Which Spectrum analyzer would you get?

[pascal_sweden]

But even when leaving the question of Siglent's firmware quality aside, there's always the 2nd hand market. While the performance of the SSA3000X seems to be perfectly fine for a low-end SA, there's still the option of buying a 2nd hand big brand SA for the same or even less money, with a larger bandwidth and much better RF performance. And it would be silly to not at least consider that option.

You have a point! I will definitely consider looking at used HP gear or R&S gear.

Considering used HP spectrum analyzers. There seems to be such a wide variety of different models.
Some examples: HP 8559A, HP 8560E, HP 8590A, etc.

Any guidelines on what their respective age is, where to start, and which one to pick?

[KJDS]

The 856X series has better phase noise and better (narrower) resolution bandwidth than the 859X series. That's usually reflected in the price. I suspect that it would usually end up cheaper buying an E4406A if you wanted to look close in at modulation and an 859X for broadband sweeps. It's rare for either the 6x or 9X to come with the tracking generator. If a business is buying something to measure insertion loss then it usually also wants to measure return loss and it's far more effective buying a Network Analyzer so that there's reasonable accuracy in the measurement.

Generally Anritsu and R&S kit will be cheaper than the equivalent HP, however support is often more difficult to come by.

If you need a tracking gen, then either a noise source can be used to give an indication, or else if there is a 1st LO output then this can be mixed with a fixed signal to give a tracking gen.

[rf-loop]

But even when leaving the question of Siglent's firmware quality aside, there's always the 2nd hand market. While the performance of the SSA3000X seems to be perfectly fine for a low-end SA, there's still the option of buying a 2nd hand big brand SA for the same or even less money, with a larger bandwidth and much better RF performance. And it would be silly to not at least consider that option.

You have a point! I will definitely consider looking at used HP gear or R&S gear.

Considering used HP spectrum analyzers. There seems to be such a wide variety of different models.
Some examples: HP 8559A, HP 8560E, HP 8590A, etc.

Any guidelines on what their respective age is, where to start, and which one to pick?

8590A  throw away, do not even think
8559A  throw away this boat anchor exept if you want collect for museum (also need be careful with its mechanics because it is very fragile after it come old)
8560E  it have still value in use (depending also options)

Do not  buy if seller can not do tests and do these tests what you ask and you need know what to ask. Exept if you know these equipments so that the thumb is not in the mouth if need repair, adjust etc,  and you can smell out what is more like ok and what is in "awoid" class.  These are best to buy locally and if there is possible to do own tests.

Btw, R&S do really nice SA's.

(HP) Look serial number prefix and add 60 (im not sure if this works with all models and years)

[VK5RC]

The sheer number of 856x SAs out there (quite a bit of website fixit  if needed) and not too many unobtanium parts and HP-AG-KS tendency to keep manuals on their website has me voting with my feet to an older HP SA.
The Siglent and Rigol gear do represent amazing value.

[joeqsmith]

OP seems to want to do some sort of network analysis with the TG option.  The 3589 has a built-in generator and some limited measurements.  Really, I need to write some software for it but never had the need.  Has software for SOT.   

Plots  showing a short and 50 ohm terminator.   

Oops. Showed the same picture twice due to lack of coffee.   So here is the actual cal'ed 50 ohm then a crappy old Ethernet 50 ohm terminator.  As you zoom in on a Smith chart, it switches to a polar chart for some reason.   Last picture showing a Showa SF0922B 9.0MHz crystal filter.   

For me, the down side to this old beast is that it is limited to 150MHz. 

[Performa01]

But then, last image.

This old gear beats this unknown modern SA clearly. Why. Answer is very simple and straight. Phase noise.  (Btw, this phase noise level can not reduce using better External Reference- major part of pn  do not come from reference )
If do MinHold  and TraceAverage long time it can show -90dBm in this same setup but this is so borderline that I do not want even show. Even this -80dBm here need bit averaging for better and more accurate result.

Well, it’s about time that I pull out some real high end SA then 
Also an “unknown” instrument, no extra resources, no life, no nothing. Just high end. 

Despite having such a fantastic instrument, I lack the pair of high end RF-synthesizers required to take on the challenge of placing a tiny signal in the middle of a modulation spectrum. So I had to make do with just one signal, but with a little more complex modulation signal (XXX_-11dBm_10MHz_AM_15_20_50Hz_max)



This screenshot shows a -10.7dBm 10MHz signal (marker 1) that is amplitude modulated with three signals at the same time:

1. 20Hz pulse, 10% duty cycle (marker 2) in an attempt to resemble what joeqsmith has shown us.
2. 15Hz sine (marker 3) just for the “little extra”.
3. 50Hz mains hum (marker 4) which cannot be missing in any high end test setup!

Max. detection and no video averaging.
Please note the vertical scaling which is 12dB/div. so you can see the extra 20dB down to -130dBm. Otherwise you might have missed the mains hum… 

[tautech]

Does the Siglent have a way to demodulate and listen to AM, FM...?     

Yep.
There's no mention of it in the datasheet however the manual describes it:

3. Earphone Jack
The analyzer provides AM and FM demodulation. Insert the earphone to the jack to acquire the audio output of the demodulated signal. You can turn on or off the earphone output and adjust the volume via Demod ->Volume.

The audio jack is on the lower front panel alongside the USB socket, easily visible in the website pics.
http://www.siglentamerica.com/pdxx.aspx?id=1546&T=2&tid=227

[joeqsmith]

Does the Siglent have a way to demodulate and listen to AM, FM...?     

Yep.
There's no mention of it in the datasheet however the manual describes it:

3. Earphone Jack
The analyzer provides AM and FM demodulation. Insert the earphone to the jack to acquire the audio output of the demodulated signal. You can turn on or off the earphone output and adjust the volume via Demod ->Volume.

The audio jack is on the lower front panel alongside the USB socket, easily visible in the website pics.
http://www.siglentamerica.com/pdxx.aspx?id=1546&T=2&tid=227

Dang, I bet it even sounds better than the 496! 

I am looking on ebay and drooling...   Lot of cheap SAs out there, so much better than I have.   Still I would get a digital one.  I wonder if any of them can decode digital TV.  Maybe I could get it past the wife then. 

[tautech]

Does the Siglent have a way to demodulate and listen to AM, FM...?     

Yep.
There's no mention of it in the datasheet however the manual describes it:

3. Earphone Jack
The analyzer provides AM and FM demodulation. Insert the earphone to the jack to acquire the audio output of the demodulated signal. You can turn on or off the earphone output and adjust the volume via Demod ->Volume.

The audio jack is on the lower front panel alongside the USB socket, easily visible in the website pics.
http://www.siglentamerica.com/pdxx.aspx?id=1546&T=2&tid=227

Dang, I bet it even sounds better than the 496! 

I am looking on ebay and drooling...   Lot of cheap SAs out there, so much better than I have.   Still I would get a digital one.  I wonder if any of them can decode digital TV.  Maybe I could get it past the wife then.

Not if she can't see the pictures. 

[pascal_sweden]

In that case you would need to get an MPEG analyzer from Tektronix or the like
http://www.tek.com/mpeg-test-video/mpeg-analyzer

Note that this only works on unencrypted video streams.

Most video streams of your cable provider will be encrypted.
The headers of the MPEG-2 streams are in the clear, but the video payload is encrypted.

[jjoonathan]

Does the Siglent have a way to demodulate and listen to AM, FM...?     

Yep.
There's no mention of it in the datasheet however the manual describes it:

3. Earphone Jack
The analyzer provides AM and FM demodulation. Insert the earphone to the jack to acquire the audio output of the demodulated signal. You can turn on or off the earphone output and adjust the volume via Demod ->Volume.

The audio jack is on the lower front panel alongside the USB socket, easily visible in the website pics.
http://www.siglentamerica.com/pdxx.aspx?id=1546&T=2&tid=227

I remember seeing a "Demod Time" setting in a screenshot of the SSA3000x somewhere. If it works like "Demod Time" in the DSA815, then it can't demod while performing a sweep, so it will play radio for the demod time, interrupt radio to perform the sweep, and repeat.

[tautech]

Does the Siglent have a way to demodulate and listen to AM, FM...?     

Yep.
There's no mention of it in the datasheet however the manual describes it:

3. Earphone Jack
The analyzer provides AM and FM demodulation. Insert the earphone to the jack to acquire the audio output of the demodulated signal. You can turn on or off the earphone output and adjust the volume via Demod ->Volume.

The audio jack is on the lower front panel alongside the USB socket, easily visible in the website pics.
http://www.siglentamerica.com/pdxx.aspx?id=1546&T=2&tid=227

I remember seeing a "Demod Time" setting in a screenshot of the SSA3000x somewhere. If it works like "Demod Time" in the DSA815, then it can't demod while performing a sweep, so it will play radio for the demod time, interrupt radio to perform the sweep, and repeat.

Exactly ^^^

Here's a shot using the human body as an antenna for the most powerful Auckland AM station that's ~15 km away.



Audio was crystal clear on #1 volume setting for the bits heard at the end of each sweep. 
Subject station:
http://www.newstalkzb.co.nz/news/

[rf-loop]

Does the Siglent have a way to demodulate and listen to AM, FM...?     

Yep.
There's no mention of it in the datasheet however the manual describes it:

3. Earphone Jack
The analyzer provides AM and FM demodulation. Insert the earphone to the jack to acquire the audio output of the demodulated signal. You can turn on or off the earphone output and adjust the volume via Demod ->Volume.

The audio jack is on the lower front panel alongside the USB socket, easily visible in the website pics.
http://www.siglentamerica.com/pdxx.aspx?id=1546&T=2&tid=227

I remember seeing a "Demod Time" setting in a screenshot of the SSA3000x somewhere. If it works like "Demod Time" in the DSA815, then it can't demod while performing a sweep, so it will play radio for the demod time, interrupt radio to perform the sweep, and repeat.

Yes. And, of course.
For listening AM or FM then need stop sweeping. This is single pathway machine, not radio with rf band panorama display.
Tune center freq to station and press zero span.  Depending station  AM or FM on. note, RBW filter setting! This is just as radio IF filter width. If someone do not know how to do, just start point for AM use 10kHz RBW  and for (wide)FM example 100kHz or EMI filter 120kHz.

One day I did many things in lab without needing SA at all, mostly paper work. IJust for fun I use SA as radio and listening some SW commercial stations and some FM.   
Sure it can not do what my Racals etc can do, or not even what cheap Tecsun can do.
There is no SSB, there is not beat osc for CW and so on.

And then very important thing is that there is not any kind of automatic RF and Audio level control. (if go more deep there is some but not what radio need). With fading signal there need continuously adjust attenuator, PA and Audio level.

Note: With antenna do not destroy SA input circuits. Use some protection. Without protection it is gambling.
Also it is good to realize that input is always widely open for full  frequency band. If you listen 12MHz weak AM station  and there is high signal in GHz range it also come directly to mixer.
Maximum: RF power 33 dBm, 3 minutes, input attenuation >20 dB !

[G0HZU]

I don't see why Siglent, unless it is defective, would not match main performance characteristics of HP/Agilent or Anritsu on such limited, low frequency range of 0.01 - 2.1 GHz. 

I guess it depends on how hard you look and where you look for the limits of the performance of any analyser.

If I had the Siglent analyser here I think I could find several aspects of its performance that are going to be much worse than a classic HP/Agilent analyser from >=15 years ago. But some things would hopefully be better with the Siglent because the Siglent analyser has a modern digital IF.

One aspect of the Siglent analyser that worries me a bit is that they have used a frequency doubler in the first LO. This would make it harder to maintain a low internal spurious qty/level.

[pascal_sweden]

Please define "several aspects". If there are several aspects, you can at least name some specific aspects, to have the figures on the table. Even better, get both units, and make the measurements to confirm it with a test report.

[G0HZU]

Mismatch uncertainty, far out phase noise, attenuator performance, dynamic range at low frequencies, internal spurious levels and qty, external spurious responses.

[nctnico]

Mismatch uncertainty, far out phase noise, attenuator performance, dynamic range at low frequencies, internal spurious levels and qty, external spurious responses.

Would it be possible to determine (or at least make an educated guess about) things like these based on the datasheet of a typical spectrum analyser or do you really need to measure this yourself?

[KJDS]

Please define "several aspects". If there are several aspects, you can at least name some specific aspects, to have the figures on the table. Even better, get both units, and make the measurements to confirm it with a test report.

One crucial thing with a spec ana is the spurious free dynamic range.

No matter what signals you add in the front end, then so long as they are not above the top of the screen then there should never be a spurious signal visible. There usually is, somewhere, however it will often take some searching for. Also, signals outside the specified frequency range shouldn't ever cause a signal to appear on screen.

Let's take a look at a classic scheme for a 1GHz analyzer. The input will come into a mixer via some switched attenuators to be mixed with the first LO. The first LO will typically run from 1.4 to 2.4GHz to give a 1.4GHz 1st IF. Put in a signal at 1.4GHz and it will leak through the mixer into the 1st IF, so it needs a lot of filtering, probably at least 70 to 80dB of filtering. Similarly, frequencies out at 3GHz could also be mixed down to 1.4GHz when the 1st LO is at 1.6GHz, so the front end filter needs to stretch a long way up with a lot of rejection.

Additionally, there will be a point where the third harmonic of the input will mix with the second harmonic of the LO to land at 1.4GHz. The mixer therefore needs to be sufficiently high performance in order to ensure that the spurious it generates are of a very low level. High performance mixers aren't cheap, and need driving with a high level LO signal so the amp to drive it becomes more expensive and the shielding requirements to stop the higher level signal going places it shouldn't become more demanding. If you know the architecture, you can calculate where the 3-2 harmonic will fall, put in a signal of the appropriate level and go hunting for the spurious. On a cheaper instrument it's much easier to find.

The other issue is phase noise and frequency stability. This is far less of an issue on modern instruments than it was in earlier times. Fractional N PLLs made things much easier, then with a digital IF it becomes simpler still, just step the LO every few MHz and let the digits sort out the IF filtering.

There's lots of other issues too, a few of which have been highlighted in the previous post, but if you want to make a measurement of a filter insertion loss and the filter only has a 12dB return loss and the analyzer only has a 20dB match then you're left with some horrible uncertainty of 1dB, which is hopeless if your filter only has 1dB insertion loss.

[G0HZU]

Mismatch uncertainty, far out phase noise, attenuator performance, dynamic range at low frequencies, internal spurious levels and qty, external spurious responses.

Would it be possible to determine (or at least make an educated guess about) things like these based on the datasheet of a typical spectrum analyser or do you really need to measure this yourself?

The datasheet can be a good guide but there will often be some weasel wording to hide the areas where the analyser will be weak. For example, the dynamic range is often quoted within certain frequency bands and is left undefined in other places. Also, some datasheets quote guaranteed specs and others quote typical specs (or both) and some show nominal figures.

Note that the weasel wording tactics are used by all manufacturers including the big names so I don't mean to imply that this is only carried out by Siglent or Rigol or anybody else

[rf-loop]

There is so many near carrier tests so here is bit more far from carrier.  0-5MHz

Trace A
In attached image there is 989MHz 0dBm carrier (checked that SA measure peak and result is 0dBm. This measured using P-Pk detector using 20dB attenuator setting (also checked with HP Power meter so that least not significant error)

Ref level changed to -80dBm.

Then measured sideband noise figure using Video Average detector (also Noise marker function need it)
Because signal is 0dBm Marker levels can read also as dBc/Hz  (in this function Siglent use also correction to 1Hz noise bandwidth using 2.5dB correction what is because RBW gaussian style shape and detector etc things need normalize for (rayleigh distribution) noise . More deep explanation can find example from HP/Agilent app notes)

RBW and VBW 10Hz.

SA with its  internal standard reference.

Signal source HP8644B with its  internal Optional High Stability OCXO.

Cable between G and SA was bit better shielded Sucoflex for keep possible environment noise away.

Trace C:
All same as trace A but SA input terminated (Narda 50ohm N.)

Both traces also averaged using trace average.

Not very bad but if they some day launch version where also 1Hz RBW is activated lets hope they can also reduce phase noise.

[Jester]

I can get a HP 8566 locally (supposedly) in working order for $400.

Is this a good unit?

Anything to look for?

I want to use this for pre-compliance EMI testing

[G0HZU]

Probably best to start a separate thread about the HP8566B... But FWIW I have one here and you could use this analyser for EMC/EMI testing but it doesn't have an internal quasi peak detector. You have to buy an external box for this.

The HP8566B is a fabulous old analyser from HP in terms of RF performance but it is also very big and extremely heavy at about 50kg. It also makes an awful lot of fan noise and consumes a lot of power. You don't want to use this analyser in a small room unless you can live with the fan noise and the heat it gives off.
It's also going to need fairly regular maintenance in terms of swapping out old failed parts (usually wet tantalum caps) and the CRT is a weak point in that the CRT can lose brightness and focus over time.

I've had mine for about 5 years and it is still running fine although it's been apart for minor repairs/tweaks several times. None of the repairs were costly in terms of parts.
My advice would be to look elsewhere if you just want to do EMC/EMI testing but $400 is a very low price for a working HP8566B. I'd expect it to be tired/faulty/damaged for that price?

[G0HZU]

I don't see why Siglent, unless it is defective, would not match main performance characteristics of HP/Agilent or Anritsu on such limited, low frequency range of 0.01 - 2.1 GHz.

If you look at rf-loop's phase noise plot for the Siglent analyser at 1GHz then the phase noise at 100kHz offset is -98dBc/Hz.

On my old 1800MHz Advantest analyser here, the (measured) phase noise for a 1GHz test signal at 100kHz offset is -128dBc/Hz . So that is 30dB better than the Siglent. This analyser was first produced in 1981. It's a massive (and ugly) thing, even bigger and heavier than the classic HP8568B and it cost a lot more money when new than the HP8568B. It uses a big and expensive YIG oscillator for the LO and this is why it has very low phase noise at 100kHz offset.

Obviously, it's a bit unfair to expect the Siglent to match the Advantest's YIG oscillator but there is a 30dB margin in noise performance here...

[pascal_sweden]

What is the difference between HP8568B and HP8566B?

Is there an easy number and letter scheme to recognize the differences between all these model numbers? Some end with A, others with B or E. Some start with 856x, 859x, 892x, etc.

[G0HZU]

Probably best to read the official HP literature about the HP8568 and HP8566 but the HP8566 was HP's ultimate 'doomsday' analyser designed for use up past 20GHz where performance was more important than the huge dev costs or the massive purchase price. It's dev work traces back to the late 1960s and dev work on the HP8568 began later, some time in the early 1970s. The HP8568 was a 1500MHz version of the 22GHz HP8566.

http://hpmemoryproject.org/technics/bench/8568/bench_8568_home.htm

These HP8566A and HP8568A analysers were released on sale in the late 1970s and caused an absolute sensation. They were miles ahead of anything else at the time. The B version of each was released a few years later and it has a Motorola 68000 based processor card and this runs the GUI and display a bit slicker than the A version that used HP's own processor.

They look very similar on the outside and the top display section is supposed to be the same. But there are lots of differences in the RF section and not much in common in the signal path. But both analysers were regarded as the industry standard (for their respective frequency ranges) for about two decades.

The HP8560 is a portable  (but still very heavy) analyser and this covers up to 2.9GHz.
http://equipland.com/objects/catalog/product/extras/6497_8565E.pdf

The HP8561 is similar to the 8560 but it covers up to 6.5GHz.
The HP8562 is similar to the 8560 but it covers up to about 13GHz.
The HP8563 is similar to the 8560 but it covers up to about 22GHz.

The A versions have 'all analogue' RBW filtering but the E versions are newer and ise digital RBW filters for the narrowest RBW filters and this means they are a bit more accurate and sweep faster when these filters are selected.

Note: You need to be wary of the B versions of these little portable analysers here because I think they omit the front end image/preselector filtering on the upper frequency ranges on the B versions. I've never seen/used a B version but I'd avoid the B version if I were you