Which Spectrum analyzer would you get?

[pascal_sweden]

Now here comes the big question: does that extra 30dB justifies the difference in weight of a ton and the extra bench space of half a meter?

Is it really that critical to have that low phase noise for most applications at those frequencies?

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...

[G0HZU]

Now here comes the big question: does that extra 30dB justifies the difference in weight of a ton and the extra bench space of half a meter?

Is it really that critical to have that low phase noise for most applications at those frequencies?

It depends on what you are designing, I guess. Back in the early 1990s at my place of work I was designing a lot of high performance RF downconverters and I was originally supplied with an HP8568B analyser (supplied free by the customer)  to help the design proving. So synthesiser phase noise and downconverter spurious testing demanded a very high performance analyser.

The HP8568B didn't have the same phase noise performance at 100kHz offset as the TR4172 Advantest analyser and also the HP8568B didn't have the same IP3 performance or 2HI performance as the Advantest analyser. So the customer supplied me with a free £££ TR4172 analyser to replace the HP8568B because I needed the phase noise performance and also the SFDR of the TR4172 when operating down at the IF output frequency of my downconverters.

That doesn't mean the TR4172 is a better analyser than the HP8568B, but it did outclass it in certain areas. Basically, I needed every last drop of performance these analysers could deliver and neither of them were good enough for some tests. Often one was better than the other for certain tests. But they were still very special pieces of equipment and nothing else could compete with them apart from the 400MHz Marconi 2382 analyser. That had very good performance too.

Here at home I've still got the very same TR4172 analyser and the very same HP8568B that the customer donated 25 years ago. They were declared obsolete at work many years ago and I now own them

[pascal_sweden]

Do some of these HP8566B or HP8568B come with a built-in Tracking Generator option, or does this always have to be purchased separately as a rack option to be put on top of the display?

What is the model number of the Tracking Generator that fits best for the HP8566B and HP8568B respectively?

[G0HZU]

Do some of these HP8566B or HP8568B come with a built-in Tracking Generator option, or does this always have to be purchased separately as a rack option to be put on top of the display?

What is the model number of the Tracking Generator that fits best for the HP8566B and HP8568B respectively?

HP produced a (big external) generic Tracking Gen module for all of the HP856x range (except the HP8568) in the form of the HP85644A and HP85645A .
There is an older external T gen module that supports the HP8568B but I've forgotten the number.

But the HP 85644A and HP85645A are VERY expensive items. They were well over $20k USD when new 20-25 years ago and they still fetch big money today. I've got one here but I rarely use it in T Gen mode with my HP8566B. You connect it up via several BNC coax cables and then enter the analyser model number in the front UI by scrolling through all the analysers that are displayed/supported.

http://cp.literature.agilent.com/litweb/pdf/85645-90020.pdf

You can also use it as a crude signal generator as you can see below with my old HP8566B. It's almost as big and heavy as a HP8560 portable analyser!
I think it also supports some of the HP859x range of portable analysers.

[cncjerry]

Ditto, I have them both as well, 8568b and 8566b. 

Big, noisy, hot, etc, but they still have great performance and the best bang for the buck compared to new if you find one that has been well maintained.  $400 is too little for an 8566b.  If it looks complete I would snatch it up in a heartbeat.  It's worth more than that as a parts unit.

If I have the signal generation capabilities, I will setup the signals that I see in threads like these to see how the old HP analyzers compare.  Rarely does anything give better performance.  These new Siglents have some interesting capabilities for the money.

[KE5FX]

Do some of these HP8566B or HP8568B come with a built-in Tracking Generator option, or does this always have to be purchased separately as a rack option to be put on top of the display?

I built a nice one for my 8566 (described here), and you can also use one of DKD's modules to make the construction process easier.  But like G0HZU says, the "official" ones are still very pricy, especially the 85645A. 

My Tektronix 496P with a 5MHz span has a minimum of 10KHz RBW before it displays UNCAL.   It's not stable enough where I could decrease the span and RBW to manually take the data, nor is the resolution of the readout near good enough to track the tuning.  The readout is just good enough to let you know it's around a GHz.     

Set your TIME/DIV switch to 'Auto.'

[KE5FX]

Probably best to read the official HP literature about the HP8568 and HP8566 but the HP8566 was HP's ultimate 'doomsday' analyser

Interesting choice of adjectives -- that was its in-house code name.

[cncjerry]

Probably best to read the official HP literature about the HP8568 and HP8566 but the HP8566 was HP's ultimate 'doomsday' analyser

Interesting choice of adjectives -- that was its in-house code name.

Somehow that codename is so appropriate.

The TG for the 8568 was an 8444a with option 059 which added a 500Mhz internal oscillator.  I have one and it works well.  Also note that some of the later 8568B models went to 1.8G.

[pascal_sweden]

Taking into account the price tag of a working HP8566B spectrum analyzer in good condition, together with a working HP85645A tracking generator in good condition, the Siglent SSA3021X with built-in tracking generator, seems the most viable option, for hobby spectrum analyzer use

[dxl]

Taking into account the price tag of a working HP8566B spectrum analyzer in good condition, together with a working HP85645A tracking generator in good condition, the Siglent SSA3021X with built-in tracking generator, seems the most viable option, for hobby spectrum analyzer use

You can't compare that - the HP8566B and HP 85645A is a combination up to 22GHz (the TG up to 26.5GHz). I guess the Siglent doesn't do 22Ghz.

[G0HZU]

Probably best to read the official HP literature about the HP8568 and HP8566 but the HP8566 was HP's ultimate 'doomsday' analyser

Interesting choice of adjectives -- that was its in-house code name.

I must confess that it wasn't my choice of adjectives. So no coincidence here

I've seen the 8566 referred to as the 'doomsday machine' and the 'Ultimate Spectrum Analyser' in several docs from HP over the years. I assume they used these terms because it probably cost a lot of time/money to develop whilst the engineers were let off the leash to make the 'ultimate spectrum analyser'. I wonder if the various (non engineering) depts at HP affectionately coined it 'doomsday machine' because the project took a long time and consumed a lot of NRE costs. Maybe they had a few Sci Fi or Star Trek fans amongst the staff? 

[G0HZU]

Here's a couple of images of my old TR4172 at 1GHz with a test tone to show the phase noise at 100kHz offset. The first plot shows the phase noise at -124dBc/Hz but the noise level is limited by the 95dB log range of this old analyser.
So to measure the noise more accurately I've turned down the ref level 15dB in the second plot. This brings the noise well inside the accurate range of the logamp. Noise is best measured at least 15dB inside the effective range of a logamp. So this is why I've done this. It looks a bit scary but this is a known/valid/accepted method on this type of swept analogue analyser and you can see the phase noise is -128dBc/Hz at 100kHz at 1GHz test frequency.

Sadly, my camera does not take images of CRT screens very well so the images look a bit tired and gloomy. The display does look much nicer than this in reality when set to normal brightness. If I turn up the brightness the camera just blooms and the camera makes the trace quality look bad.

You might wonder what use this level of performance is if you have to resort to the ref level trick to overcome the limitation of the 95dB log amp range.. But when used like this the analyser was very useful when testing various VHF and UHF synthesiser designs for phase noise at 100kHz offset.

[KE5FX]

Here's a couple of images of my old TR4172 at 1GHz with a test tone to show the phase noise at 100kHz offset. The first plot shows the phase noise at -124dBc/Hz but the noise level is limited by the 95dB log range of this old analyser.
So to measure the noise more accurately I've turned down the ref level 15dB in the second plot. This brings the noise well inside the accurate range of the logamp. Noise is best measured at least 15dB inside the effective range of a logamp. So this is why I've done this. It looks a bit scary but this is a known/valid/accepted method on this type of swept analogue analyser and you can see the phase noise is -128dBc/Hz at 100kHz at 1GHz test frequency.

Pretty common to deliberately overdrive the analyzer a bit when measuring noise.  I've never used a TR4172 myself, but a user of my PN app sent me a patch for it a few years ago, along with a baseline plot:



Unfortunately he used a relatively weak test signal (-20 dBm) and 10 dB of RF attenuation.  So even though he overdrove the reference level by 10 dB, he got worse results than he should have beyond 30 kHz.

[cncjerry]

Maybe they should have called the 8568 the doomsday analyzer...

[pascal_sweden]

Shouldn't every spectrum analyzer come with a Tracking Generator by default, if you need to buy it anyhow separately later on?

Can you SA experts list a short overview of:

1) Tests that can be perfectly done without any TG at all, and neither any workaround to simulate TG;

2) Tests for which you strictly speaking need a TG, but where you can use a (cheaper) workaround approach such as an external device to take the role of TG;

3) Tests that can't be done at all without TG, and where a workaround is not feasible or recommended with regards to expected test results and accuracy;

[KE5FX]

Shouldn't every spectrum analyzer come with a Tracking Generator by default, if you need to buy it anyhow separately later on?

Can you SA experts list a short overview of:

1) Tests that can be perfectly done without any TG at all, and neither any workaround to simulate TG;

2) Tests for which you strictly speaking need a TG, but where you can use a (cheaper) workaround approach such as an external device to take the role of TG;

3) Tests that can't be done at all without TG, and where a workaround is not feasible or recommended with regards to expected test results and accuracy;

Short answer: if you need a tracking generator, you probably really want a network analyzer. 

Tracking generators are mainly good for aligning filters and making basic scalar measurements (magnitude only, no phase) on antennas. 

If you have a sweep generator, you can run essentially any test you could run with a tracking generator, but it will take (perhaps literally) forever.

TL;DR: if your budget is limited, consider a sweep generator, and if it's less limited, consider a VNA. 

[cncjerry]

You can also use the Peak Hold or Max Hold function with a noise source to get a view of a filter on an SA.  Never used a noise source with a reflection bridge on an antenna but it might work as well.

[PartialDischarge]

Id definitively buy a siglent because its hackable, will have a large user base of SDR/ham radio user enthusiasts, digital technology, its fast, nice screen, multiple cursors or USB support. So that its phase noise is worse
than some models from the 80s..., well I'm not building a satellite transceiver anyway... in short the siglent would meet my needs.

[Alexei.Polkhanov]

If you care about colors and USB rather than Noise Figure it makes me wonder what do you use Spectrum Analyzer for? My spectrum analyzer has LCD display with Windows 98 - I can play Tetris on it. People that need SA for "HAM Radio use", whatever it means, probably want to see that little moon bounce or asteroid bounce signal not being buried in noise.

You can take a photo of screen with your camera and color it up in Photoshop - can you bring the Noise Figure down?

[nctnico]

You can always hack a color TFT display into test equipment with a monochrome screen:

[PartialDischarge]

If you care about colors and USB rather than Noise Figure it makes me wonder what do you use Spectrum Analyzer for? My spectrum analyzer has LCD display with Windows 98 - I can play Tetris on it. People that need SA for "HAM Radio use", whatever it means, probably want to see that little moon bounce or asteroid bounce signal not being buried in noise.

You can take a photo of screen with your camera and color it up in Photoshop - can you bring the Noise Figure down?

I'll bite. I do care about phase noise, as I said the specs of the siglent are good enough for me, hell its *way* better than the Rigol 832 that lots of hobbyist use and no one seems to complaint about. I'm not trying to optimize one particular figure in the equipment, but get an overall "pass" or "very good" in each feature.

[PartialDischarge]

By the way, just to demonstrate that im not biased towards color screens or even against HP look at what I just bought, an audio spectrum analyzer from the 80s, an HP 3580a
An awesome piece of technology with digital storage,tracking generator, frequency log scale, 1dB/10dB/div or linear vertical scale,  and battery power option   

[G0HZU]

Unfortunately he used a relatively weak test signal (-20 dBm) and 10 dB of RF attenuation.  So even though he overdrove the reference level by 10 dB, he got worse results than he should have beyond 30 kHz.

Yes, the DANL for the TR4172 is typically -146dBm/Hz and up at 1GHz it's more like -145dBm/Hz.

With the 10dB attenuator selected this will degrade to -135dBm/Hz. So for a -20dBm test tone this becomes -115dBc/Hz so it's no surprise to see the noise plateau at this level in that plot.

However, that analyser looks to be a couple of dB worse than this and I wonder if it has a problem? A while back my TR4172 developed an age related fault in the IF1 section where the adhesive they used for some RF absorber had gradually eaten away at a printed resistor trace (used as the bias resistor for the IF1 amplifier) and this made my analyser slowly go deaf and it also lost a lot of SFDR.
I removed the adhesive and used RTV instead and I fixed the part of the PCB with the damaged resistor. The noise floor improved back to where is normally was and the system IP3 went back up to normal. I think the production people at Advantest glued the absorber in place and immediately assembled and sealed the IF1 module and the adhesive then caused corrosion as it cured and the chemicals stayed entombed inside the module. It also rotted the nearby PCB traces making them go soft like rotten paint and it also caused all the plated screws to corrode. The sections without the adhesive still looked shiny and new inside.

The same thing happened inside the attenuator module. The adhesive was used on some RF absorber ant it rotted through an adjacent PCB trace and the analyser then lost an attenuator setting (now repaired). So I suspect that every TR4172 will have this issue at some point. I have a spare TR4172 here and the same thing was happening in both modules in that one too (but not as bad, as the PCB traces were soft but still intact)

[KE5FX]

However, that analyser looks to be a couple of dB worse than this and I wonder if it has a problem? A while back my TR4172 developed an age related fault in the IF1 section where the adhesive they used for some RF absorber had gradually eaten away at a printed resistor trace (used as the bias resistor for the IF1 amplifier) and this made my analyser slowly go deaf and it also lost a lot of SFDR.
I removed the adhesive and used RTV instead and I fixed the part of the PCB with the damaged resistor. The noise floor improved back to where is normally was and the system IP3 went back up to normal. I think the production people at Advantest glued the absorber in place and immediately assembled and sealed the IF1 module and the adhesive then caused corrosion as it cured and the chemicals stayed entombed inside the module. It also rotted the nearby PCB traces making them go soft like rotten paint and it also caused all the plated screws to corrode. The sections without the adhesive still looked shiny and new inside.
The same thing happened inside the attenuator module. The adhesive was used on some RF absorber ant it rotted through an adjacent PCB trace and the analyser then lost an attenuator setting (now repaired). So I suspect that every TR4172 will have this issue at some point. I have a spare TR4172 here and the same thing was happening in both modules in that one too (but not as bad, as the PCB traces were soft but still intact)

Interesting, that does seem plausible.  Sounds like acetic acid in action...

[cncjerry]

You can always hack a color TFT display into test equipment with a monochrome screen:

Hey nicotine, how did you do that?