Siglent SSA3000X and SSA3000X-Plus Spectrum Analyzers

[Bob S]

- The screenshot save windows is visible in the middle of the screen on the saved screenshot. That way it covers a good amount of the spectrum.

I second that! 

[Wuerstchenhund]

I don't now about Europe offhand but here in North America the customer can certainly obtain an RMA from us and send a warranty repair directly to Siglent in Ohio. No reseller needs to be involved unless a customer wants them to be.

Our goal is to repair any defective instrument within 10 days of receiving it. We take this goal very seriously.

That's good to hear (and Siglent should really do the same in Europe), but that covers the hardware only. What about firmware problems? I doubt they'll be fixed within 10 days.

- The screenshot save windows is visible in the middle of the screen on the saved screenshot. That way it covers a good amount of the spectrum.

I second that! 

Which suggests it's pretty much the SDS2000 all over again - they clearly don't test their firmware (otherwise they would have found that bug), and relying on user feedback to find bugs, which may or may not get fixed at some indefinite point in the future.

[michael2]

   WTF? 

I own a Siglent SSA3021X for a few days. I made a firmware update to the actual 7.01 because I bought the TG option...

Here is a "screenshot"  of the tracking generator

Yellow trace is a selfmade BNC cable 1 meter, the pink one is bought "high quality" SMA cable 1 meter, too.



What a fail!

[nctnico]

I don't now about Europe offhand but here in North America the customer can certainly obtain an RMA from us and send a warranty repair directly to Siglent in Ohio. No reseller needs to be involved unless a customer wants them to be.
Our goal is to repair any defective instrument within 10 days of receiving it. We take this goal very seriously.

That's good to hear (and Siglent should really do the same in Europe), but that covers the hardware only. What about firmware problems? I doubt they'll be fixed within 10 days.

You are truly evil! I already wrote a similar comment but choose not to click 'post' 

Regarding the tracking generator: can someone verify that with a different spectrum analyser? It looks so bad that it must be a measurement error!

[TheSteve]

I assume it can be normalized?

[markone]

Regarding the tracking generator: can someone verify that with a different spectrum analyser? It looks so bad that it must be a measurement error!

It's still within specs ;-)



Also Rigol's TG is not that brilliant ...



I wonder about the performance of incoming SSA3000X improved version.

[bson]

Hmm, I think 1-2dB is fine.  But I'd like to see the output level calibrated to zero out the peaks, so it's consistently 0-1dB low instead of going high.  Just to avoid overloading external mixers and amplifiers in parts of the spectrum.

[markone]

Hmm, I think 1-2dB is fine.  But I'd like to see the output level calibrated to zero out the peaks, so it's consistently 0-1dB low instead of going high.  Just to avoid overloading external mixers and amplifiers in parts of the spectrum.

Good point, in this regard we have also to add amplitude precision error margin : 0.7dB



If not a full review at least it would be nice to see the this instrument "stressed" to main spec numbers, it would not require so much time ...

[michael2]

Hmm, I think 1-2dB is fine. 

Yes, I think so, too. Because this is including 1 meter cable and two adapter from N to SMA.

But maybe I found an other issue, I would like to call it the "40 MHZ problem": I measure relatively high and sharp peaks (RBW 10 Hz) above DNL at 40 Mhz and harmonics, no matter if preamplifier on or off. Some harmonics are really bad, they are 10db above DNL. The critical frequencies are (Mhz):

40, 80, 120, 160, 200, 240, 280, 520 (!!!), 560, 880, 960 (!!!), 1040 (!!!), 1120 (!!!), 1200, 1560 (!!!)

Frequencies marked with (!!!) are the really bad ones. Because this is my fist spectrum analyzer and I obviously have no EMC shielded room, I don't know if this tone are radiated into the analyzer or picked up internally.

 

[markone]

40, 80, 120, 160, 200, 240, 280, 520 (!!!), 560, 880, 960 (!!!), 1040 (!!!), 1120 (!!!), 1200, 1560 (!!!)

Frequencies marked with (!!!) are the really bad ones. Because this is my fist spectrum analyzer and I obviously have no EMC shielded room, I don't know if this tone are radiated into the analyzer or picked up internally.

Are you getting this with 50ohm termination on input ?

Can you show a screenshot and list all test conditions ?

[Wuerstchenhund]

I don't now about Europe offhand but here in North America the customer can certainly obtain an RMA from us and send a warranty repair directly to Siglent in Ohio. No reseller needs to be involved unless a customer wants them to be.
Our goal is to repair any defective instrument within 10 days of receiving it. We take this goal very seriously.

That's good to hear (and Siglent should really do the same in Europe), but that covers the hardware only. What about firmware problems? I doubt they'll be fixed within 10 days.

You are truly evil!

I know   

I already wrote a similar comment but choose not to click 'post' 

 

Regarding the tracking generator: can someone verify that with a different spectrum analyser? It looks so bad that it must be a measurement error!

I agree, it looks way to bad for even a mediocre TG. People often forget that cables and adapters (and if they are clean and connected properly, i.e. not loose) can cause major errors.

[Wuerstchenhund]

But maybe I found an other issue, I would like to call it the "40 MHZ problem": I measure relatively high and sharp peaks (RBW 10 Hz) above DNL at 40 Mhz and harmonics, no matter if preamplifier on or off. Some harmonics are really bad, they are 10db above DNL. The critical frequencies are (Mhz):

40, 80, 120, 160, 200, 240, 280, 520 (!!!), 560, 880, 960 (!!!), 1040 (!!!), 1120 (!!!), 1200, 1560 (!!!)

Interesting. Can you provide a screenshot of the spectrum from say 10MHz to 1.8GHz, with a somewhat narrow RBW if possible?

Frequencies marked with (!!!) are the really bad ones. Because this is my fist spectrum analyzer and I obviously have no EMC shielded room, I don't know if this tone are radiated into the analyzer or picked up internally.

You don't need an EM shielded room, just put a N short or a 50 ohms N terminator on the input.

[bson]

One obvious question : up to what SPAN width FFT mode is supported ?

This is another interesting question that's worth investigating if someone with the instrument in their possession should find the time to do so.

[michael2]

  So I made some more experiments regarding the 40 MHz issue. It looks like that attenuator, input termination and pre amplifier have no influence on this issue. The peaks are practically the same high above the average noise.  Because in the current firmware (7.01) screenshots are broken, I didn't make some.

So, please can somebody check with these settings: Center 80 MHz, span 90 Mhz, RBW and VBW 100 Hz, attenuator 50 dBm, trace A maximum hold, trace B average 8 times, detector positive peak.

After 8 sweeps I got the following results:

Average noise: -60.5 dBm
Peak 40 MHz: -46.38 dBm
Peak 80 MHz: -54.80 dBm
Peak 120 MHz: -56.94 dBm

It would be nice if someone could check because I have to decide to return this unit or not.

[markone]

So, please can somebody check with these settings: Center 80 MHz, span 90 Mhz, RBW and VBW 100 Hz, attenuator 50 dBm, trace A maximum hold, trace B average 8 times, detector positive peak.

After 8 sweeps I got the following results:

Average noise: -60.5 dBm
Peak 40 MHz: -46.38 dBm
Peak 80 MHz: -54.80 dBm
Peak 120 MHz: -56.94 dBm

-60.5 dBm average noise floor @ 100Hz RBW ?

I guess those are not sweeps made with open/terminated input, could you confirm this?

If present with undrived input those are internal fixed "birdies", probably present on all currently produced instruments.
 

[nctnico]

@michael2:
With the attenuator set to 50dB the noise is more likely to be in the ballpark of -110dBm. The question you should ask yourself is: 'Is the spectrum analyser useful to you in the state it is currently in'. If the answer to that question is NO then you must return it. Do not make the mistake in thinking Siglent will fix the bugs because they won't!

[G0HZU]

-60.5 dBm average noise floor @ 100Hz RBW ?

Seems about right to me (for the trace level on the display)  if I assume the preamp is off and there is 50dB attenuation and 100Hz RBW, ppeak det and max hold.

I would maybe have expected to see -65dBm 'typical' but -60.5dBm isn't that far off.

But I've never used this analyser so I'm only going by the datasheet specs. Also, I wouldn't have used that combination of span/BWsettings/detector to look at small signals relative to noise.

[Siglent America]

Yes, the 50 dBm of attenuation will certainly reduce the sensitivity of your measurement. You also have a built-in preamp that will increase the sensitivity and lower the noise floor because it is built in and the analyzer 'knows' when it is turned on.

Please contact Siglent directly - or through the dealer from where you purchased it - if you believe are having a problem. 

[G0HZU]

I worked it out from the datasheet spec of DANL = -137dBm/Hz with the preamp off in this frequency range.

Add 50dB for the attenuation and then add another 20dB for the 100Hz RBW. So you get -67dBm.

Then factor in the choice of positive peak detector and max hold and you can expect the displayed trace level to pump up several dB over several sweeps.

[markone]

I worked it out from the datasheet spec of DANL = -137dBm/Hz with the preamp off in this frequency range.

Add 50dB for the attenuation and then add another 20dB for the 100Hz RBW. So you get -67dBm.

OK, for some reason i was assuming preamp-on so the reading would have been way off.

I'm wondering if it's correct to report the best condition DANL data as "average" in the main feature list ...

[michael2]

Ok, I made a test series using the settings from the datasheet and contacted the distributer. So I will report in a few days what the result is.

Regarding the advertising -161 dBm/Hz typical, yeah this is annoying. All the nice writing in specifications just bothers me. To be fair, without my 40 MHz peaks the DANL is even in some ranges 2 dB below the typical specifications.

So we will see what would be happen the next days. I will report. Stay tuned.

[markone]

To be fair, without my 40 MHz peaks the DANL is even in some ranges 2 dB below the typical specifications.

Have you already checked spurius specs ?




What frequency span does FFT mode support ?

[tautech]

New Firmware has just appeared on Siglent websites:
http://www.siglentamerica.com/USA_website_2014/Firmware&Software/firmware/SSA0703.rar

[ytterligare]

Hello, I've been following the thread for weeks : I'm about to get a SA, and the ( budgetwise) options are the Rigol 815 Tg and Siglent SSA3021X with TG option....mainly for amatorial use ( I'm an OM).
Are there any news on how to tell if the TG is the improved one ? I'm quite in a rush to get the SA ASAP, but not sure the one I will get will be the updated one...

Thank you

Andrea

[Siglent America]

Hello, I've been following the thread for weeks : I'm about to get a SA, and the ( budgetwise) options are the Rigol 815 Tg and Siglent SSA3021X with TG option....mainly for amatorial use ( I'm an OM).
Are there any news on how to tell if the TG is the improved one ? I'm quite in a rush to get the SA ASAP, but not sure the one I will get will be the updated one...

Thank you

Andrea

Hi Andrea,
The new TG version is not yet shipping and I believe it will begin to ship in late April.
As a note, many customers are currently using the existing model and simply normalize the non-flatness of the TG. I believe a photo of the TG plot appears earlier in this thread (Note that the scale was set for 1 dB/division). Normalizing is accomplished with 1-2 button presses and most users perform normalization to subtract out the effects of the coax, connectors, couples, etc. in the system under test anyway. So, it is something most people do regardless when testing the frequency response or VSWR of an device being tested. The normalized curve will look ruler flat.
We will notify everyone of the new serial numbers which will signal the new TG versions being shipped.
Steve