Low Cost PCB's Low Cost Components

Author Topic: typical specifications that an analog SA beats a FFT 'instantanous' SA?  (Read 1671 times)

0 Members and 1 Guest are viewing this topic.

Offline CopperCone

  • Frequent Contributor
  • **
  • Posts: 344
  • Country: us
Are there any typical specifications that a swept SA beats a real time spectrum analyzer (i.e. modern signal hound).

DANL does not seem to be it.

like, comparing a HP70000 to a modern PSX series (or greater), perhaps the new signal hound.
« Last Edit: June 14, 2017, 01:37:13 PM by CopperCone »
 

Offline KE5FX

  • Frequent Contributor
  • **
  • Posts: 624
  • Country: us
    • KE5FX.COM
It depends on what models you're comparing.  You may find that the front end components are better in the old analog SAs.  They'll be less sensitive (because they didn't tend to include preamps, which are common in units sold today) but they may have less internally-generated IMD and harmonic distortion, depending on a large variety of factors.  I don't think the differences will be huge, but they'll exist.

The wideband phase noise floor of the older models with YIG-tuned LOs will likely be quite a bit better than modern SAs that rely on monolithic PLL synthesizer chips with integrated VCOs.  Close-in phase noise should be about the same, again depending on models.  SDR-based models like Signal Hound's BB60 and upcoming SM200A are theoretically capable of much better close-in noise performance since the demands placed on their LO synthesizers are much less stringent.

The top-of-the-line analog models from the 80s-90s era will have much better spurious performance than most of the low-cost models on the market today.  If you see a signal on a Tek or HP analyzer, you can safely assume the signal is actually there.  If you see one on a small USB-based SA, it might or might not be.   That's what tends to happen when you get rid of 20 pounds of internal shielding.
 

Offline technogeeky

  • Frequent Contributor
  • **
  • Posts: 403
  • Country: us
  • Older New "New Player" Player Playa'
Are there any typical specifications that a swept SA beats a real time spectrum analyzer (i.e. modern signal hound).

DANL does not seem to be it.

like, comparing a HP70000 to a modern PSX series (or greater), perhaps the new signal hound.

Isn't the obvious answer (but perhaps so obvious, that we're excluding it): bandwidth?

Secondly, my first two guesses were basically the features at the limit -> small RBW, namely:
  • 1Hz RBW (the lowest RSA I've seen is 10 Hz, so this isn't far out)
  • Zero-Span mode (though RSAs can reconstruct this, can't they?)

I am interested in the is question too, and I have a few guesses; but since I didn't know for sure, I figured I'd just collect links to some of the state of the art examples of units, or examples of possibly greatest-of-all-time units.

Real-time Spectrum Analyzers (RSA):

Mixed (Swept-Tune Spectrum Analyzers with RSA components):

Swept-Tune Spectrum Analyzers:
« Last Edit: June 14, 2017, 03:50:23 PM by technogeeky »
 

Offline CopperCone

  • Frequent Contributor
  • **
  • Posts: 344
  • Country: us
Maybe its the HP 70000 for best in class.. Its large.. that or the other HP 8566A.

I am very interested in the new signal hound. It has 1THz/Second sweep speed.
Like I said in the other thread, I wonder what the POI will be on the new signal hound. I think the best POI's on the market are sub microsecond.

The RTSA king might be this?
http://www.keysight.com/en/pcx-x205211/real-time-spectrum-analyzers-rtsa?cc=US&lc=eng

you get 500MHz of real time bandwidth. I wonder if old stuff can hold a candle to PSX series. But it looks like your dynamic range is low (spur free).
« Last Edit: June 14, 2017, 03:16:07 PM by CopperCone »
 

Offline w2aew

  • Super Contributor
  • ***
  • Posts: 1629
  • Country: us
  • I usTa cuDnt speL enjinere, noW I aR wuN
    • My YouTube Channel
Quote
1Hz RBW (the lowest RSA I've seen is 10 Hz, so this isn't far out)
Zero-Span mode (though RSAs can reconstruct this, can't they?)

The Tek RSA5100B series can go down to 0.1Hz RBW.  RSAs can show you RF amplitude vs. time in a user defined BW, and the 5k can also do this in Real-Time, giving you up to 50,000 zero-span traces/sec, to see those infrequent transient time-domain disturbances.

Quote
...you get 500MHz of real time bandwidth. I wonder if old stuff can hold a candle to PSX series.

The RSA7100A gives you 800MHz instantaneous BW, plus long-term IQ streaming recording to RAID, optionally built into the unit:
« Last Edit: June 15, 2017, 05:44:05 AM by w2aew »
======================================
YouTube channel: http://www.youtube.com/w2aew
FAE for Tektronix
 

Offline CopperCone

  • Frequent Contributor
  • **
  • Posts: 344
  • Country: us
 :o

maybe we can afford one in 20 years
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2228
  • Country: gb
On its first frequency band (0 to several GHz?) the typical amplitude flatness across (say) a 100MHz span should be better on a decent swept analyser like the HP8566B when compared to a typical RTSA set to a 100MHz span. A healthy HP8566B should perform very well in this test but a typical RTSA will be quite poor in comparison. A lot depends on how well the RTSA has been calibrated/corrected internally. The HP8566B will excel at this test because of the swept architecture :)

Quote
1Hz RBW (the lowest RSA I've seen is 10 Hz, so this isn't far out)
I think the old Tek RSA3408A can go down to 1Hz RBW and below 0.1Hz in FFT mode on a narrow span.
 

Offline vk6zgo

  • Super Contributor
  • ***
  • Posts: 3729
  • Country: au
Not quite on topic, but some "digital sweep" SAs (not FFT type) like the IFR one I used a few years back have problems displaying a carrier which rapidly changes frequency over a large range.

The stored data to the DUT's PLL board was lost, & the PLL output frequency went from  433MHz to 470MHz(apparently its default frequency), in a very short time.

The  IFR's sample rate was such that this appeared as a series of fixed carriers across that range, sending us chasing our tails looking for "spurii".

When we looked at it with an "El Cheapo" Atten analog SA, the problem was immediately obvious.

Would an FFT display see the variation correctly?

 

Offline w2aew

  • Super Contributor
  • ***
  • Posts: 1629
  • Country: us
  • I usTa cuDnt speL enjinere, noW I aR wuN
    • My YouTube Channel
Not quite on topic, but some "digital sweep" SAs (not FFT type) like the IFR one I used a few years back have problems displaying a carrier which rapidly changes frequency over a large range.

The stored data to the DUT's PLL board was lost, & the PLL output frequency went from  433MHz to 470MHz(apparently its default frequency), in a very short time.

The  IFR's sample rate was such that this appeared as a series of fixed carriers across that range, sending us chasing our tails looking for "spurii".

When we looked at it with an "El Cheapo" Atten analog SA, the problem was immediately obvious.

Would an FFT display see the variation correctly?

With proper setup, most likely. But, even better is that the RTSA can seamlessly capture/record the signal vs. time, and then let you view spectrum vs. time (spectrogram), as well as frequency, phase and amplitude variations vs. time (similar to FM, PM, AM demod). 
======================================
YouTube channel: http://www.youtube.com/w2aew
FAE for Tektronix
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 8925
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
On its first frequency band (0 to several GHz?) the typical amplitude flatness across (say) a 100MHz span should be better on a decent swept analyser like the HP8566B when compared to a typical RTSA set to a 100MHz span. A healthy HP8566B should perform very well in this test but a typical RTSA will be quite poor in comparison. A lot depends on how well the RTSA has been calibrated/corrected internally. The HP8566B will excel at this test because of the swept architecture :)

You seem to be speculating ("should be"?) -- can you find any data to support this claim?

I'm unconvinced: oscilloscopes sample at higher rates, over wider bandwidths, and maintain excellent flatness.  The aberrations that you would expect to find, are common to both architectures (i.e., frequency response errors due to transmission lines, PCB traces and connectors, protection circuitry, mixers, amplifiers, ADCs), so neither one should be particularly susceptible to quirks.

Indeed, the analog sweep SA might be worse, because it has elements the oscilloscope doesn't: IF filters.  These need to be very flat, or calibrated properly.  (The swept RSA might have a bigger challenge with this, as it has wider IF bandwidth, so perhaps the filters need to be higher order for the same sharpness and flatness.  The direct sampling RSA is just an oscilloscope married to an FFT, so isn't a special case.  I... don't know if anyone actually markets "DSRSAs".)

Whatever the case, the nameplate spec is the only thing that matters.  If it's good enough, then it's good enough. :)

Tim
Seven Transistor Labs, LLC
Electronic Design, from Concept to Layout.
Need engineering assistance? Drop me a message!
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2228
  • Country: gb
Quote
You seem to be speculating ("should be"?) -- can you find any data to support this claim?
You are comparing a scope to a spectrum analyser. I think the thread is about an old school swept spectrum analyser vs a modern real time spectrum analyser. i.e. an RTSA consisting of an RF converter followed by a digital IF.

The swept analyser has a fixed RBW filter (and therefore a fixed IF with stable gain) and the passband ripple will therefore be a function of the frontend/mixer flatness. The HP8566B is typically very flat on its first range. Especially if you select a 100MHz span from within 50MHz to 2GHz.

By contrast the FFT based spectrum analyser will inevitably have ripple within its IF passband filters (these could be 10MHz or 20MHz or even 100MHz wide). The best RTSA analysers will have corrections and self calibration functions to minimise this effect but I've not seen one that can compete with an HP8566B for flatness in a typical 100MHz sweep.

Note: On the upper frequency ranges of the HP8566B the YIG preselector will spoil the passband flatness. That's why I limited the claim to the first RF range of the HP8566B :)

 



« Last Edit: June 17, 2017, 08:23:33 AM by G0HZU »
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2228
  • Country: gb
The other gotcha with some modern RTSA's that have a very wide real time BW is that they may not have any image rejection when used on the upper range of the analyser (typically >4GHz)

I think the 'wide IF' Tek RTSAs we have at work maintain a reasonably good image rejection across all ranges. However, some Agilent/Keysight RTSAs don't have any image filtering in the analyser front end if the 'wide' real time BW option is selected.

 

Offline KE5FX

  • Frequent Contributor
  • **
  • Posts: 624
  • Country: us
    • KE5FX.COM
The other gotcha with some modern RTSA's that have a very wide real time BW is that they may not have any image rejection when used on the upper range of the analyser (typically >4GHz)

I think the 'wide IF' Tek RTSAs we have at work maintain a reasonably good image rejection across all ranges. However, some Agilent/Keysight RTSAs don't have any image filtering in the analyser front end if the 'wide' real time BW option is selected.

Interesting, and surprising.  Some of their specialized instruments don't have any image rejection -- notably the E4406A transmitter tester, which shares some guts with the PSA but is explicitly intended to look at a single signal at a known frequency -- but a general-purpose SA?  YIG filters might be too slow and narrow for the latest and greatest instruments, but for the money they're charging, they can afford a switched filter bank.
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2228
  • Country: gb
Yes, I think the Agilent/Keysight PXA analyser switches out all preselection when a very wide IF BW option is selected on the upper frequency bands. The YIG is too narrow so it gets bypassed.

At work nearly all of our RTSAs in the engineering labs are made by Tek and I think they all have decent image rejection. The most popular RTSA is the Tek RSA3408A and we have loads of them but this only has a 36MHz RTBW. But we also have Tek RTSAs with 110MHz RTBW. eg the RSA6114A. I think this still has image filtering on all bands but I can't be certain. I rarely use it as it is not nice to operate. The UI is very strange and highly frustrating to use.

I have a Tek RSA3408A here at home and it definitely has decent image rejection above 3.5GHz on the upper ranges as I just checked with a few tests. But it only has 36MHz RTBW.

The flatness isn't great on this instrument if used on a 100MHz span.  On a 36MHz span it is much better if it has had a recent internal calibration but it still can't match the HP8566B here.

« Last Edit: June 17, 2017, 08:32:23 PM by G0HZU »
 

Offline w2aew

  • Super Contributor
  • ***
  • Posts: 1629
  • Country: us
  • I usTa cuDnt speL enjinere, noW I aR wuN
    • My YouTube Channel
All of the Tek RSA3000, 5000 and 6000 series units use switched bandpass filters for preselection - so they preselected at all frequencies, up to the highest RT bandwidth for these modes (165MHz for the 5000). 
======================================
YouTube channel: http://www.youtube.com/w2aew
FAE for Tektronix
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf