SA's: Anritsu MS2711/12E vs R&S FSH3 vs Agilent N9340B: Input wanted

[ivaylo]

A decent old school bench analyser would show about -130dBc/Hz at 1MHz offset at 1GHz. A good one would get closer to -140dBc/Hz.

Do you mind listing models (any manufacturer) which fit this characteristic?

[Mr Simpleton]

HP 8560 series
HP 8566/68 series
All have good phase noise and are a pleasure to use 

[videobruce]

You forgot to add, they are 30 year old boat anchors & power hogs with CRT's.
.

I also (very casually) measured a fairly decent 'semi portable' Agilent spectrum analyser the 26GHz 8563EC

How much was that when new $40k (US)??

[1design]

Actually I missed what are you going to use this for, as old SA do not support so many measurements as the new ones do, especially regarding digital demodulation, things like burst demodulation, EVM and other constellation data, gated sweeps, SEM, ACP, etc. Take this into consideration when choosing such an instrument. I really like the channel power, occupied BW etc. for digitally modulated signals; it saves me time over hand calculations.

BR

[videobruce]

OTA broadcast, mostly TV. Also 2-way communications & WiFi. No cellular. Some RF sniffing.

[videobruce]

G0HZU;
I was only looking at the 1 MHz offset since that was the most common spec. I now see, the 10kHz offset really drops that number way down on some.
Regarding this phase noise, looking back over the specs of a few of these the Rigols (815 & the 1020) are -80dBc/Hz at a 10khz offset. The 1030a is -88dBc
The 9340E is -87 dBc, but the 2712E is -100dBc (-110dBc typical).
That 2711D is a much older design.

What applications would a lower phase noise benefit the most. And what applications would it benefit the least?

[G0HZU]

High phase noise masks weak signals (it raises the noise floor)  if there is one 'big' signal on the screen.

So having low phase noise is really useful if you want to be able to look at small signals that are very close to big ones.

See the image below for an extreme case. The black trace is an old school (huge/heavy) bench analyser from Japan from 25-30 years ago. The signal going into the analyser is from a low noise sig gen at 1GHz. The span is 2MHz and the RBW is 10kHz. The blue trace is my handwritten overlay of what you would see on a MS2711D on the same span with the same RBW because it has such high phase noise.

You can see how the little portable analyser will not be able to display small signals that are close to the carrier because they are covered by all that noise.

However, don't get too worried because for portable monitoring on a small aerial you are unlikely to be seeing such a huge variation of signals and the little portable analyser will not have huge signals masking the performance like this. So for typical portable use then don't worry...

But for bench use you might want to look for tiny spurious signals from a transmitter and the little portable analyser will struggle a bit. You can use narrower spans and lower RBW settings to overcome some of this but it means the measurements aren't as quick and easy as with a decent bench analyser.

[videobruce]

don't get too worried because for portable monitoring on a small aerial you are unlikely to be seeing such a huge variation of signals

Unless it is connected to a outside antenna which it can and will be.
That is exactly what I have had a problem with. Overload from nearby TV xmitters (7 miles away). It raises the noise floor depending on settings.
Then that is a biggie, at least for me.

Then how does DANL come into play?

[The Electrician]

Here are images showing the performance of a Rohde&Schwarz FSH4, and an Anritsu MS2713B

I applied a -10dBM, 1 GHz signal from an old Agilent 8657B siggen.

The SA settings were 10 kHz RBW, 1 kHz VBW, 5 MHz span, preamp off, attenuator off.

You can see that the Anritsu is overloaded.

The sweep speeds are definitely slower than a bench SA, but they seem reasonable to me.

I think that as time goes on and DSP processors get faster and lower power consumption, we'll see faster sweep speeds.  I'm impressed at just how long the battery life is on these units--several hours typically.

[1design]

Phase noise is important when looking at closely spaced CW or narrowband signals, for wideband signals you must also look for TOI distortion, which will come into account as soon as there is more than one tone present. This is a very practical parameter that informs you about the linearity of you SA. For example, broadcast stations used in uW backhaul might have 50-60dBc IMD3, so you need a very good SA to measure such systems as they use adaptive predistortion. On the other hand, WIFI routers have an IMD3 of only about 20-30dBc which is easily visible by any SA. I will try to take some pictures next week to help you see the difference, like The Electrician did.

BR

[videobruce]

First off, I would like to thank the all the members, especially the recent additions. The information is very welcome. 

The Electrician; Very interesting. You wouldn't happen to have a Rigol DSA1030A (or a 1030) and/or a Agilent N9340B would you?? Of course you know this is throwing a wrench into the works. 
One thing you have to admit, the 27xx series has a killer display, the exact opposite of the FSH3.

[G0HZU]

I had a go at overlaying the phase noise profiles of all the analysers mentioned.

The red trace is the FSH4 based on the plot from the Electrician.

The green trace is the typical phase noise of the 2711E according to the datasheet and the yellow trace is the typical phase noise of the 2713E according to the datasheet.

You can see that the 2711E is quite noisy compared to the others (apart from the older 2711D in blue)

I think I would be happy to accept the phase noise performance of the FSH4 and the MS2713E for portable use but the 2711E (green) looks too noisy. However, maybe it will be better than the green trace when it is measured on a real unit?

[G0HZU]

The other thing that looks weak on the Anritsu is the spec for harmonic distortion. It seems to get weaker with lower frequencies.

It states 0dB attenuation and a -30dBm tone and a typical internal harmonic would be -60dBc. Again, a decent bench analyser will give a typical result that is much better than this.

A 'reasonable' bench analyser will have a 2HI of about +50dBm at the mixer so for a -30dBm signal the harmonic would be -80dBc. A strong bench analyser will have a 2HI closer to +70dBm so the harmonic would be -100dBc for the above test.

Usually, you are supposed to allow a 15-20dB margin on the internal harmonic level if you want to make a reasonably accurate measurement of a second harmonic level. Otherwise the internally generated harmonic can cause high measurement uncertainty. If you allow an 18dB margin then the uncertainty falls to about 1dB. So the 2711E would be good to -42dBc if you measured with a -30dBm fundamental signal at the mixer and wanted to keep the uncertainty down near 1dB.

This would be OK for typical ham radio use because the harmonic rejection requirements are very sloppy on the ham bands but it's not so good if you wanted to measure something with a decent spec like -65dBc. In this case you would have to use extra attenuation on the Anritsu and use a narrower span and RBW to claw back the required range.

None of the above really matters for portable use but if you want to use the analyser on a bench for more serious harmonic rejection testing you might hit a few problems

[The Electrician]

First off, I would like to thank the all the members, especially the recent additions. The information is very welcome. 

The Electrician; Very interesting. You wouldn't happen to have a Rigol DSA1030A (or a 1030) and/or a Agilent N9340B would you?? Of course you know this is throwing a wrench into the works. 
One thing you have to admit, the 27xx series has a killer display, the exact opposite of the FSH3.

Sorry, but I don't have access to a Rigol or Agilent. 

I also haven't ever had access to a R&S FSH3, but the FSH4 is a newer unit, and I would expect better performance all around.

The FSH4 display has good resolution, but it's not as bright as the Anritsu.  The Anritsu is "majorly" bright (as my young nephew would say).

Another thing I notice is that the numbers for sweeps/second on the Anritsu don't make sense to me.  I set it up to look at a local TV station (a digital signal here in the U.S.) at around 600 MHz.  Using a 5 MHz span and 10 kHz RBW, the instrument reported a sweep time of 900 mS, but the update rate on the display is much faster than that; I would estimate about 50 mS.  I see this behavior constantly; the unit is much faster than it indicates.  They must mean something different than what I would expect by the term "sweep time".  The FSH4 doesn't behave like this.  It reports a sweep time that is consistent with what you see on the display.

Both units exhibit the behavior you would expect.  For wide spans and narrow RBW, it gets slow. 

For other settings, I've seen the FSH4 report sweep times as low as 22 mS, and for similar settings the apparent performance of the Anritsu is just as fast even though it reports a much longer time.  I can't explain this.

R&S makes another "semi" portable SA, the FSL3:

http://www.rohde-schwarz.us/en/products/test_and_measurement/aerospace_and_defense/test_and_measurement_instruments/spectrum_signal_analyzers/FSL.html

 which I have occasionally had access to, and I've seen that unit exhibit sweep times of as low as 2.5 mS!  It has a battery option, too.

The N9340B looks good; maybe you can get the Agilent rep to show you one.  Their spec of .1 second for a full 3 GHz sweep (RBW unknown) is outrageously fast!

The Anritsu, for a 3 GHz full sweep, 3 MHz RBW, reports 14.785 seconds, but visually I estimate it's taking about 1.5 seconds.

[videobruce]

The Electrician;
Could you comment on those questions in those two screen shots of yours in my previous post?

I'll look into the FSL3.

[The Electrician]

Sorry.  I didn't realize you had annotated the images with questions; I can't see that in the thumbnails. 

WRT to the Anritsu, when I said it was overloaded, I was basing that on the message that says "ADC Over Range..."  I assume that the red colored portion of the spectrum might be affected by the overload, and by coloring that portion red the analyzer is warning us that perhaps that portion is not trustworthy.

About the FSH4, I guess the FSH4 has a different PLL characteristic in the reference oscillator circuit, leading to a different shape and rolloff rate in the phase noise.  It looks like Anritsu did a somewhat better job with their design.

[videobruce]

How do you get the images embedded in the post like you did?

Regarding the pattern on from the Anritsu, those 'sidebands' if you can call them that, could it be since there is a overload situation, the scope reduced them so the entire signal would fit on the screen? (Hope that made sense)

[The Electrician]

How do you get the images embedded in the post like you did?

I'm not sure how you're supposed to do it, but what I do is attach them and Post, leading to thumbnails like you have.

Now you have to go back and modify your post to add each image.  Right click on one of the thumbnails and cntrl/C to capture the link.  Then click in the upper right of the post on "Modify".  Then, where you want the image, type "[i.m.g]", paste the link address, then type "[/i.m.g]".  Finally, "Save".  Repeat for each thumbnail.  Leave out the decimal points in [i.m.g]; I had to put them in to be able to show the img command.

+Regarding the pattern on from the Anritsu, those 'sidebands' if you can call them that, could it be since there is a overload situation, the scope reduced them so the entire signal would fit on the screen? (Hope that made sense)

I'm not sure about that.  On both analyzers, as I reduced the input, the "hump" of phase noise gradually reduced and one was left with a normal looking spectral display.  There didn't appear to be anything happening like you describe.

[The Electrician]

OMG!!!  Have a look at the movie on this page:

http://www.rohde-schwarz.us/en/products/test_and_measurement/aerospace_and_defense/test_and_measurement_instruments/spectrum_signal_analyzers/FSW.html

Phase noise of -137dBc at 10 kHz offset (1 GHz carrier)!!   

You get what you pay for.

[G0HZU]

Phase noise of -137dBc at 10 kHz offset (1 GHz carrier)!!   

Going a bit off topic here, but you have to be wary of modern (marketing) spec claims for phase noise at 10kHz offset. The Agilent PXA can get close to this figure... eg it can get to something like -133dBc/Hz but it does this by pushing out the loop bandwidth to something like 700kHz

It is a special type of phase noise cleanup mode on narrow spans,

So you get -133dBc/Hz at 10kHz (because of the high noise cleanup inside the loop BW) but the phase noise at 600kHz offset will be -120dBc/Hz. But you don't easily spot this penalty because this increased hump of noise up at 600kHz is off the screen. The Agilent marketing bods will make this downside less obvious in the marketing specsheets 

When you select a wider span it will turn this wide loop BW mode off and this cleans up the phase noise at 600kHz offset to be more like -140dBc/Hz. The phase noise at 10kHz in this 'normal' mode will be about -110dBc/Hz.

I strongly suspect that R&S are resorting to the same wide loop BW trick with the FSW.

[Mr Simpleton]

I may have access to FSH3 and FSL if you want/need some tests done!
Try to see if there is any Agilent stuff around, but doubt it... Mostly old heavy HP, as carry around has not been a primary ting.

I told you how I feel about the FSH3, but sure can do some screen captures if needed, and if I can find the RS232-optical interface.

I think newer S/A are overpriced... new hardware should bring price down and performance up... now with the Rigol they seem to use the integrated PLL/VCO from Agilent, and this gives mediocre p/n. Pity, they should have spent some more developement on the LO.

Have you had a look at the Hameg HMS? Sits in between R&S and Rigol both in performance and price, and now owned by R&S

[G0HZU]

Dunno if this helps but I quickly set up a demo of Agilent's phase noise enhancing method by applying it to a PLL board I have here that I can control/program for different loop characteristics.

The plots below are of a single loop PLL running at 1500MHz. In the first plot I've selected a normal loop bandwidth for the PLL on the test board and you can see that the close in phase noise is quite high but it tapers away in a normal fashion.

In the second plot I've dynamically altered the PLL loop bandwidth. I've not changed the phase detector frequency, I've just altered the PLL characteristics by reprogramming it for a different loop characteristic. This is all done in software i.e. I don't have to change any components on the PCB to get this change.

You can see how I can now claim lower phase noise at a 15kHz offset for my little test board!

But the penalty is the hump of noise at a 150kHz offset. The phase noise is now WORSE at 150kHz to get the noise 'better' at 15kHz.

Agilent appear to do something like this on the synthesisers in the PXA analyser but the hump of noise is at 600kHz offset.

Anritsu could well be using this trick on the MS2713E because this technique is making its way into lots of test gear. My HP/Agilent ESG 4000D sig gen has this feature as a user control. I can make the close in phase noise look really good but the far out phase noise gets much, much worse...

[The Electrician]

Imagine if the image I posted for the FSH4 were 500 kHz span rather than 5 MHz.  See the hump close to the center freq?  Looks similar, doesn't it?

[videobruce]

The Electrician;
Get a load of the size of that scope and the size of the display! 
Gotta love looking at  pieces of gear worth $50k or so.

I thought only the CEA (consumer electronics) were the ones that pull stunts like that.

[videobruce]

I may have access to FSH3 and FSL if you want/need some tests done!

Yes, that would be beneficial.
Yes, I have ruled out the FSH3, but not the 4, but I'm afraid it is out of my price range. At least they at least went with a 640x480 display. I wish Agilent would redesign the 9340 and get rid of the video game look. 
.

I think newer S/A are overpriced... new hardware should bring price down and performance up... now with the Rigol they seem to use the integrated PLL/VCO from Agilent, and this gives mediocre p/n. Pity, they should have spent some more developement on the LO.

Agreed! Regarding Rigol are you referring to all of their SA's or just the 1030/1030A?
.

Have you had a look at the Hameg HMS? Sits in between R&S and Rigol both in performance and price, and now owned by R&S

No, but anyone have any "vouchers' lying around? 
Looks like you have to buy the ECM option to get the preamp.