Best option to measure phase noise

[mehdi]

Hi.
I'd like to measure the phase noise of my RF signal generator, however I don't have a spectrum analyzer with good-enough phase noise; Like at least -110dBc/Hz at 10KHZ offset at 1GHZ.
(My best spectrum analyzer is SignalHound BB60D which uses a superhet architecture and also doesn't have good phase noise characteristics)
I am looking for something affordable (below 10K) and preferably not too old/heavy (like ancient HP devices)
These are the only options I know:
1. SignalHound SP145 (10K. released just last month)
2. Aaronia ECO 18 (not released yet. Not much data available on its phase noise, but previous models using OCXO had good phase noise)

Anything else I've missed?

[jwet]

I'll attempt to answer since you've got no responses.  I don't know what your experience level is and what level of phase noise you're trying to measure but good phase noise measurements at low levels (good sources) are generally expensive and difficult.  HP/Agilent and Rohde & Schwartz have great app notes and R&S even has good you tube videos.  Here are a few tidbits of advice.  Maybe others can chime in and enforce or disagree- this is how this board generally goes.  Comments follow.

1. If you have high a quality piece of gear that that you're concerned about, you can rely on the manufacturer's specs.  Phase noise is a design type spec and if the SG works generally and passes calibration/selftest, etc, there is no reason to think that the phase noise is worse than the specs.

2. The RS and HP App notes go through it better than I can, but for the brute force method using just a good SA, you will need an SA that has a 10-20 db better spec that your DUT.  This is difficult at low phase noise levels.  There is also the problem of dynamic range.  You've got this big carrier and you are trying to measure small noise components close in.  One technique is to put an ultra narrow notch filter at the carrier frequency, this eases the DR problem significantly- you're just measuring residuals similar to THD measurements.  This filter itself can be expenisve.  You also have issue in the brute force technique of the true shape and BW of the piece of offset spectrum you're measuring.  The HP and RS notes go through this, it's not difficult but isn't obvious either.  Some good SA's might account for this internally.

3. There is a clever technique that you can look up (again look at HP or RS) that is called "the cross correlation method".  This method uses two simultaneous measurements using two uncorrelated receivers feeding an autocorrelator.  Some clever math makes a lot of the measurement system's phase noise drop out and you're left with just your DUT phase noise.  This is the way most good phase noise analyzers work rather than just brute force.

4. Hams and others without a NASA budget can infer phase noise with techniques like reciprocal mixing where you build a receiver with your source as LO and A-B test it with a really low phase noise source and see what your close in noise looks like.  One of the old Ham standards was the HP8640, a 50 pound HP beast that has pretty exceptional phase noise to this day.  They can be had for a few hundred and maybe a few hundred more to send to a Cal Lab for a look.  Journals like QEX, a specialty technical Ham journal would be a good place to start.

5. If you really need to measure phase noise, you might consider renting a good Agilent or RS box.  You might be able to get a short rental time of a week or use the instrument at the supplier's site for a day.  I would think that you can get your hands on a good instrument for < $1000 USD/Day if you look around.

I don't know of any general purpose instrument that would be useful and not very specialized.


Good luck.


John

[mehdi]

Thanks for your detailed answer John.

HP/Agilent and Rohde & Schwartz have great app notes and R&S even has good you tube videos.

Yes, I have been watching/reading those recently (plus other resources like TheSignalPath)

If you have high a quality piece of gear that that you're concerned about, you can rely on the manufacturer's specs.

using just a good SA, you will need an SA that has a 10-20 db better spec that your DUT.

That was pretty much my conclusion so far: I won't get access to something affordable enough to measure e.g my OCXO's phase noise, and even for the signal generators I am building, the difference between my DUT's phase noise and the spectrum analyzer I was looking at (SignalHound SP145 and SM200C) is still less than 10dB in some offsets.

[Kleinstein]

For a few discrete frequencies one can build a oscillator + mixer setup to convert the DUT frequency to a lower frequency range and this way higher relative phase noise, so that than the SA may be good enough. For something like an OCXO this may need having 2 similar OCXOs to test the sum of the phase noise of the two.

[jwet]

This is what I was generally referring to in my number 4 response above of how Hams approach this.  The idea of using fixed OCXO modules is very good.  I don't have much direct experience but see fixed TCXO/OCXO modules in trade magazines with phenomenal specs.

[Solder_Junkie]

You can use crystal filters to notch the fundamental leaving noise either side.

This is a simple version, cascading notch filters will give better results. A web search will probably bring other circuits.

http://www.ve2azx.net/technical/XTAL_NOTCH_ckt.pdf

SJ

[G0HZU]

Here's an old plot of a crystal notch filter I designed a few years ago for measuring phase noise. The plot below is the initial simulation based on the 2 port s-parameter data from the crystals. The filter uses 6 crystals and the plot below is only 10kHz wide. You can see that this notch filter is much sharper than other designs and the insertion loss and return loss of the passband is really good.

The design was optimised on a computer for best shape factor and narrowest bandwidth. I did build this filter using real crystals a couple of years or so ago and it performed as per the simulation. I keep meaning to make a second version centred on 10MHz. I have the crystals here but haven't had the free time to make it yet.

[G0HZU]

I managed to find the prototype 9830kHz notch filter and I have just measured it with a VNA.

See the plots below. The full notch depth of >70dB can only be seen on a 2kHz span. I really should make up the 10MHz version as it could be used to look at various 10MHz crystal oscillators.

[Gerhard_dk4xp]

I do not consider notch filters a good idea. You usually need 10 dBm or more to see the
far-out grass on even good, specialized spectrum analyzers. A crystal will not survive that
very long. A crystal is usually specified for 0.5 mW on the resonant frequency, for long-term
stability even less. 10 dBm is 20 times that level.

If you are serious with bleading edge OCXOs, cross correlation is a must. That allows you
to measure oscillators that are 20 dB better or so than the 2 references that are needed.
You get  only the noise that is contained in both measurement channels at the same time.
The ADC and the reference noise is averaged away.

I use a Miles Design Timepod for this. It has the drawback that it is limited to 30 MHz.
I think there are faster/better successors for this, from Microchip and someone else
whose name escapes me right now.

I'm working on a 15 GHz two-channel dual-conversion down converter for it. First IF
is 900..928 MHz with SAW filters, 2nd IF is 0 to 28 MHz. OK, the 2nd IF starts at 5 MHz
because of Mini Circuits ring mixer limitations. And then I cannot get the LT5553 for
the high band, which is unobtainium, so I'll have to use the HMC-220 for now.

If I did not have the Timepod, I'd probably build this machine from Holmes,
which consists mainly from some evaluation boards.
<   http://www.aholme.co.uk/PhaseNoise/Main.htm     >

Cheers, Gerhard   DK4XP

[Co6aka]

Leif Asbrink, SM5BSZ, has been doing a lot of experimenting with measuring phase noise, and other noise, over the past five years. His "Nerds#" series of videos is fascinating:

https://www.youtube.com/@sm5bsz289/videos

[G0HZU]

I do not consider notch filters a good idea. You usually need 10 dBm or more to see the
far-out grass on even good, specialized spectrum analyzers. A crystal will not survive that
very long. A crystal is usually specified for 0.5 mW on the resonant frequency, for long-term
stability even less. 10 dBm is 20 times that level.

If you are serious with bleading edge OCXOs, cross correlation is a must. That allows you
to measure oscillators that are 20 dB better or so than the 2 references that are needed.
You get  only the noise that is contained in both measurement channels at the same time.
The ADC and the reference noise is averaged away.

I use a Miles Design Timepod for this. It has the drawback that it is limited to 30 MHz.
I think there are faster/better successors for this, from Microchip and someone else
whose name escapes me right now.

I'm working on a 15 GHz two-channel dual-conversion down converter for it. First IF
is 900..928 MHz with SAW filters, 2nd IF is 0 to 28 MHz. OK, the 2nd IF starts at 5 MHz
because of Mini Circuits ring mixer limitations. And then I cannot get the LT5553 for
the high band, which is unobtainium, so I'll have to use the HMC-220 for now.

If I did not have the Timepod, I'd probably build this machine from Holmes,
which consists mainly from some evaluation boards.
<   http://www.aholme.co.uk/PhaseNoise/Main.htm     >

Cheers, Gerhard   DK4XP

I think a lot depends on how challenging the phase noise spec is to be measured. I've had very good results using a narrow crystal filter as well as the crystal notch filter. I generally drive the notch with a carrier at 0dBm meaning the analyser measures phase noise with no need for any correction offsets.

The notch filter typically requires a mixer and LO ahead of it and this is usually the weak link.
Also, the crystal notch filter method is no good if the aim is to measure phase noise at < 500Hz offset from the carrier. At offsets >500Hz, it is one of the most cost effective ways to boost the phase noise measuring capability of a spectrum analyser.

In this case, the OP is looking to have an instrument that has a phase noise spec of -110dBc/Hz at 10kHz offset from a 1GHz carrier. I don't think the notch filter is the weak link here.

The LO could be something as simple and cheap as an LMX2572 PLL/VCO chip. This can offer low phase noise close to carrier even up at 1GHz. Obviously, there are better options but the price will go up a lot. I think the LMX2572 chip (on its own) is about £30.

[hendorog]

You might be able to cheat a bit by using Noise Floor Enhancement since your requirements are not very strict.
The SignalHound SA124B (and SA44B assuming it performs similarly) might be good enough to meet your needs when combined with NFE and are relatively inexpensive.

Procedure is something like this from memory. Measurements are done with lots of averaging.
Measure a really clean signal as a reference, and convert the measurement to mW.
Then measure your DUT and convert the measurement to mW.
Subtract the reference from the DUT and convert it back to dBc/Hz

I got this from here:
http://literature.cdn.keysight.com/litweb/pdf/5966-4008E.pdf

Bit more info here where I had an attempt at using it.
https://www.eevblog.com/forum/testgear/phase-noise-head-to-head-siglent-ssa3032x-vs-hp-8566b-vs-signalhound-sa124b/msg1628197/#msg1628197

* Reading this again it is pretty cryptic I fully admit, it was a while ago and no guarantee I have done it correctly

[tggzzz]

Can I put in a mention of a book I like: Prof Thomas H Lee [1] wrote a rather good [2] book "Planar Microwave Engineering, A Practical Guide to Theory Measurements and Circuits"[3].

That has two chapters on phase noise, one on phase noise in oscillators, and one on different ways to measure phase noise. The latter starts with "In keeping with the guiding philosophy of this book, phase noise measurements methods suitable for the weekend experimenter are presented in addition to techniques more commonly used in professional laboratories."

[1] to be found contributing to TekScopes

[2] i.e. despite having many closed-loop integrals, it also discu is sufficiently easy to read that I can fool myself into believing I could understand the topics - given a decade or two

[3] think of it as the TAoE for RF

[Solder_Junkie]

For ARRL members, the November/December QEX magazine has an article called “Phase noise measurement revisited” by Dennis Sweeney, WA4LPR.

SJ

[mehdi]

and one on different ways to measure phase noise.

Yes, it provides 3 methods:
* PLL-based phase detector
* Delay line discriminator
* Resonant discriminator

However, they are discussed very briefly, and with too much theory.
There is also a book called "RF & uWave Measurements" by "Shiv Prasad Tripathy", again very briefly (2 pages) touching this subject,

[tggzzz]

and one on different ways to measure phase noise.

Yes, it provides 3 methods:
* PLL-based phase detector
* Delay line discriminator
* Resonant discriminator

However, they are discussed very briefly, and with too much theory.
There is also a book called "RF & uWave Measurements" by "Shiv Prasad Tripathy", again very briefly (2 pages) touching this subject,

I disagree about it being too much theory. Without general/generic theory, specific practices would only be cargo-cult engineering.

Of course the practice is also necessary, and manufacturers provide application notes about the practice of using their equipment.

[cncjerry]

A number of us built this system and it far exceeded expectations: http://www.aholme.co.uk/PhaseNoise/Main.htm

Total cost with references, splitters, filters,  and power supplies was less than 2k.

Jerry