New SSA3021X exhibiting weird behavior

[ironcurtain]

I got this unit from the well known Siglent distributor based out of Germany. As far as I remember, it did not come with the screen protector, which I realized was there in Dave's unit after watching his video. I don't want to inflate the existent thread for the SA with my own issues related to my unit, so I'm posting this separately hoping it might help others with similar problems.

I don't want to doubt the seller, so for now I have not made any assumptions.

The noise floor seems much higher in my unit than in Dave's, (see attachment). This is with a 50ohm termination attached and without.

Note: I followed the standard hack to open up the options, but the calibration data was kept properly. That should not be the issue here, as the file was never touched/tampered with.

Needless to say I have barely played with it, and it hasn't seen any sort of abuse or dumb mistakes, so no damage to the frontend possible on my side. I have a calibrated HP 8924c that I could use to test the RX. I will upload pictures of that later.

Anyone knows what might be happening?

[tautech]

Here's a typical screenshot from the SSA3000X thread



Yours looks very similar.

You don't need a 50 ohm termination unless you need more attenuation than the 51 dB available internally.
Spend some time to become familiar with the UI and correct usage and there's many screenshots in the SSA3kX thread that show what it's capable of.

[hendorog]

Noise floor directly relates to the RBW setting, and your screenshot shows RBW at 3MHz - which is very wide.

Reducing the RBW by a factor of 10 will reduce the noise floor by 10dB.

Tautech's pic shows RBW at 1MHz - which is narrower and so slightly lower noise floor.

If you reduce the span, the RBW will automatically reduce too, if you leave RBW set to auto. You will see the noise floor magically drop.

Have a play around with the RBW setting on a narrow span - say 10 kHz - and you should get the picture.

[ironcurtain]

I did more tests with my calibrated HP 8924c, attached the images. 31MHz FM modulated signal, spot on frequency, -130 to -100 dBm, with different RBW settings, mostly set to 10khz.

[ironcurtain]

Noise floor directly relates to the RBW setting, and your screenshot shows RBW at 3MHz - which is very wide.

Reducing the RBW by a factor of 10 will reduce the noise floor by 10dB.

Tautech's pic shows RBW at 1MHz - which is narrower and so slightly lower noise floor.

If you reduce the span, the RBW will automatically reduce too, if you leave RBW set to auto. You will see the noise floor magically drop.

Have a play around with the RBW setting on a narrow span - say 10 kHz - and you should get the picture.

I was just about to post the screens with RBW properly adjusted to different settings 


@tautech, sure, UI familiarity is key, but a lone screenshot of someone else's unit without specific, precise context, is worthless. They could have a dud from the same batch as mine for all we know. This is not so much a UI issue but confirming with other non biased forum members that my unit is within spec.

The documentation from Siglent is also terrible, then again I come from HP & co world, where that excels.

How are you guys saving calibrated presets (ex. how do you test cable setups to save and preload attenuation/offsets?, say, a 40db 100w power attenuator and N type microwave flex cables, is there an automated procedure to do this like other brands have?)

[hendorog]

Narrow up the span as you can't tell much about frequency accuracy with a span that wide.

For example to set the centre freq to 31MHz, and then use a span as wide as you need - say 10MHz to start with, then drop it further.

Other benefit is that the sweep time will reduce, and so it will become more practical to drop the RBW further :- which will reduce the noise further and allow you to see the signal again after it has fallen into the noise floor.

[tautech]

Noise floor directly relates to the RBW setting, and your screenshot shows RBW at 3MHz - which is very wide.

Reducing the RBW by a factor of 10 will reduce the noise floor by 10dB.

Tautech's pic shows RBW at 1MHz - which is narrower and so slightly lower noise floor.

If you reduce the span, the RBW will automatically reduce too, if you leave RBW set to auto. You will see the noise floor magically drop.

Have a play around with the RBW setting on a narrow span - say 10 kHz - and you should get the picture.

I was just about to post the screens with RBW properly adjusted to different settings 


@tautech, sure, UI familiarity is key, but a lone screenshot of someone else's unit without specific, precise context, is worthless. They could have a dud from the same batch as mine for all we know. This is not so much a UI issue but confirming with other non biased forum members that my unit is within spec.

The documentation from Siglent is also terrible, then again I come from HP & co world, where that excels.

How are you guys saving calibrated presets (ex. how do you test cable setups to save and preload attenuation/offsets?, say, a 40db 100w power attenuator and N type microwave flex cables, is there an automated procedure to do this like other brands have?)

When hendorog had my 3032X for a bit his comments about the usability surprised me some as he said it was very similar to HP's he has. Much of how it works is outlined on an often quoted HP/Agilent/KS app note AN150.

I've not bothered saving specific test setups other than Power ON = Last.
Your cables and adapter's influence on a DUT can be very quickly nulled with Normalize....it takes just seconds.

Just use it and get to know it. 

[tautech]

FYI
Here's a Default Preset screenshot from my 3032X

[hendorog]

Looking at the SA mode on the SVA here, there are a couple of other options aside from Normalise, for correcting for cables/adapters etc:

* Ref offset function. Just offsets the reference to make it think that 0dBm is the value that you enter. So if you enter 10dBm reference offset, then the SA thinks a 0dBm input is 10dBm. This would correct for an external 10dB attenuator for example.
* Correction function. You can create, save and load up to 4 selectable corrections based on a number of frequency/amplitude points. You could use this to model a particular cable over a wide frequency range. Need to enter the points manually though - or upload a file I guess.
* Trace math. You can use this to subtract the effect of a cable if you can make a measurement both with and without it.

I am not sure if the correction function is present in the SSA or not. It is under the Amplitude button on the SVA.

[tautech]

I am not sure if the correction function is present in the SSA or not. It is under the Amplitude button on the SVA.

It is and in the same place as SVA.

[luc-on7kz]

hello,
can some one help me.
the last time i have making a lot of tests with my ssa3021 hacked equipment.
I have see a strange behavior, are the persons who can simulate it so i will know if it is a hardware problem or a bug into the software
span: 50Khz
BW: auto =300Hrz
detect: All tested = same
trace: All tested = same
Preamp on or out = same
can some one do the same test to confirm if it is a hard or software fault or the man between the chair and the test equipment 

[rf-loop]

hello,
can some one help me.
the last time i have making a lot of tests with my ssa3021 hacked equipment.
I have see a strange behavior, are the persons who can simulate it so i will know if it is a hardware problem or a bug into the software
span: 50Khz
BW: auto =300Hrz
detect: All tested = same
trace: All tested = same
Preamp on or out = same
can some one do the same test to confirm if it is a hard or software fault or the man between the chair and the test equipment 

If you search forum this is also previously been on the board.

Using some other settings and signal you can find more "noise steps" just like in Keysight image 3.3. Where is "noise floor after autoranging".

With your settings it is possible also turn it from FFT to normal (slow) sweep and this phenomenon disappear

I have seen this phenomenon long time ago (when SSA was just published and I get one quite early) and  it is Perhaps something like this until proven otherwise:

Keysight (Spectrum Analysis Basics – Application Note 150,  page 38) :
Figure 3-3 illustrates the sweeping behavior of the X-Series analyzers. The single-pole prefilter allows the gain to be turned up high when the analyzer is tuned far from the carrier. As the carrier gets closer, the gain falls and the ADC quantization noise rises. The noise level will depend on the signal level frequency separation from the carrier, so it looks like a step-shaped phase noise. However, phase noise is different from this autoranging noise. Phase noise cannot be avoided in a spectrum analyzer. However, reducing the prefilter width can reduce autoranging noise at most frequency offsets from the carrier. Since the prefilter width is approximately 2.5 times the RBW, reducing the RBW reduces the autoranging noise.



Figure 3-3

[RFDUK]

Hello.
Well spotted!
My SSA 3032X manufactured January 2018 displays an identical screen to yours.
Best regards,
Martyn g8fek.

[luc-on7kz]

Thanks for the replay and the explication.
Greetings Luc