Siglent SSA3000X and SSA3000X-Plus Spectrum Analyzers

[dpenev]

@nctnico it is - 120dbm/hz noise power density, not - 120dbm. I think I am far above the noise floor
@Tautech all was done with 8.1 will update and redo
Meanwhile I found agilent app note for noise marker they do a few compensation. http://www.ab4oj.com/test/docs/5966-4008E.pdf
Wondering if Siglent implementation is OK

Thanks!

Sent from my MI NOTE Pro using Tapatalk

[rf-loop]

About noise level measurements.


Noise level need corrections due to distribution (Rayleight distribution) and correction for gaussian type and rectangle filter etc. Noise dB/Hz is so that there is 1Hz wide "window". Just like you se noise and then cut 1 Hz wide slice from it.  But because SSA can not do it, independent of brand. Is it R&S, Keysight or what ever, it use  filters what are typically "gaussian like" (exept that.... but not here...)

This is overall nice explanation how SSA3kX works. (when you read compatible things there)   http://cp.literature.agilent.com/litweb/pdf/5952-0292.pdf
Whole chapteer 5 is good to read from start to end.

Finally there need do some approximation and without nitpicking just find some factor what is "near enough" right.  In Siglent it looks like this correction factor constant is 2.5 or 2.51dB.


This image is just one example for pick up this factor. (previously I have made more many different tests and all tests are in line with this one.  Previously in this forum this factor have also noted.  It is also handled in other Agilent app note (150)

In image there is 2 traces. A and B.  They are overlayed and B is top.
SSA3kX FW 1.2.8.2

Marker 1  read trace A level  dBm
Marker 2 read trace B level but show normalized level for dBm/Hz   (1Hz)
RBW is 10kHz.  So subtract 40dB
Add noise 1Hz normalization factor ~2.5dB  Who knows if this is exactly right for just these filters and other things. It can be exactly right in very rare cases but still I'm quite sure it is mostly wrong (but enough close right for normal use). Nearly all is wrong and what ever you see on the screen is still lie, sum of errors, what ever you try or how ever perfect human made machine you do it show all wrong - more or less. Test and measurement world is this.  But if it is enough right we can happy use these.   
It is just "close enough to the truth" for this kind of purposes what is typical use of this kind of SSA's.
Slide rule as calculator and HP T141+plugins  as SA and ITT as DMM, they are nice, they do not fool user with last digits after decimal point...

[bozidarms]

http://vma-satellite.blogspot.pt/2017/02/siglent-ssa3021x-new-firmware-p0802.html

Hi Vitor,
now seen on your Satellite Blog, looks very promising - congratulation

Regards,
Bozidar

[rf-loop]

http://vma-satellite.blogspot.pt/2017/02/siglent-ssa3021x-new-firmware-p0802.html

Hi Vitor,
now seen on your Satellite Blog, looks very promising - congratulation

Regards,
Bozidar

Same here. 

[michael2]

After a few months without my hobby, I decided to look for a new firmware for my SSA, and found this note:

1. This firmware must be upgraded from V1.2.8.1 or later. If your SSA3000X has an earlier version, please to V1.2.8.1 first and then proceed to uplate to the latest version.

Hmm. I don't have V1.2.8.1.     And Siglent doesn't provide a link to this older firmware.   

Please, can somebody help me and send me the older firmware?   

Thank you in advance!

[tautech]

After a few months without my hobby, I decided to look for a new firmware for my SSA, and found this note:

1. This firmware must be upgraded from V1.2.8.1 or later. If your SSA3000X has an earlier version, please to V1.2.8.1 first and then proceed to uplate to the latest version.

Hmm. I don't have V1.2.8.1.     And Siglent doesn't provide a link to this older firmware.   

Please, can somebody help me and send me the older firmware?   

Thank you in advance!

All previous versions of firmware for any Siglent model or series are listed in this page:
http://www.siglentamerica.com/gjjrj.aspx?id=15

Latest versions are at the top.
Scroll down or next page until you find the one you need.

[Bicurico]

Hi,

Here is a small demo of the work in progress. Not very useful at the moment, as lots of funtionallity still missing.

Pic 1: Current state.
Pic 2: VMA Simple Spectrum Analyser for ADF4350/ADF4351/MAX2870 devices.

Regards,
Vitor

[bozidarms]

Vitor,

how fast is actually your application (specially the waterfall diagram)?

Regards,
Bozidar

[Bicurico]

Pretty fast, at least compared to the other version for the small Chinese devices.
It depends on the settings, though, especially RBW and VBW.
For reasonable settings it is very fast. If you have a SSA3021X, try it yourself.

Regards,
Vitor

[rf-loop]

I think it is useful to start separate new topic just for this VMA - SSA. 

If we continue it here it may lead endless talking about this software this and that mixed with only SSA3000X things. Who can then find anything what is about SSA3000X itself or about your nice development work..



My opinion is that it is useful to start own topic for this VMA-SSA  and not so much mess this thread. If we mix these things here and if there come many messages for this, then it is very hard for forum vistors find real things for SSA3000X itself.  Just like "mod"  is also in separate thread.

@Bicurico   this is nice work and I hope this can develop to very useful software - it is hard work.

---------------
Edit / Add:
New thread for this nice work by Bicurico can find here:

    https://www.eevblog.com/forum/testgear/vma-simple-spectrum-analyser-for-siglent-ssa3021x/

[bozidarms]

Yes, i totally agree with rf-loop -
 own topic for this VMA-SSA will be perfect,
 so Vitor, please open a new topic,
 your software definitely deserve that

Regards,
Bozidar

[dpenev]

Hi Gents,

During upgrade (from 7.07 to 8.1 and then up to 8.2) the progress bar I get goes up to 40% and then instrument reboots.
I get logo which stays for a while and the instrument  reboots again into the normal working mode.

Is this the same behavior you have during the upgrade?

Thanks!

[tautech]

Hi Gents,

During upgrade (from 7.07 to 8.1 and then up to 8.2) the progress bar I get goes up to 40% and then instrument reboots.
I get logo which stays for a while and the instrument  reboots again into the normal working mode.

Is this the same behavior you have during the upgrade?

Thanks!

Yes, to date this has happened in all firmware updates. It is normal.
My 3032X demo unit arrived with 7.01 and with every update since, SSA3kX has auto double booted while updating.

[bozidarms]

Another successfully upgraded SSA to 8.2 
Thanks to the community
Regards,
Bozidar

 

[rf-loop]

Hello,

I am after accurate noise measurements.
I have applied flat White Gaussian noise -120dBm/Hz to the analyzer.
I have done two measurements with and without pre-amplifier and I get about 0.35dB difference.
See attached picture. Probably this PA offset can be fixed?
Anyone with experience with the instrument calibration?

Thanks
Dimitar

Real "factory" calibration need so many things and so on... 

But, if really want do some own corrections and if really ( I mean really) know what to do there is one method.  But if not have enough accurate and trusted signal sources I do not recommend.

There is inside system one file. In this file there is correction table for PA. It have 806 data points over whole frequency bandwidth (3.2G)

This file is in "calib" folder.  File name is cali_pa

It need telnet connection and knowledge how to operate there.

Best is first take backup from all important data there.

After then make own edited cali_pa file. Then change it to system. And keep original also there (renaamed) so that these can later swap when need/want.

In this image is one very rough experiment.
cali_pa data table edited very roughly for get this figure.
One data point in table is in form 0x00000000
In my unit this part of table is roughly around 0x00001000   and edited table have first steps same.
After then 5 data points 0x00002000 and after then next 5 positions 0x000007FF and then next fiive 0x00001000 and then just where is image right side start next five  0x000017FF  but these are not visible..

It looks like cal daata do not affect linear and also it depends frequency and I suspect it is very different with individual preamplifiers... 

Around 100MHz and around 0x00001000 value  0.1dB is roughly (hex) 1F ~ 23

trace A: PA on, just SSA noise without signal and modified cali_pa in use.
trace B: Sweep with external sweeper 1 - 80MHz!  (this sweeper is NOT level accurate) but there can se same figure. (last ~20MHz rise around 10dB due to harmonics etc during sweep. Trace mode "max hold")

This is only for give some idea.....  (but carefully, it is very easy to do more damage there than useful... )

[dpenev]

Hello,

Another things that puzzle me.

My setup:
-120dBm/Hz flat white Gaussian noise
I have tried to measure using "Noise marker" and setting few RBWs.
Each time I keep the VBW same as RBW and average traces enough to get low variance (0.5dB)
Detector is "Average"
RBW=10KHz(100 avg traces)  Noise marker=  -120.1dBm/Hz (trace variance +/-0.5dB)
RBW=3Khz (100 avg traces)    Noise marker= -120dBm/Hz     (trace variance +/-0.5dB)
RBW=1KHz (100 avg traces)    Noise marker= -120dBm/Hz     (trace variance +/-0.5dB)
RBW=300 (3 avg traces)          Noise marker= -117.8Bbm/Hz  (trace variance +/-0.5dB)
RBW=100 (1 avg traces)          Noise marker= -118.3Bbm/Hz  (trace variance +/-0.5dB)

For the RBW 300Hz and 100Hz my measurement is about 2dB too strong. Do you think it is a bug?
Should I prefer big RBW and many trace averaging or small RBW and less  trace averaging?

I also have compared the dBm readings on the screen at the marker and what the "Noise marker" calculates
for example I see my trace at -95dBm with the RBW=300Hz  and I expect power density -95-10*log10(300)=-119.77 dBm/Hz
but the noise marker calculates it to  -117.8 dBm/Hz (2dB too strong)
It is interesting how the analyzer is calculating the "Noise marker"?

Things are getting even more confusing as I noticed that what I see on screen differs from the regular dBm marker with sometimes 1dB
See the attached picture. So I am not sure I should believe the trace and grid or the marker.

Any comments on all this is welcome.

Thanks
Dimitar
 

Hello,

I have redone the above measurements using latest firmware 8.2
The above values I get the same.

Then I have used another white noise source to check (General purpose arbitrary waveform generator)
and I got more equal noise power estimation at RBW=10KHz and 300Hz.

Then I realize that my original noise although random is actually limited record length which is repeated (34Hz repetition rate)
and so discrete spectrum. This is most probably make my noise marker 1.5dB off at low RBWs?
The noise sampling rate is much bigger then my measurement frequency range.   

To summarize, my experiment may be useful for someone.
(Or someone can explain the results)   
The difference in the noise power spectrum measurements for RBW 10KHz and 300Hz is: (for a few input noise level values)

Source A (white noise but the record is limited in time and repeated with 34Hz)
~ -100 dBm/Hz    1.5 dB
~ -120 dBm/Hz    2 dB
~ -133 dBm/Hz    1.6 dB

Source B (AWG) better results in the specified Analyzer tolerances.
~ -80 dBm/Hz      0.8dB
~ -133 dBm/Hz      0.5dB


Thanks
Dimitar

 

[dpenev]

Hello again

Doing some more experiments I am observing something I believe is an issue
I was measuring my -120dbm/Hz white noise (two noise sources have been tested ) and decide to compare two spans 10MHz and 50MHz using RBW=VBW=300Hz
Each time I was measuring the noise marker around 5MHz.
See the attached plot.  The old trace was for the 10MHz span and I have measured noise power -120dBm/Hz @5MHz
The new span is 50MHz , all other settings the same
The new partially plotted trace measures -117.47 dBm/Hz @ 5MHz so 2.5dB more.
Why the analyzer shows different power at the different span settings?

This issue I don't see for RBW=VBW=10KHz 
 

[rf-loop]

Here is two images.Both have two traces.
All have Detector=VideoAverage
All have exactly same pseudorandom noise level input.
All have same center freq, 5MHz
All have RBW=VBW=300Hz
Image where is span 100kHz and 500kHz it have both traces sweep mode "sweep" and both traces set for same sweep time around 54s (manually forced).
Image where is 1MHz and 10MHz span, it have both traces sweep mode "FFT" and 1MHz span have (automatic) 3.8s sweep time and 10MHz span sweep time is (automatic) 42.8s

I can not see any significant differences  between different spans from 100kHz to 10MHz span.
(also I have tested many times and also bit different settings. All results in same ballpark.)

Is it possible you have some  error in your test system or user error or some other problem. Wit these your exercise tests I do not believe your SSA is broken without real trusted evidences.


A span 100kHz, B span 500kHz



A span 1MHz, B span 10MHz

[dpenev]

Hi rf-loot,

Thank you for the measurements.
I have double check and clearly see the issue.
See attached.
Yellow measurement is for the span 0 - 50MHz
Pink trace is for the span 0 - 10MHz

In your case you have not checked the 50MHz range.
Can you set wide band white noise at the analyzer input (the signal level you have used I think should be OK)
Measure with the same settings you already used from 0 - 10MHz and then 0 -  50 MHz.

Thanks
Dimitar

[rf-loop]

Hi rf-loot,

Thank you for the measurements.
I have double check and clearly see the issue.
See attached.
Yellow measurement is for the span 0 - 50MHz
Pink trace is for the span 0 - 10MHz

In your case you have not checked the 50MHz range.
Can you set wide band white noise at the analyzer input (the signal level you have used I think should be OK)
Measure with the same settings you already used from 0 - 10MHz and then 0 -  50 MHz.

Thanks
Dimitar

After now I have better imagine what was your test. I have done now some fast tiny checkings. More deep tests are time consuming together with documenting and I do not have more than 24h in my clock for day. If I can find 28h/day clock from some shop it may help..

I have not now images to show my tests but now I have seen it with my SA and my suspect is: there must be some kind of bug. It also looks like Error depends start and stop frequencies (and span). Small change in span and error looks like disappear and with bit different settings it show this error.
Just two examples (starting 0Hz):   Stop freq 50MHz not ok,  49.3Mhz ok,  48.36MHz not ok... and so on.
I will do some more tests and report to Shenzhen.

[videobruce]

Many of these have already been reported I'm told by Siglent, but since I already composed a list of needed 'changes' I'll post them here for convenience. Note " 1" is skipped since it requires a separate entry due to it being what I consider fairly severe!

2.Trying to use the Normalize function with the Tracking Generator when the 'Units' are set to anything other than dBm will not work. Normalize is grayed out. One has to remember to go back and change the setting to dBm unlike other SA's that do this automatically (Rigol for starters). Note, dBmV has been common for receive measurements for Signal Level Meters (MATV/CATV etc) for as long as I can remember.

3. Tracking Generator menu: Remove the "Normalize" soft key on the 1st page since all it is is a .next page' go to button and move the four entries on that 2nd page to page one. There will still be a vacant soft key left over. This eliminates jumping back and forth between pages something they have done well to eliminate over the Rigol menu structure.

4. No mechanical power button. I see no reason why ALL test equipment shouldn't have one. Yes, they can be connected to switched power strips, but having a true removal of power w/o physically unplugging can also be a safety factor. Keep the standby button, but I'm sure there is room on the rear panel for a rocker switch.

5. The 'standby mode' is a power hog by todays standards. Five watts running 24/7 shouldn't be acceptable anymore. That's 1/6th of the operational current (28-29 watts). I assume this has wake on LAN or similar? Fine, but does it have to be that much of a hog? How about a option to disable and 'wake' mode to get the draw down below 1 watt (EPA standard)? At least that would be a somewhat quick fix, assuming it can be done thru the firmware.

6. Need a option to blank out the ridiculously long default name/number assigned to saved screen shots. Considering that the time & date is in the instrument directory and when copied to a flash drive, there isn't any real need for it to be the file name. Deleting it one digit at a time is even more annoying.

7. Those annoying front feet that collapses ever time they are extended after one picks up the instrument by the handle when it brushes the desk or table top is really annoying. Copying a good idea is one thing, but copying a bad idea is another.

8. How about a carrying case option? One that is designed to allow actually using the SA while still in the case. Mind you, I'm not suggesting climbing a tower, utility pole or rooftop, just for remote monitoring etc where AC is available.

[videobruce]

Apparently everyone uses auto settings for Sweep, RBW & VBW since no one here has apparently noticed this somewhat severe firmware problem as follows. 
This has been reported and confirmed!

To justify my reason to declare this a #1 is this; the need for speed for real time monitoring complex waveforms. Namely ATSC (8VSB) for OTA (Over The Air) broadcast TV for just one example. Many analyzers allow adjusting sweep speeds in manual mode and they stay set no matter what is done otherwise,. But once set, the setting doesn't change (and shouldn't) if it is in 'Manual' mode.

Knowing that there isn't any free ride, in this case accuracy, something usually has to suffer which is acceptable in this case since missing changes in the signal for the sake of accuracy isn't. Manual modes are provided for a purpose. They allow operator choices which they should. You don't have to use them, but they are still there. I for one prefer to take advantage of a supplied option. 

Setup conditions (easy to duplicate):

1. Place the RBW, VBW & Sweep settings all in manual mode. Set the two bandwidths to a middle setting. Settings for the attenuator or units shouldn't matter. No signal is required.

2. Change the Sweep to the fastest setting (usually 24ms)

3. Set the frequency range to whatever you want, the full bandwidth is fine though I never have a need for that. My ranges are usually between 50 & 500 Mhz.

4. Go to the BW page and start changing either of the two adjustments paying attention to the sweep speed. You can use either the 'jog' dial or the up/down buttons. You will see it drop to the next slower speed.

5. Now, here where it gets more interesting, return to the previous setting. Note the sweep speed still slowing instead of returning to where it was. Everytime you make any adjustment to either the RBW or the VBW in either direction the sweep slows down to the next fixed setting forcing one to return to the sweep page and crank to back to the fastest setting allowed. Trying to get a respectable display is a real chore jumping back and forth between BW & Sweep.

This brings up a few questions;
A. Why is it changing in the first place since all settings are in or supposed to be in manual mode??
B. Why are these preset sweep speeds fixed at these 'odd' figures? Some mathematical formula and/or a limitation of the system??

With further testing I did find a couple of exceptions where the speed reversed direction, but only to a certain extent.

And no one noticed this at all??   

I trust this was spelled out enough to get a handle on it. If it wasn't please ask where I lost anyone.

[Bicurico]

I think you came across the same issue I mentioned vaguely here: https://www.eevblog.com/forum/testgear/siglent-ssa3000x-spectrum-analyzers/msg1012414/#msg1012414

[videobruce]

when I start to play with Frequencies, Span, Amplitude, Sweep and Trace options, after some time everything gets slow. Even if I set the parameters back to reasonable values, the refresh rate is low. If I press the PRESET button, however, things speed up again. So I tried to memorize those PRESET settings and on the next slow down entered them manually: the SA was still slow. Pressing the PRESET made it run faster again.

I would say not since you didn't make any reference to RBW & VBW, nor to it the speed constantly dropping. 

[videobruce]

I have never meet this exept in situations when I have changed some parameter what affect speed. I mean example if I have adjusted VBW, after adjusting it is manually forced until I change its mode back to auto.

That referred to a one time manual change then back to auto.

Among a number of older, far more expensive SA's I either demoed or actually owned, there were many that had deal-breaking 'quirks'. The Anritsu MS2712E for example had a similar problem, about as bad and surely quirky where sweep speed was inconsistent with other settings, so was a Agilent N1996A. Sweep time would report one thing, the screen clearly showed different.