Spectrum Analyzer, which one to get?

[amdnra]

Hi. I've been looking to get a Spectrum Analyzer.
The choices at this point are:
Used Anritsu MS2721B
Used Rohde & Schwarz FSEA30 20Hz to 3.5GHz
New Rigol DSA832E-TG
New Siglent SSA3032X
All include the Tracking Generator
Trying to get something with an LCD display and stay around $2500 give or take
Any ideas or advice? Thanks
 

[DaJMasta]

That first one is sticking out like a sore thumb... it's a great unit (I've got one with no TG), but it's in a very different class as a portable SA.  You gain the ability to run off battery and save space on the desk, but you loose sweep speed responsiveness - the minimum sweep speed on the latest firmware is 100ms/sweep.  It's very nice to have the bandwidth too, but it's not well suited to some applications with very short transmissions or looking at very wide spans.  If you're looking for portable SAs, maybe the Rohde & Schwarz FSH series is worth looking at - the 6GHz model (not the current version) should be in that price range.

The others will give you a lot faster sweep speed, even if their DANL isn't all that different, but come at the cost of bandwidth, desk space, and portability.  It's also worth noting that the options packages can be very important for particular applications, so if you're interested in a particular kind of transmission or demodulation capability, that should weigh heavily on your choices.

Worth mentioning that an 3.5GHz FSEA20/30 should not be running you $2500, at least not in the US, and for that money you may be able to strike at the higher 7GHz model (or the FSIQ line).  Could be worth considering the Anritsu/Advantest benchtop models in a more comparable form factor to the FSEA, you'd be getting more performance for your money if the portability isn't important.

The Rigol and Siglent have a nice compromise on desk space, but not the portability, and offer good speed and noise performance.  They've both been pretty well investigated on the forums and have the advantage of being new and having much more reasonably priced options (though perhaps not as comprehensive options as some of the older, larger models).  Good choices, but in terms of raw bandwidth and performance, new instruments rarely compare well to similarly priced used instruments.

[nctnico]

Also look at Advantest and include their 'signal analysers'. Contrary to Agilent Advantest seems to bolt signal analyses onto a regular spectrum analyser so good deals can be had.

[usagi]

My Used spectrum analyzer buyer's guide might be able to help you 

If you want bang for buck it's hard to beat the R&S CMU 200. It's a big rackmount unit though.

I also have some used R&S FSH6 for sale.

[amdnra]

DaJMasta: What's the latest firmware rev for the Anritsu? If you know
So...The Rohde & Schwarz is pretty big and heavy
The Anritsu is not so fast or capable
And from the forums it appears like the Siglent is better than the Rigol (better specs)
The 'winner' seems to be the Siglent. Would you agree?
Thanks for your input

nctnico: Thanks for the tip

usagi:Thanks for pointing out the guide. Informative.
About the R&S CMU200, Is that a Spectrum Analyzer? if so, What's the bandwidth?
Would the FSH6 suffer from the same limitations as the Anritsu (see reply above by DaJMasta)?
How much are they?

[DaJMasta]

Most portable SAs will have that sort of sweep time limitation, not sure if the FSH series has the same lower limit, but I would expect the same ballpark performance (you can probably find videos of them being used to get a general idea).  Firmware is complicated because it gives me four numbers... Package 1.51, OS 3.21, Base 5.71, SPA 5.73 - software updating is easy though, it's got USB ports so you stick in a memory stick with the new one and hit a few buttons (and the download is free from anritsu when registered, if I remember right).  While there is the sweep speed limitation, a portable SA is more likely to have a preamp option, will probably have better input protection, etc.

While portable SAs do definitely have speed limitations, the extra bandwidth is nothing to shake a stick at - you usually pay a pretty big premium for extra bandwidth, so the 7.1GHz it offers is considerably above the other options (and a bit above the R&S 6GHz instruments which may still fit the price point).


I think the FSH6 series should run in the $2000 US ballpark - there was recently a government auction with a few dozen of them, and I'd expect it to be more of a buyer's market on ebay as a result.  I think prices are already down a bit from several months ago, but it still may dip a bit.

[amdnra]

DaJMasta: What's your opinion on the R&S CMU200?
and...
"So...The Rohde & Schwarz is pretty big and heavy
The Anritsu is not so fast or capable
And from the forums it appears like the Siglent is better than the Rigol (better specs)
The 'winner' seems to be the Siglent. Would you agree?"
Comments?

[DaJMasta]

Never used one.  They've been popular for a bit because they tend to be cheaper than many standard SAs while also including some demodulation capability, but I wasn't too concerned with that when I was looking, so i didn't look into them deeply.

[Bicurico]

The R&S CMU200 or R&S CRTU are great devices for amateurs.

They offer a pretty good spectrum analyzer with an incredible fast refresh rate. The freqeuncy range goes all up to 2.7GHz and overall the performance is R&S style.

However, these devices are NOT spectrum analyzers! They were made for the testing of mobile phones and mobile network. The spectrum analyzer and signal generator just happen to coexist with that functionality.

The drawbacks are:

1) The spectrum analyzer functionality is kind of reduced: no modern measurements, no waterfall diagramm (spectrogram), no screen capture to USB, no RJ-45 for remote operation.
2) The signal generator is a signal generator. Great to have, but it is NOT a tracking generator!
3) These devices were popular for their very affordable price, when Nokia (or other) plants closed and people selling them apparently were not aware of the spectrum analyzer/signal generator function. They can still be purchased at great prices - especially in the USA, but the avareage "buy it now" price in Europe is a bit too high, in my opinion.
4) They are USED devices, which means: out of calibration or at least the last calibration was years ago, possible defective components, cosmetic damage
5) They are huge! They occupy 4 slots in a 19" rack. You could line up around 5-6 Siglent SSA3021X behind each other within the same space.

The Siglent SSA3021X is a game changer, when it comes to affordable spectrum analyzers.

They are sold at a very competitive price, can be hacked for full options and increased frequency range (3.2GHz).

The screen is huge, GUI is intuitive and responsive, firmware is pretty decent and it basically just works as expected.

Sure, if you follow all threads in this forum, you will find people complaining about bugs and performance, but in general, the devices ship within specs and are of great quality.

At this price you won't find anything similar. Period.

It is a game changer, because many hobby users are now wondering if it is still reasonable to purchase a second hand device at the same cost as a brand new Siglent SSA3021X, which comes with 2 years of warranty.

I have both: SSA3021X and R&S CRTU. I am an amateur and like to explor the RF for transmissions, test modulators, search for satellite feeds, amateur radio transmissions, etc.

I use both devices:
- I like the CRTU for it's simplicity, responsiveness, fast refresh rate. Also, it seems more robust. You get different RF input connectors, which accept different signal power levels, so it gives me more confidence that I won't damage anything.
- I like the SSA3021X because of it's sheer amount of functionality, waterfall diagram (great for monitoring). Also, Siglents great SCPI implementation allowed me to program my own spectrum analyzer application for this device. My software allows for additional measurements like automatic realtime satellite identification (more info: http://vma-satellite.blogspot.pt/2017/03/vma-simple-spectrum-analyser-for.html).

To sum it up:

- If you want a spectrum analyzer now and the SSA3021X specs are within your needs, that's the device to go for, in my opinion.
- If you want a spectrum analyzer for hobby/amateur/educational purposes you may want the SSA3021X or you may want to search for a good deal on a second hand unit. In this case, I am not sure if the CMU200/CRTU is the right device. Perhaps it is better to look for a "real" spectrum analyzer, but make sure it is modern (TFT screen, don't go for anything with a CRT!) and you are sure it is not broken or specs. See the thread on this forum about purchasing a second hand spectrum analyzer.

Hope this helps.

Regards,
Vitor

[amdnra]

Vitor: Thank you for taking the time for such an excellent reply and content. Really appreciate it. Lost of good information.
Now let me ask you a SA question: If you have a signal sitting at, say, -20db and you add an 'Attenuation Level' of -25db
then the signal on the screen should shift and show the signal sitting at -45db. In other words, the addition of attenuation should
show up as lowering the signal on the display by that much. Let me know if this is the case. It sounds reasonable but I'm just not sure.
Please advise.

[nctnico]

With some patience you can get really good deals on Ebay. My current spectrum analyser is an Advantest R3477 which has a TFT touch screen, screencaptures to USB , network connectivity and last but not least: 15GHz bandwidth. It cost me the same as the basic version of the Siglent SSA3021X.

[RoGeorge]

Seems like everybody is in the SA fever these days.
I was looking for an SA last week too. Found some nice equipped R&S CMU200/CMRU units from India then, at a closer look, I was put down by some specs I don't fully understand:

Can anybody please comment this CMU/CMRU specs?
Examples from CMU200 specs http://www.upc.edu/sct/documents_equipament/d_175_id-448.pdf

- pag 7/48, CMU200 Phase noise <-100dBc, 1Hz (SSB f<2.2GHz, la offset 20 KHz to 250 KHz). Is this OK? I don't know the expected performance for GHz range, but a Rigol DG4102 DDS has a Phase noise <-115dBc, 1Hz (SSB f=10MHz, la offset de 10 KHz)

- pag 10/48 DANL <-95...-100dBc. I don't understand why dBc instead of dBm, why 1KHz instead of 1 Hz normalized band of 10-100 Hz?
E.g. the smallest Rigol SA, DSA815, have less than -110 dBm, while an R&S from the FPC1000 series have DANL less than -130...-150dBm, and this outstanding performance is without preamplifier.

I'm a total RF noob, so I might be totally missreading the specs. Are those just average, or the specs are very good but I don't understand them properly?
For a quarter million USD machine (in its glory days), I am expecting for top notch performance, but does it still is top-notch?

Whoever understand those datasheet numbers, please clarify them for mere mortals.
 

[tmbinc]

(I still owe Bicurico a half-working CRTU; and the primary reason is that they are a mess to ship, especially when disassembled.)

The CRTUs can be really cheap when obtained at an auction. They regularly don't sell or go for $100. But don't be fooled - once you add all the costs (including the stuff you bought because you were already bidding at that auction, but don't really need), you pretty much pay the ebay price...

They don't have stellar RF specs, but they are pretty damn useful for the mentioned reasons. For example, they have the wideband RF power sensor, the RF signal generator, a few basic signal analysis functions (high resolution frequency selective counter) that can be pretty useful.

What they cannot do is more advanced signal analysis. But if you want that, then nothing beats an SDR equipment or a (higher-end) scope connected to the IF output.

Yes, they used to cost a quarter million $, but that was not because of their RF performance, but because of the specialty of the software that they are used with (GSM/WCDMA tests); they are a massive pain to develop with only a few users worldwide. R+S pretty much took whatever RF hardware they had (CMU200) and made a few tweaks, because at $250k (or whatever their price was), it simply doesn't matter if the hardware is not optimized.

Unfortunately that doesn't mean that you got $250k worth of RF equipment...

[amdnra]

Anybody feel like giving this question a shot?
Now let me ask you a SA question: If you have a signal sitting at, say, -20db and you add an 'Attenuation Level' of -25db
then the signal on the screen should shift and show the signal sitting at -45db. In other words, the addition of attenuation should
show up as lowering the signal on the display by that much. Let me know if this is the case. It sounds reasonable but I'm just not sure.
Please advise.Thanks

[DaJMasta]

Could depend on the analyzer and whether the attenuator was internal.  For example, if you don't have an internal attenuator or you're using a -25dB attenuator on the input signal, you'll see the -20dB signal at -45dB provided the attenuator is rated at the frequency in question and the SA can see that input level (and I don't know of one that can't).  If you're using an internal attenuator, it probably won't display -45dB, it will display -20dB.  Since it knows what attenuation is switched in, it tells you what the signal is at the input (this is how you get high power input to still be readable on the SA).  In the same vein, if you have an external 20dB amplifier, your -20dB signal (and the noise floor) will be 20dB higher on the SA, but if it's an internal preamp that's being switched in, the SA will display the signal level at the input and correct automatically for the gain of the preamp.

[tautech]

Anybody feel like giving this question a shot?
Now let me ask you a SA question: If you have a signal sitting at, say, -20db and you add an 'Attenuation Level' of -25db
then the signal on the screen should shift and show the signal sitting at -45db. In other words, the addition of attenuation should
show up as lowering the signal on the display by that much. Let me know if this is the case. It sounds reasonable but I'm just not sure.
Please advise.Thanks

Exactly this ^ for SSA3000X models to a max attenuation of 51 dB.
Adding external attenuation must be manually deducted from the signal level. 

[usagi]

the R&S CMU200 can safely eat up to +47dbm, one of the reasons I got one. as pointed out the cmu doesn't have a true tracking generator, but the signal generator can be used to (slowly) sweep so you can still do stuff like characterize filters. most of the used ones are loaded to the gills with options for cellphone testing, since they were used by equipment manufacturers and telcos. you are getting what used to cost $300,000 or more for around $1000. the signal generator is really nice, very flexible with many modes and options.

the FSH6 doesn't have the sweep time limitations of the anritsu. per the manual min sweep time is 1ms and it sweeps 600mhz in 20ms. it's a true 6ghz spec analyzer with tracking generator. if you're interested, pm me.

[rf-loop]

Anybody feel like giving this question a shot?
Now let me ask you a SA question: If you have a signal sitting at, say, -20db and you add an 'Attenuation Level' of -25db
then the signal on the screen should shift and show the signal sitting at -45db. In other words, the addition of attenuation should
show up as lowering the signal on the display by that much. Let me know if this is the case. It sounds reasonable but I'm just not sure.
Please advise.Thanks

Exactly this ^ for SSA3000X models to a max attenuation of 51 dB.
Adding external attenuation must be manually deducted from the signal level.

Yes, but then also:
Just use SSA level offset setting for correct external attenuator. Only disadvantage is that unlike internal step attenuator every step, external attenuator do not have flatness and level calibration data table inside SSA for automatic correction. But if user can trust external 20dB attenuator is enough flat over frequency band then user can set this offset. (if external attenuator frequency response is not enough flat but known, then user can make correction table and enter this table to SSA so that it can do also wanted flatness corrections for this individual external attenuator or other circuit as external amplifier etc..

[amdnra]

To clarify: I originally should have said "Internal Attenuation Level" (as opposed to it being external).

Could depend on the analyzer and whether the attenuator was internal.  For example, if you don't have an internal attenuator or you're using a -25dB attenuator on the input signal, you'll see the -20dB signal at -45dB provided the attenuator is rated at the frequency in question and the SA can see that input level (and I don't know of one that can't).

If the attenuator in question is Not internal, How's the SA going to 'know' and compensate for the measurement and show -45db?

If you're using an internal attenuator, it probably won't display -45dB, it will display -20dB.

So, is the operator supposed to keep track of what attenuation level is set on and mentally compute for the true input signal strength? I would have thought the SA automatically scales the display for the internal attenuation factor.

 

Since it knows what attenuation is switched in, it tells you what the signal is at the input (this is how you get high power input to still be readable on the SA).

How is the SA going to 'tell' me what the signal is at the input? By displaying a number off to the side OR by displaying a graph showing the signal at the proper level?

[nctnico]

So, is the operator supposed to keep track of what attenuation level is set on and mentally compute for the true input signal strength? I would have thought the SA automatically scales the display for the internal attenuation factor.

No, the SA will compensate the display for whatever internal attenuation or gain it uses.

[amdnra]

If you're using an internal attenuator, it probably won't display -45dB, it will display -20dB.

Therefore his statement above that I quoted is incorrect? That is what I thought...

[nctnico]

If you're using an internal attenuator, it probably won't display -45dB, it will display -20dB.

Therefore his statement above that I quoted is incorrect? That is what I thought...

It is written in a confusing way. If you apply a +20dBm signal to a spectrum analyser it will display this as a +20dBm signal no matter what internal attenuation or gain it uses. The spectrum analyser knows it's internal attenuation and gain settings so it compensates the numbers on the display so you don't have to worry about that.

[amdnra]

It is confusing alright ! 
I was playing with an Anritsu MS2721B Spectrum Analyzer at work and it has an Internal Attenuation Level control.
I fed it a signal of -25db.
Then I changed the Int Att Level to 10db, 20db and then to 0db but it did not affect the graph or apparent signal strength...It remained at -25db
Then of what use is this internal Att Level control? Or is the unit not working properly (I wouldn't be surprised if that was the case)?

[CJay]

Same with all 'problems', define your requirements and then match the specs.

You can get a *lot* of SA second hand for your budget but, for my needs and if I had the budget, I'd definitely be paying the Siglent a lot of attention (and have done, there may be a sell off of 'stuff' to fund the purchase of one).

Bear in mind that service info and even software/firmware for a lot of recent SA's and indeed lots of other test gear is just not available to anything other than 'replace this board', return to base level so buying second hand without a warranty might become a very expensive lesson.


I've been extremely lucky and got great bargains but always with the worry that they'd be scrap, I have now got two CMU200s and find them excellent for my current needs but they lack a proper tracking generator (there is a seller on eBay who has designed a TG for them) and are only good down to 10MHz and up to 2.7GHz which may or may not become a problem for me soon.

[nctnico]

It is confusing alright ! 
I was playing with an Anritsu MS2721B Spectrum Analyzer at work and it has an Internal Attenuation Level control.
I fed it a signal of -25db.
Then I changed the Int Att Level to 10db, 20db and then to 0db but it did not affect the graph or apparent signal strength...It remained at -25db
Then of what use is this internal Att Level control? Or is the unit not working properly (I wouldn't be surprised if that was the case)?

The attenuation level is usefull when dealing with the noise floor of a spectrum analyser. If you attenuate a small signal too much it may dissapear under the noise floor. If you can see the attenuator is on, you can switch it off or (if available) enable a pre-amplifier. Most spectrum analysers set the attenuator to 10dB by default to have some input protection.