

https://othernet.is/products/morfeus-1

First, I checked the output on a real-time spectrum analyser for signal purity. The signal is clear of any spurs but the frequency is out compared to the spectrum analyser (up to 1.5 GHz). I checked the accuracy of the spectrum analyser with a SRS PRS-10 Rb/GPSDO. The analyser is spot on, so the moRFeus is out.

The results seem pretty good. A clean signal consistent with the spectrum analyser. About 93 dBc/Hz at 10 kHz at 1 GHz and 83 dBc/Hz at 10kHz at 3 GHz. The measurements are about 10 dBc/Hz worse than the specification. However, I would be more convinced of the specification if there was a phase noise plot or identification of the signal synthesiser chip.

1 GHz

3.3 GHz

Anritsu MG3700A vector signal generator

Next, I measured the phase noise of my Anritsu MG3700A vector signal generator (to 3 GHz). Again, I observed the signal at 1 GHz on the spectrum analyser. It is a reasonably clean signal except for two sidebands of 1700 Hz from the carrier. Again, the carrier frequency was off, 250 Hz too low. The mixed result from the spectrum analyser illustrates why it is useful to check it before measuring phase noise.

At 1 GHz, the phase noise is -112 dBc/Hz at 10 kHz and -110 dBc/Hz at 3 GHz. The signal is noisy below 10 kHz, presumably due to the 1700 Hz spur.

The results initially disappointed me, but on checking the GHz phase noise of other vector signal generators, their performance is ballpark, depending on how much one spends. The Phase Noise for a Siglent SSG5000X RF Signal Generator is quoted as  -120 dBc/ Hz @ 1 GHz, 20 kHz. I have included the phase noise plots for the  R&S SMW200A Vector Signal Generator for comparison, as the current gold standard. It is pretty awesome but I dread the price (> US$100k?). 

1 GHz

 

3 GHz

R&S®SMW200A Vector Signal Generator GHz phase noise

Conclusion

The downconverter for the PN2060A phase noise analyser seems to work well, at least for noisier sources. I need to find the noise floor of the PN2060A when used with the downconverter. The noise of the VCO in the local oscillator signal source seems to be a limitation. From chip specifications, they are usually around -110 to -120 dBc/Hz at 10kHz in the GHz range. My understanding is that cross-correlation phase noise analysers improve the noise floor by around -20 dBc/Hz, so the resultant noise floor is roughly -130 to -140 dBc/Hz. The R&S instrument uses a very low phase noise VCO but still struggles at higher frequencies. On the other hand, most signal generators struggle to get below -120 dBc/Hz at 10 kHz in the GHz range.

The BG7TBL WB-SG1 wide-band signal generators are a convenient way to generate both the local oscillator signal and the 10 MHz reference signals for the PN2060A. They do not have an adjustable output and at 0 dB, it is too low for a non-active mixer, hence the need for an active mixer. Using its reference OCXO for the PN2060A simplifies the number of devices and power supplies needed and may reduce noise a little, but is probably swamped by the VCO phase noise. They are very easy to use.