Siglent SSA 3021X plus high phase noise

[angelcruiz]

Hello,

I've got an SSA 3031X which, according to its datasheet and confirmed by several tests that I've found in this blog and somewhere else, should have a phase noise of around 98 dBc/Hz @ 10 KHz offset (specified with 1GHz input signal).

The problem is I can't get better than 85 dBc/Hz as seen in the attached picture.

The tests conditions are:

Unit: SSA 3021X Plus
SW ID: 3.2.2.5.1R1
Source: Siglent SDG 2042X
Input signal: 0 dBm, 10MHz
Span: 50 KHz
RBW: 100 Hz
VBW: 1Hz
15 Averages

Despite the datasheet test conditions being with a 1GHz input signal, several posts in this blog confirm the stated phase noise using input signals of 10 or 20 MHz.

I don't know whether the fact that my unit has been hacked has somehow messed up the internal PLL parameters that shape its phase noise, I have a defective unit or there's something else I'm doing wrong. Still, I've tested with multiple configurations (RBW and VBW of 1Hz, 20, 30 and 100 MHz input signals, etc) and I always measure no less than 85 dBc/Hz phase noise.

Do you have any clue?
Thanks in advance,

Angel.

[Orange]

I just checked on my SSA3032X (non plus)

I get about 100dBc/Hz at 10MHz  @10kHz offset, BW 100Hz
Signal source is a SDG1062 at 10MHz, 0dBm

Mind you that this is not a plus model.

Seems that you have an issue with the SA or maybe the generator is at fault

[Orange]

On more thing if you use the SDG2042X with external 10MHz clock it might also increase the noise, the PLL in the generator has a known problem.

[tautech]

Try installing the latest FW:
https://int.siglent.com/upload_file/zip/firmware/Spectrum_analyzer/SSA3000X%20Plus_V3.2.2.6.0R7_EN.zip

[angelcruiz]

Sorry, to be specific, the installed FW was SVA1000X_V3.2.2.5.1R1.

I've upgraded to SVA1000X_V3.2.2.6.0R7 and the results are still the same.

I've attached a noise floor test with RBW of 30KHz (50 Ohm termination input) since it seems to be also high for about 5 to 10 dB.

Could it be that the SVA calibrations are bad for an SSA or something like that?

[Orange]

From an old SSA3032X, which is properly calibrated within specs.....
Looks almost identical. Compared to the SVA the noise floor of the SSA is about 3.5 dB lower

[tautech]

1. OP analyzer settings duplicated best I can see. Source = SDG6022X

2. With BW settings equal.

Try a Preset to factory settings.

Edit to add
Factory fresh SVA1032X

[angelcruiz]

I think the noise floor is indeed within specs, just a little bit down.

The phase noise is definitely bad, but it gets better going from 10 MHz (-85 dBc/Hz) to 1 GHz (-96 dBc/Hz, attached, ERASynth micro source). The datasheet tests conditions for PN are at 1 GHz but I'd expect it to be as good at lower frequencies.

There's no change after a factory reset or using an external 10 MHz reference either. I think it might be related with the PLL loop filter parameters. Do you know if that could be the case?

[tautech]

There's no change after a factory reset or using an external 10 MHz reference either. I think it might be related with the PLL loop filter parameters. Do you know if that could be the case?

No idea, sorry. Not seen much in the way of discussion about this.

Do consider factoring in the Error and Accuracy spec on P11.
https://int.siglent.com/u_file/download/22_08_23/SSA3000X%20Plus_DataSheet_DS0703P_E04C.pdf

Never looked as to how it compares with a SVA so there's a little exercise for you.
https://int.siglent.com/u_file/download/22_08_23/SVA1000X_DataSheet_DS0701X_E05C.pdf

Then there's the fact you hacked it and saved ~ $ 1800 however you did get a single port VNA for your troubles.

[807]

Hello,

I've got an SSA 3031X which, according to its datasheet and confirmed by several tests that I've found in this blog and somewhere else, should have a phase noise of around 98 dBc/Hz @ 10 KHz offset (specified with 1GHz input signal).

The problem is I can't get better than 85 dBc/Hz as seen in the attached picture.

The tests conditions are:

Unit: SSA 3021X Plus
SW ID: 3.2.2.5.1R1
Source: Siglent SDG 2042X
Input signal: 0 dBm, 10MHz
Span: 50 KHz
RBW: 100 Hz
VBW: 1Hz
15 Averages

Despite the datasheet test conditions being with a 1GHz input signal, several posts in this blog confirm the stated phase noise using input signals of 10 or 20 MHz.

I don't know whether the fact that my unit has been hacked has somehow messed up the internal PLL parameters that shape its phase noise, I have a defective unit or there's something else I'm doing wrong. Still, I've tested with multiple configurations (RBW and VBW of 1Hz, 20, 30 and 100 MHz input signals, etc) and I always measure no less than 85 dBc/Hz phase noise.

Do you have any clue?
Thanks in advance,

Angel.

My display at 10MHz looks almost idendical to yours. But I'm using a Marconi 2022E. Don't know what it's phase noise is like.

It's better at 1GHz.

[Orange]

Some more pictures from an SSA3032X

Looks better at 10MHz

[drew23]

Hi All.

I have a PN2060A phase noise analyser and have made a blog post on it, together with phase noise results for the 10 MHz TCXO output of my Siglent SSA3021X and Rigol RSA3015N.

The Siglent is okay at -141 dBc/Hz @ 10 KHz and I think in spec but the Rigol is considerably better at -167 dBc/Hz @ 10 KHz.

In a later, draft post, the phase noise for a Keysight MXA N9020A Vector signal analyser is -163 dBc/Hz @ 10 KHz. This was a very expensive instrument in its day and depended heavily on its TCXO for performance. So the Rigol value is pretty good for a much cheaper instrument.

I am not sure of the method used in the specifications of the Siglent. Presumably, it is the output of the tracking generator at 1 GHz. The PN2060A phase noise analyser can only measure to about 130 MHz at baseband, without a downconverter. I have just moved house and my gear is disorganised. However, I might try measurements at a higher frequency.

Regards Drew VK4ZXI

https://vk4zxi.blogspot.com/2023/07/an-economical-way-to-measure-phase.html

https://vk4zxi.blogspot.com/2023/08/have-pn2060a-phase-noise-analyser-will.html

[RoV]

I have both an SSA3021X+ "upgraded" to SVA1032X and an SDG2042X "upgraded" to SDG2122X, so I tried to replicate your configuration.
Additionally, I tested a KVG OCXO, also 10 MHz.
My OCXO is fitted with an output attenuator and gives ~-11.7 dBm on the fundamental, so initially I tried both generators at -11.7 dBm. Same settings as OP, except for input attenuator at 10 dB.
Results were ~identical for the two generators and phase noise at 10 kHz ~99 dBc/Hz
(OCXO)

(yellow OCXO, pink SDG)
By increasing the output level of the SDG at 0 dBm and with SSA input attenuator at 20 dB, results remain consistent (OCXO trace is same as before, taken with 10 dB attenuation)
(yellow OCXO, pink SDG)

[807]

Just measured mine again at 10MHz, but this time using a Marconi 2024 generator which has much better phase noise than the 2022E I used in earlier post.

The close-in noise is better on this generator, but I get similar reading to my previous one at just -86.66dBm/Hz @10kHz.

Is there a problem with mine & angelcruiz's analyzers? Why are we getting such pronounced "hills" compared to others?

edit...I tried the measurements at -10dBm, but results were the same.

btw...mine has been "upgraded", but not to a VNA

[MRF237]

Hi all, this is just an attempt to 'close' this topic.

I had my unit for 8 months or so before I ran into this issue. It was a bit indirect. I used the digital modulation analysis and my EVM readout on a QPSK signal was a lot higher than two other SSA3000x+ units (same signal). This appeared to be due to the higher phase noise on my unit. My unit also showed a DANL which was _just_ according to spec, where the other units were on the 'typical' values.
Using Google I arrived here and learned that my unit was not unique.

After some more reading and testing, I contacted Siglent, and they had a look at my unit, but they could not 'fix' it. They stated that the machine was fine and according to specs, as Siglent only specifies the phase noise at 1GHz.
I have also learned that it has nothing to do with the software “improvements” that many have done...

So, with my current knowledge: if you have a device showing this high phase noise at lower frequencies, you will have to live with it....

If not, the SSA3032X-R is an alternative. It has a different RF architecture, not showing this particular effect. I could exchange my SSA3021X+ for one of these and it is really a nice device! It has other issues, of course, as you might expect...

(Fun fact: later, I found some older posts on this blog from a user who ran into the same issue, did the same research and exchanged his unit for an SSA3032X-R as well, with the same arguments .)

[807]

It's not software related, so I wonder what has changed in the hardware? Was it an actual circuit change or perhaps a different batch/supplier of components? Perhaps the different performing units have certain range of serial numbers?

Although I love this analyzer, it still irks me to know that I have one of these units that has a 12dB higher phase noise at 10MHz. I know it meets the spec at 1GHz, but when you see what others are getting at 10MHz, it makes me wonder what the reason is. From what's been posted on here, it seems to be one measurement or the other, with nothing in between. So it doesn't appear to be a spread of tolerances, which would give a spread of measurements.

[MRF237]

it seems to be one measurement or the other, with nothing in between

Yes, indeed, good point. I compared mine with two other units (a SSA3021X+ and a SSA3032X+), both with shoulders at 10 MHz as on the 'good' screenshots on this page.
My 'bad' unit had HW version 00.00.00 where the 'good' ones both had 01. My calibration date was 2022-12-15, quite old as I bought my unit in March 2024!

[ozkarah]

I replicated the test condition with my SDG2042X and my SSA3021X Plus converted to SVA1032X.
For reference I would like to share my findings which are in par with the measurements by @tautech.

Unit: SSA 3021X Plus
SW ID: 3.2.2.6.0R7
Source: Siglent SDG 2042X
Input signal: 0 dBm, 10MHz
Span: 50 KHz
15 Averages  (8 on RBW:1Hz-VBW:1Hz)


RBW:100 Hz  VBW:1 Hz


RBW:30 Hz  VBW:30 Hz


RBW:30 Hz  VBW:1 Hz


RBW:1 Hz  VBW:1 Hz

[807]

it seems to be one measurement or the other, with nothing in between

Yes, indeed, good point. I compared mine with two other units (a SSA3021X+ and a SSA3032X+), both with shoulders at 10 MHz as on the 'good' screenshots on this page.
My 'bad' unit had HW version 00.00.00 where the 'good' ones both had 01. My calibration date was 2022-12-15, quite old as I bought my unit in March 2024!

My "bad" unit is also HW 00.00.00 I bought mine from Telonic in UK in April 2023. My calibration date was very close to yours on 2022-12-09

[807]

I replicated the test condition with my SDG2042X and my SSA3021X Plus converted to SVA1032X.
For reference I would like to share my findings which are in par with the measurements by @tautech.

Unit: SSA 3021X Plus
SW ID: 3.2.2.6.0R7
Source: Siglent SDG 2042X
Input signal: 0 dBm, 10MHz
Span: 50 KHz
15 Averages  (8 on RBW:1Hz-VBW:1Hz)

(Attachment Link)
RBW:100 Hz  VBW:1 Hz

(Attachment Link)
RBW:30 Hz  VBW:30 Hz

(Attachment Link)
RBW:30 Hz  VBW:1 Hz

(Attachment Link)
RBW:1 Hz  VBW:1 Hz

What is the HW version of your unit?

[rf-loop]

Kind recommendation:

For who ever who do these tests.

If you use ExtRef, please do also perfectly same test using SSA own internal reference. If no difference detected (except small freq. difference) then it do not matter what result you display (related to phase noise but perhaps good to inform that test result using internal reference  is same)  but if there is difference please inform both results.

[807]

Kind recommendation:

For who ever who do these tests.

If you use ExtRef, please do also perfectly same test using SSA own internal reference. If no difference detected (except small freq. difference) then it do not matter what result you display (related to phase noise but perhaps good to inform that test result using internal reference  is same)  but if there is difference please inform both results.

Yep. Good point. As far as my results are concerned, I get no difference between internal or external reference. My previous examples were with a Leo Bodnar GPS reference. The attached screenshot is using the internal reference.