Author Topic: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer  (Read 32875 times)

0 Members and 1 Guest are viewing this topic.

Offline PinkusTopic starter

  • Frequent Contributor
  • **
  • Posts: 768
As I often benefit from this forum I thought, I should pay something back use the opportunity to compare two devices which are probably of interest for many people.

Short comparison between Rigol DSA815-TG and Siglent SSA3021X Spectrum Analyzer

I had both devices only for a few hours together on my bench, thus there was not real detailed review possible. However it was enough time for a quick comparison

Conclusion
Basically it is a give and take between both devices. As Siglent copied the Rigol in terms of user interface, most task are done identically on both devices. So it comes down to details where different people may have different opinions.
In short I can say: the Rigol sometimes is offering a few more options/settings where the Siglent on the other hand generally shows a 15-20dB lower noise floor (though also Siglent has options, the Rigol does not have e.g. waterfall display). While Siglent or Rigol may add additional functionality to the firmware in the future, Siglents advantage of the lower noise floor cannot be canceled out by Rigol.

Edit: Some newer versions of the Rigol DSA815 seem to have a improved (lower) noise floor and I also forgot to self calibrate both units before comparing them. Thus you better expect the Siglent to have 'only' a 10-15dB lower noise floor than the Rigol. Unfortunately I cannot redo the measurement on the Rigol anymore.


Costs (in Euro incl. 19% VAT).
Rigol DSA815 without Tracking Generator: 1422€
Siglent SSA3021X without Tracking Generator: 1724€
Rigol DSA815 with Tracking Generator: 1660€
Siglent SSA3021X with Tracking Generator: 1925€

For the Siglent, the Tracking Generator can also be purchased afterwards as a license code for 201€. With the Rigol you have to decide upfront for TG or not TG as the non TG version misses the internal hardware (not only the connector).
Also other software options can be added such as EMI, adv. measurements, VSWR. They can add up another 1663 Euro to the Rigol and 1414 Euro to the Siglent.
All options are available as a trial on both units for 48/50 hours (running hours of the analyzer). Thus if you do not forget to switch the unit off after using it, they will be available for quite some time for you (if you are not a heavy user).

Siglent and Rigol both offer a PC software for helping with EMI measurements. The Siglent software is free, the Rigol software needs a license which you only will get from Rigol (for free), if you proof the purchase of the EMI license for 593 Euro (no matter if your trial time is still active).
Both provide (at least in Europe) 3 years full warranty

Data overview
Frequency
Rigol DSA815-TG: 9 khz – 1.5 Ghz
Siglent SAA3021X: 9 khz – 2.1 Ghz
DANL without pre amplifier:
Rigol: 100 khz – 1Mhz: -90dBm; 1Mhz – 1.5 Ghz: -115dBm
Siglent: 100 khz – 1Mhz: -111dBm; 1Mhz – 2.1 Ghz: -136 up to -141 dBm
DANL with pre amplifier:
Rigol: 100 khz – 1Mhz: -130dBm; 1Mhz – 1.5 Ghz: -135dBm
Siglent: 100 khz – 1Mhz: -137dBm; 1Mhz – 2.1 Ghz: -154 up to -161 dBm
DANL: just a rough (approx.) overview; see datasheets of both units for exact data
Screen:
Rigol: 8”, 800x480 Pixel
Siglent: 10”, 1024x600 Pixel
Size:
Rigol: 362 x 179 x 128mm; 4.25 kg
Siglent: 393 x 207 x 117 mm; 4.6 kg

Check the datasheets for detailed data!

...continued in the next message.
« Last Edit: September 21, 2016, 08:48:53 am by Pinkus »
 
The following users thanked this post: videobruce, ElectricGuy, tautech

Offline PinkusTopic starter

  • Frequent Contributor
  • **
  • Posts: 768
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #1 on: June 19, 2016, 03:07:47 pm »
Unboxing:
The Rigol has a better quality appearance regarding housing/knobs etc.; the Siglent looks and feels kind of cheapish (plastic quality etc.). Both devices are heavy and have a similar weight (4.3kg/4.6kg), however the Siglent housing and display is much larger. The Rigol came without a screen protection (foil), where the Siglent had one, however between protection foil and display were a lot or dust particles, so either the Siglent factory is very dusty or somebody removed it and placed it back. Both come with a CD, power cord, calibration sheet and quick guide, the Siglent contains also an USB cable, the Rigol doesn’t.

The Rigol is shipped in a double box with foam, where the Siglent came in a single box with foam. Regular shipment would be OK for both but the Rigol will definitely survive even extreme situations (such as USPS shipping).

Turn it on
I leave the “take it apart” to Dave – he is much better in this. I am more the “turn it on” guy, though without reading any manual first  ;).
Turning the Siglent/Rigol unit on takes approx. the same time (23s/21s). The fan noise of both units is of a low frequency. It is auditable but not really disturbing. The Rigol seems to be louder by a tiny tiny bit, but this might be subjective. Basically they are identical.
After a while the blown out air is pretty warm on both units, so I would not advise to lower the fan speed for noise reduction.

« Last Edit: June 19, 2016, 03:25:21 pm by Pinkus »
 
The following users thanked this post: ElectricGuy

Offline PinkusTopic starter

  • Frequent Contributor
  • **
  • Posts: 768
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #2 on: June 19, 2016, 03:11:23 pm »
The following will only scratch the surface.
What I will do here is only showing different screens with the same input signal from both devices to show functions, noise level, trace quality etc.

On the first glance and after pressing a few buttons it is obvious, that the Siglent developers took the Rigol device as a template. Many menus/screens are 90% identically layouted, though Rigol often put more effort into a nice look and feel (graphical elements) where Siglent just showed the necessaries. Though for working with the unit this is not important as the needed information is always displayed anyway.

The Siglent has 8 menu buttons (right side of the screen) plus one named “back”. The Rigol comes with 7 menu buttons plus two back and forth buttons which allows them to spread the menu on several screens where the Siglent is missing this this feature – it has to show everything on one screen.

Not counting the menu buttons (right side of the screen) and the number entry area, the Rigol comes with function 22 buttons where the Siglent comes with 23 buttons.

The button functionality is a bit different on both units, the Siglent is using single buttons for some functions where the Rigol is sometimes combining two functions to one button.

The Rigol combined functionality/settings: Bandwidth with Detector; Sweep with Trigger; Trace with Pass/Fail. Siglent has individual buttons for Detector, Limit (Pass/Fail), Display, Couple (a kind of short cut to see/set the coupled parameters of RBW, VBW, Attenuation, CF Step and Sweep time on one screen) and Mode (which today offers only: Reflection Measurement).
The Rigol on the other hand has two buttons the Siglent is missing: User Key, Print Setup. The user key allows the user to define a shortcut to an often used screen/menu. Direct printing is unknown to the Siglent, thus no such option.

Obviously the different usage of buttons/functions has advantages and disadvantages. The Rigol allows to display more functions by the back/forth buttons in their menu structure where the Siglent always has only one page of available functions but is using the main function keys in a better way (e.g. Detectors are just one click away). Though the Rigol is more flexible as it allows to add more and more functions to their firmware as they are not limited to 7 buttons per screen).

I am not sure yet what I like more.

Photos/Screenshots:
1) Button layout Siglent vs. Button layout Rigol
Example of menu enhancement:
2) Siglent always only have one menu (of course sub-menus when a blue triangle is showed)
3) Rigol may have 1 or 2 or (seldom) 3 pages of menu entries.

Advantage of Siglent: you always see whats available; at Rigol you must search or you are wondering what else is on page 2. Advantage of Rigol: more flexible with additional functionality as they can add as much functions as they want in the future.
« Last Edit: June 19, 2016, 04:34:19 pm by Pinkus »
 
The following users thanked this post: bookaboo, ElectricGuy, nugglix

Offline PinkusTopic starter

  • Frequent Contributor
  • **
  • Posts: 768
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #3 on: June 19, 2016, 03:12:58 pm »
A closer look
It is notable that the Rigol is a more matured product. The Rigol does have some more options buried in the menu screens the Siglent is missing. Though these are usually no important options and Siglent might add them to later firmware versions.

Speed: With larger spans the Rigol often scans the frequency span quicker than the Siglent, however with more narrow spans and/or? narrower bandwidth, the Siglent ofter is quicker. There was not time to setup and fill a detailed table to compare the speed.

As there was no time for a detailed look and comparison of both devices. I often only went through some menus and made screen shots. They may help you as additional information if you want to decide between both devices. I always used the same signal and used the same setting on both devices.

Oh, regarding screen shots: creating screen shots on the Rigol is pretty annoying. Not only that they take ages, after pressing “save” the screen is completely filled with the file menu where you have to enter a file name. But only after pressing “OK” the screen closes and then the current screen (which has been –slowly– updated in the background) is being saved. The problem: it might take you easily 30 seconds to enter a file name and the screen being saved is not the one of the moment when you pressed “save”.
It then takes the Rigol 10 seconds to create a BMP shot and 23 seconds for a PNG shot.
The Siglent it more convenient: it shows a small box with an automatically created file name (set up of the current date and time e.g. “SSA_160617181838” which can be changed of course) inside the still-alive screen data screen. Not before pressing ENTER the then current screen (which you can always see and which is updated in regular speed) will be saved. The Siglent then needs only approx. 1 second for saving the screen shot. However, the Siglent only allows to save BMP, there is no JPG or PNG option (yet?).


For the following screen shots I always set up both units to the same settings. Even if the Siglent would allow up to 2.1 GHz, I always used the same frequency span as the Rigol (0-1.5 Ghz or lower depending on the signal) to allow an easier comparison.

As Rigol screenshots usually are smaller due their lower resolution, I resized them to the same horizontal size as the Siglent shots.

Screen shots:
All shots without any connection to the input.
1) Siglent at start up(up to 2.1 Ghz) - note that the Siglent has 20dB attenuator on as standard (Auto)
2) Rigol at start up - Rigol has 10 dB attenuator as standard (Auto)
Siglent is more flat than Rigol.
3) Siglent with pre amplifier on
4) Rigol with pre amplifier on
Note: Siglent does not allow ref level of -10 with preamp on. For this we are showing the difference with a reference level of -20dB (thus you need to add another -10dB to the Siglent values if you want to compare them with the Rigol). The next two pictures are showing noise level with identical settings.
5) Siglent noise floor with 300 Khz, 100Khz, 30Khz
6) Rigol noise floor with 300 Khz, 100Khz, 30Khz
Edit: Check out also Thomas screenshot https://www.eevblog.com/forum/testgear/comparison-rigol-dsa815-tg-with-siglent-ssa3021x-spectrum-analyzer/msg965865/#msg965865. On his DSA815 device the noise level is lower. Still the Siglent is better, but not that much as shown on my screen shot.


« Last Edit: June 20, 2016, 08:48:40 am by Pinkus »
 
The following users thanked this post: rf-loop, videobruce, ElectricGuy, nugglix

Offline PinkusTopic starter

  • Frequent Contributor
  • **
  • Posts: 768
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #4 on: June 19, 2016, 03:14:12 pm »
Screenshots:
Radio station at 88.8 Mhz; showing peak table
1) Siglent also showing the station at 89.4 Mhz
2) at the Rigol the 89.4 station is buried in the noise.

Now 10 Mhz signal, 0dB
3+4) Siglent and Rigol with 4 Khz SPAN and 100 Hz RBW
5+6) Siglent and Rigol with 1 Khz SPAN and 100 Hz RBW
7+8) Siglent and Rigol with 1 Khz SPAN and 30 Hz RBW



« Last Edit: June 19, 2016, 05:06:57 pm by Pinkus »
 
The following users thanked this post: videobruce, ElectricGuy, nugglix

Offline PinkusTopic starter

  • Frequent Contributor
  • **
  • Posts: 768
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #5 on: June 19, 2016, 03:16:05 pm »
Screen shots
Showing some screens of the analyzers without a real comparision (no time for this). Just for information:
the 10 Mhz signal is not 0dB as shown above, this time it is >0 dB.
I did not fiddle with any option/settings on these screens (no time). I just entered them and made a screen shot, so the pictures are only partly meaninful and you should not judge anything. This is only for information.
Adjacent cannel power
1) Siglent
2) Rigol
Channel Power
3) Siglent
4) Rigol
Third order interpolation
5- Siglent
6- Rigol

7) Waterfall (not available at Rigol)
« Last Edit: June 19, 2016, 05:38:48 pm by Pinkus »
 
The following users thanked this post: videobruce, ElectricGuy, nugglix

Offline PinkusTopic starter

  • Frequent Contributor
  • **
  • Posts: 768
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #6 on: June 19, 2016, 04:59:20 pm »
Screenshots:
Standard frequency deviation of the tracking generator (without normalisation)
1) Siglent Tracking Generator
2) Rigol Tracking Generator

Difference of system response (no signal on input) without tracking generator (yellow) and with tracking generator switched on (magenta). Note that the Siglent does not allow to switch on the tracking generator with RBW of less than 30 Khz no matter what span is used! I did not try if there is a limit on the Rigol, at least 10 Khz were possible.
3) Siglent Tracking Generator on/off - very little interference
4) Rigol Tracking Generator on/off - mmhh, quite some interference up to 7 dB

5) Siglent spurious signals at 1.540 and 1.905 GHz.  :wtf: Can be seen only with RBW of 10 Khz or less. This is something Siglent need to work on.
The Rigol did not show this behaviour.

OK, thats it. I know a lot were not covered (e.g. EMI) but I was short in time as I only had both units at the same time for a few hours. I know I could have been better prepared for this, but what the heck

Hopefully it will help you to select a spectrum analyzer in case you need one. Not only one of the two covered products here (there are a lot more good companies with good devices out there) but it might help you narrow down where to look at to compare units.

Finally to answer the non-asked question: which one would I recommend?
Well, both units are quite good and each one will be fine for many users. The Rigol is really matured, the Siglent needs some software improvement. The Siglent covers a larger frequency span and you can purchase it without tracking generator and buy it later if you really need one. If I would be confident that Siglent would constantly work on the firmware and would bring out improved versions 1-2x per year as Rigol does, I would tend to the Siglent. However, Siglent did not really proof that they are a reliable partner so it is a draw..... Your decision!

« Last Edit: June 19, 2016, 06:27:34 pm by Pinkus »
 
The following users thanked this post: ElectricGuy, nugglix

Online nctnico

  • Super Contributor
  • ***
  • Posts: 26755
  • Country: nl
    • NCT Developments
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #7 on: June 19, 2016, 07:36:20 pm »
Off the bat: Siglent should really write screendumps as PNG (or JPEG)! BMP is so 1990.
I'm also interested in the advanced features like EMC measurements (quasi peak) because -for example- they can be handy to study how effective spread spectrum clocking is. Sometimes you can have various spread spectrum settings and it would be nice to have some idea what the right setting is and which isn't (yes, you can make things worse with the wrong spread spectrum settings).
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 
The following users thanked this post: videobruce

Offline pascal_sweden

  • Super Contributor
  • ***
  • Posts: 1539
  • Country: no
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #8 on: June 19, 2016, 07:43:11 pm »
Does the cheaper Rigol DSA700 series have an improved noise floor?
Or is it the same as the DSA800 series?
 

Offline PinkusTopic starter

  • Frequent Contributor
  • **
  • Posts: 768
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #9 on: June 19, 2016, 10:11:22 pm »
Does the cheaper Rigol DSA700 series have an improved noise floor?
Or is it the same as the DSA800 series?
The datasheets are showing the exact same values. So I would assume the 700 series are just software crippled DSA815. Note that the DS700 series is not available with a tracking generator.
 

Online tautech

  • Super Contributor
  • ***
  • Posts: 28139
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #10 on: June 19, 2016, 11:14:34 pm »
Finally to answer the non-asked question: which one would I recommend?
Well, both units are quite good and each one will be fine for many users. The Rigol is really matured, the Siglent needs some software improvement. The Siglent covers a larger frequency span and you can purchase it without tracking generator and buy it later if you really need one. If I would be confident that Siglent would constantly work on the firmware and would bring out improved versions 1-2x per year as Rigol does, I would tend to the Siglent. However, Siglent did not really proof that they are a reliable partner so it is a draw..... Your decision!
FYI
All products that need improvement have a 6 month FW review/release cycle and Siglent need ongoing quality feedback on any bugs so that they can be replicated at the factory.
Eg. a recent simple bug that was identified in an SDG2042X AWG was fixed and released in new FW in ~2 weeks.
Avid Rabid Hobbyist
Siglent Youtube channel: https://www.youtube.com/@SiglentVideo/videos
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 37661
  • Country: au
    • EEVblog
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #11 on: June 20, 2016, 02:50:28 am »
Here is the 10MHz phase noise comparison at 10KHz offset with a -110dBc source.
No contest, the Siglent wins hands down.

« Last Edit: June 20, 2016, 03:46:21 am by EEVblog »
 

Offline TurboTom

  • Super Contributor
  • ***
  • Posts: 1388
  • Country: de
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #12 on: June 20, 2016, 06:15:03 am »
Pinkus -

thanks a lot for the effort you put into this comparison! I'm well aware of how much hassle that must have been, especially thinking of an as complete view on the devices as possible in the limited time you had the DSA815 available.

Considering that "your" Rigol DSA had the 30 Hz resolution bandwidth available, it must have been an older version that had been "modified..."  ;) . It may have been defective, though, since when I reproduce the noise measurement at 300, 100 and 30kHz, I get figures almost 20dB better than what you found. Still worse than the new Siglent, but not as far off. From what I've seen before, this also matches better with screenshots of other's Rigol analyzers.

See attached screenshot.

Cheers,
Thomas
 

Offline rf-loop

  • Super Contributor
  • ***
  • Posts: 4064
  • Country: fi
  • Born in Finland with DLL21 in hand
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #13 on: June 20, 2016, 08:32:33 am »
Just this. And this is not DANL!  But sweep peed affect to P-PK levels (of course)
With this 50ms sweep time I suspect bit that Rigol can not measure reliable example 800MHz 0dBm peak at all.
(also with this speed Siglent can't)



EDIT:
Lot of talk about noise floor.

But this is quite complicated thing and also lot of misunderstooded. Specially when we compare different SA's noise floor there is so many traps. Also different spectrum may need different settings for optimal noise floor. In this it is not at all always so that compare need do with same settings. If I need pick up signal from noise I will use best settings what I can find and not at all same settings what may be ok with other equipments. If there is two equipment and need real compare for real works instead of playing then both units need set so that they give best results, independent of each others settings. If I want use Rigol dfor some special thing I work for find best optiimal settings where it can give best result. same settings in different spectrum is NOT comparable.

I can not tell which one have lower noise floor in real until I have finded best individual settings for both individual equipments.


First of all we need undestand what is noise and how different things affect in spectrum what we then see as "noise floor".  It is best I do not even start explain all things with my poor finglish.

One quite good but still not too complex introduction is in HP App note 150
http://cp.literature.agilent.com/litweb/pdf/5952-0292.pdf

« Last Edit: June 20, 2016, 10:26:41 am by rf-loop »
I drive a LEC (low el. consumption) BEV car. Smoke exhaust pipes - go to museum. In Finland quite all electric power is made using nuclear, wind, solar and water.

Wises must compel the mad barbarians to stop their crimes against humanity. Where have the wises gone?
 
The following users thanked this post: tautech

Offline PinkusTopic starter

  • Frequent Contributor
  • **
  • Posts: 768
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #14 on: June 20, 2016, 08:44:11 am »
Considering that "your" Rigol DSA had the 30 Hz resolution bandwidth available, it must have been an older version that had been "modified..."  ;) . It may have been defective, though, since when I reproduce the noise measurement at 300, 100 and 30kHz, I get figures almost 20dB better than what you found. Still worse than the new Siglent, but not as far off.
Thomas: mmmh, irritading. Should I mistakenly discredit the Rigol?
What I can say: after checking some internet photos and youtube videos I suppose, that the noise level seems to become improved a bit (but not that much) with newer versions of the DSA815. The one I had was exactly 3 years old. And yes, it is a 30Hz version ;)

I have to admit what I forgot to do, is to self calibrate both units after warming up. While the Siglent is new, the Rigol might have sat on the bench for a year or two without self-calibration (though it shows a quick calibrate now and then for a few seconds). I simply did not think about this and as I was really hurrying, I had no time for thinking.... Also I did not terminate the inputs (when open = no signal), though with an internal impedance of 50 Ohm this should not be any problem.
Dave is doing a video on both units, it should be online in a day or so. Possibly he is doing a similar comparison which you should check out.

However, the noise level sometimes seems to be reasonable on some shots (e.g. radio station) where the Rigol shows -85dB at a RBW of 3 Khz (with 10dB attenuator), so it is working correctly.

The huge difference between your shot and mine? I have no idea right now where that came from. Unfortunately I cannot double check anymore or redo the measurements. I will try to find older screen shots of this unit.

Maybe some Rigol owners are able to redo some of the pictures with the same settings in case .
For the 10 Mhz signal, just use the 10 Mhz Ref-Out output of the Rigol on the backside. Though to get a exact 0dB level you will need an external attenuator (probably unit dependend, I used 1.5dB or 2dB).
« Last Edit: June 20, 2016, 08:45:44 am by Pinkus »
 

Offline rf-loop

  • Super Contributor
  • ***
  • Posts: 4064
  • Country: fi
  • Born in Finland with DLL21 in hand
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #15 on: June 20, 2016, 08:53:08 am »
Pinkus -

thanks a lot for the effort you put into this comparison! I'm well aware of how much hassle that must have been, especially thinking of an as complete view on the devices as possible in the limited time you had the DSA815 available.

Considering that "your" Rigol DSA had the 30 Hz resolution bandwidth available, it must have been an older version that had been "modified..."  ;) . It may have been defective, though, since when I reproduce the noise measurement at 300, 100 and 30kHz, I get figures almost 20dB better than what you found. Still worse than the new Siglent, but not as far off. From what I've seen before, this also matches better with screenshots of other's Rigol analyzers.

See attached screenshot.

Cheers,
Thomas

Is it possible to see this exactly same image so that you add also example some -10dBm signal so that it is visible with tracking peak marker  in this image same time as this noise floor. If I do it with Siglent I need reduce sweep time lot of for reliable peak level.

Signal 800MHz -10dBm (cable compensated) Here manually set sweep speed up from fastest to first speed  where is give correct level. (this is faster than default)
Other image selected accuracy priority for automatic sweep speed (default set  is speed and also this speed is more slow what I use in this other image)



« Last Edit: June 20, 2016, 09:29:22 am by rf-loop »
I drive a LEC (low el. consumption) BEV car. Smoke exhaust pipes - go to museum. In Finland quite all electric power is made using nuclear, wind, solar and water.

Wises must compel the mad barbarians to stop their crimes against humanity. Where have the wises gone?
 

Offline cio74

  • Regular Contributor
  • *
  • Posts: 173
  • Country: gb
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #16 on: June 20, 2016, 01:10:37 pm »
Just spotted this

 

Offline rf-loop

  • Super Contributor
  • ***
  • Posts: 4064
  • Country: fi
  • Born in Finland with DLL21 in hand
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #17 on: June 20, 2016, 01:22:49 pm »
Here is the 10MHz phase noise comparison at 10KHz offset with a -110dBc source.
No contest, the Siglent wins hands down.



Hello Dave:

I'm confused by the Display you provided for the Siglent SA.  The results don't look like a typical Phase Noise Measurement to me.
The Phase Noise marker label says that it it is -99.31 dBm /Hz (first, shouldn't this be 'dB/Hz, or dBc/Hz', not dBm/Hz), although it looks like it would be 80.5 dBc/Hz to me if the peak is at 0 dB as shown.
Then the Phase Noise at 1 kHz appears to be at ~ -80dB (? ? ?)
The Phase Noise display for the Rigol looks typical for the slope of the noise, where as the Siglent doesn't look right to me.  What is this?

Perhaps you have a 10 MHz source with phase noise much better than anything I have, although if so, how could the Siglent possibly  have a 100 Hz BW Filter with such a super Shape Factor (~ 1:1) with this high of a ultimate attenuation in their low cost Spectrum Analyzer.

How can this be explained?  Are you using a Average Detector here.  The display does say you have a 20 Average, so it would seem so.  The Rigol says 20 Average also, but it does not appear to necessarily be a Average Detector Display.  In any case you do not want to use a Average Detector for a Phase Noise Measurement.

Please let me know what explains the Siglent's PN unusual display.

Huh, confusion, there is so many things mixed and so on... also in video...

But answer first to one question about Siglent 100Hz RBW filter. Shape factor -3dBc/-60dBc is around 1:4.4
So not any mystery there.

As can see there in attached image there is not just exactly perfect Marker settings but, enough for just this purpose to show that there is nothing special.

There is nothing unusual PN display. Just normal image what somehow imply PN. But phase noise is not usual random distributed upside down Y figure. It is very small in close carrier but after then it rise until it normally fall. Optimised for near LO. (you can find many of these test images, now also from Dave, and previously from Pinkus and also me and possible some others also somewhere.)
This same kind of pn figure can also find in many modern synthesized signal generators. And example bit like same in some Agilent SA where user can even select what PhaseNoise optimization user want use. If use <140kHz optimization it give low pn near carrier but it may even rise when go away from carrier. Like this last image  from Agilent AN-150 where yellow trace is this kind of near carrier optimized pn.

« Last Edit: June 20, 2016, 01:31:41 pm by rf-loop »
I drive a LEC (low el. consumption) BEV car. Smoke exhaust pipes - go to museum. In Finland quite all electric power is made using nuclear, wind, solar and water.

Wises must compel the mad barbarians to stop their crimes against humanity. Where have the wises gone?
 

Offline D3f1ant

  • Frequent Contributor
  • **
  • Posts: 346
  • Country: nz
  • Doing as little as possible, but no less.
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #18 on: June 20, 2016, 07:17:42 pm »
Would love to see a comparison between these units and something like Keysights 'basic' N9320.
« Last Edit: June 20, 2016, 08:27:55 pm by D3f1ant »
 

Offline rf-loop

  • Super Contributor
  • ***
  • Posts: 4064
  • Country: fi
  • Born in Finland with DLL21 in hand
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #19 on: June 20, 2016, 09:18:39 pm »

I'm confused by the Display you provided for the Siglent SA. The results don't look like a typical Phase Noise Measurement to me.
The Phase Noise marker label says that it it is -99.31 dBm /Hz (first, shouldn't this be 'dB/Hz, or dBc/Hz', not dBm/Hz), although it looks like it would be 80.5 dBc/Hz to me if the peak is at 0 dB as shown.
Then the Phase Noise at 1 kHz appears to be at ~ -80dB (? ? ?)
The Phase Noise display for the Rigol looks typical for the slope of the noise, where as the Siglent doesn't look right to me.  What is this?

Please let me know what explains the Siglent's PN unusual display.

Phase noise slope can be many kind of. Not only like upside down Y slope.

Please look example this:
http://www.analog.com/media/en/training-seminars/tutorials/MT-008.pdf

Also, Siglent pn looks just as designed and  test results are not in conflict with Siglent specifications.
Specifications do not tell pn between 0 - 10kHz from carrier. There is just Siglent present to users. Low near carrier pn.

But then, if look image from Dave video.

"although it looks like it would be 80.5 dBc/Hz to me if the peak is at 0 dB as shown"
No, not at all. Do you know what is dBc/Hz? 

There read RBW 100Hz !  If you simply "normalize" this to Hz you subtract 20dB. 
(but this is too simple due to many unknown factors. We do not exatly know RBW filter noise BW. We do not know exactly VideoAverage Detector characteristics. But if we make 2.5dB correction we are perhas quite close. etc etc and bla bla blaa.. some weak evidence that Siglent used  2.5dB is here in attached image.  (manufacturer know details and can select good correction values)

In Dave video image trace A is VideoAverage detector output. (because he have turned noise cursor on and it is derived from VideoAverage. (using correction factor. This process is explained well in Agilent Appnote 150 http://cp.literature.agilent.com/litweb/pdf/5952-0292.pdf )

In my attached image is
trace A  Positive Peak Detector and after then done trace averaging.
trace C Normal detector. Also for this also trace averaging.
trace D VideoAverage detector and also trace averaging.  (this detector is used for Noise Marker function and  for this trace what is source for Noise Marker what value is then corrected)
trace B (trace display hidden) is for "fooling" system so that I can get dBm/Hz noise marker Marker measurement  from trace A but so that trace A keep its Positive Peak Detector as source. (this IS wrong but this give more data to me how noise marker is corrected for  noise level dBm/Hz)



I drive a LEC (low el. consumption) BEV car. Smoke exhaust pipes - go to museum. In Finland quite all electric power is made using nuclear, wind, solar and water.

Wises must compel the mad barbarians to stop their crimes against humanity. Where have the wises gone?
 

Offline HighVoltage

  • Super Contributor
  • ***
  • Posts: 5453
  • Country: de
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #20 on: June 20, 2016, 09:32:44 pm »
Very nice comparison, thank you for the effort.
There are 3 kinds of people in this world, those who can count and those who can not.
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 37661
  • Country: au
    • EEVblog
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #21 on: June 20, 2016, 09:59:41 pm »
"although it looks like it would be 80.5 dBc/Hz to me if the peak is at 0 dB as shown"
No, not at all. Do you know what is dBc/Hz? 
There read RBW 100Hz !  If you simply "normalize" this to Hz you subtract 20dB. 
(but this is too simple due to many unknown factors. We do not exatly know RBW filter noise BW. We do not know exactly VideoAverage Detector characteristics.

Correct. ideally you need to know that extra info to make "proper" phase noise measurement using this technique. But for a simple measurement like this signal gen, it's usually good enough.

Quote
In Dave video image trace A is VideoAverage detector output. (because he have turned noise cursor on and it is derived from VideoAverage.

Correct. The analyser automatically changes to video average detector when the Noise market is chosen.
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #22 on: June 21, 2016, 12:35:10 am »
Quote
The Siglent SA is apparently is awesome making close in (carrier +/-1kHz) PN measurements, as well as Dave's RF source compared to those I have at home.  We do have signal generators at work that provide a clean PN source like this, but they are to rich for me at home.  And I didn't think Dave had a SG in this class either, but apparently he does.
-100dBc/Hz phase noise at 1kHz offset on a 10MHz carrier isn't exactly special. You could make a simple test oscillator 20-25dB better than this at home with a few lumped parts and a cheap MMIC amplifier.

eg you could hope to achieve

-125dBc/Hz at 1kHz offset
-150dBc/Hz at 10kHz offset
-170dBc/Hz at 100kHz offset

Or you could simply use a 10MHz crystal oscillator to get decent close in phase noise.

There are plenty of old/used/cheap sig gens that can deliver good phase noise down at 10MHz. eg the Marconi 2019 is one example although it won't be as good as the figures above. There are some HP sig gens that should be avoided if you want decent phase noise down in the HF bands eg the HP8647/8 is particularly poor down on the HF bands.

The ESG range from HP/Agilent (eg the range that includes the ESGD 4433B or the ESG 4421) are quite poor too but not as bad as the little HP8647/8.

The classic sig gen for stuff like this is the old HP8640B as it delivers very low phase noise down on the HF bands. It can easily beat the simple homebrew osc at 1kHz offset and probably beat or match the homebrew oscillator at 10kHz offset but it can't match the homebrew oscillator at 100kHz offset. Sometimes the simple circuits are the best performers :)





« Last Edit: June 21, 2016, 01:31:54 am by G0HZU »
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 3012
  • Country: gb
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #23 on: June 21, 2016, 12:54:36 am »
I've not had access to either the Rigol or the Siglent analyser but it looks to me like the Siglent analyser has a very wide loop BW on the PLL which means it gets plenty of noise cleanup close to carrier. So it looks nice on a 30kHz span.

But I suspect there will be a price for this in terms of far out phase noise. If the Siglent has a loop BW of 50-100kHz or so then you can expect the phase noise at 100kHz to be quite high. Probably in the same ballpark as the phase noise at 10kHz offset. The phase noise at 1MHz offset will probably be quite high as well. Maybe only 20dB better than the phase noise at 10kHz offset.

So that makes it a VERY noisy analyser on wide spans of a few MHz when compared to an old school lab analyser. My old 1800MHz Advantest analyser (introduced in 1981) has typical phase noise performance of -128dBc/Hz at a 100kHz offset. Nearly 30dB cleaner than the Siglent?

« Last Edit: June 21, 2016, 01:05:32 am by G0HZU »
 

Offline pascal_sweden

  • Super Contributor
  • ***
  • Posts: 1539
  • Country: no
Re: Comparison Rigol DSA815-TG with Siglent SSA3021x spectrum analyzer
« Reply #24 on: June 21, 2016, 12:15:13 pm »
What about the noise floor level on the more high-end Rigol spectrum analyzers, such as DSA832-TG, DSA875-TG, or DSA1030A-TG3?

Are these noise floor levels all the same as in the base model Rigol spectrum analyzer, DSA815-TG?
 
The following users thanked this post: videobruce


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf