Siglent SVA1015X and SVA1032X 1.5, 3.2GHz Spectrum & Vector Network Analyzers

[IM3]

What is Samp detection in your screenshots ? Never used it 

It is very common to use sample detection in noise measurements. I believe this is even 'a must' by convention.
I just forgot about it during the 2nd measurement. Fortunately it doesn't matter much because we are not measuring at 0.1 dB now.

And as far as phase noise is concerned, there are many more reports of this. Just one with a quick search;

https://www.eevblog.com/forum/testgear/siglent-ssa-3021x-sideband-angel-wings/msg3170698/#msg3170698

[tautech]

What is Samp detection in your screenshots ? Never used it 

It is very common to use sample detection in noise measurements. I believe this is even 'a must' by convention.
I just forgot about it during the 2nd measurement.

Yet you still compared apples with oranges. 

[IM3]

Yet you still compared apples with oranges. 

That's quite a bold statement, so to say.

But I have done a new measurement for you in which you will see that the noise values with the sample detector are  better than with a peak measurement. The choice of the detector is therefore important.

In my post where I used a peak measurement (by mistake) I actually put the SSA3032X-R at a disadvantage and yet it was even better than the SVA1032X.

[IM3]

Let's consider for a moment the consequences of bad phase noise.

In the 1st screenshot (SSA3032X-R) you see a carrier.
You don't really want to see the skirts at the bottom. The measured carrier is actually clean, but it is the quality of the spectrum analyzer that causes this to appear. It is phase noise from the (several) local oscillators in the analyzer.

Take a look at the 2nd screenshot.
This is an FPL1007 from Rohde&Schwarz. You hardly see skirts here. That's how you want it. (bandwidth here is 10kHz for display convenience)

Now let's introduce a 2nd carrier.

On screenshot 3 (SSA3032X-R) you see the same carrier and to the left of it a 2nd carrier of -80 dBm.

On screenshot 4 (SSA3032X-R) you can see what happens when we bring the -80dBm carrier closer to the -20 dBm carrier. It is barely visible. It disappears in the phase noise of the SSA3032X-R analyzer.

On screenshot 5 (FPL1007 R&S) you can see that with low phase noise of the analyzer the carrier of -80 dBm is still clearly visible.

Of course the FPL1007 is more expensive but don't forget that my SSA3032X-R has the same hardware as the 3075X-R. Then you no longer talk about entry-level models. Especially not if you buy a few expensive options.

[2N3055]

Of course the FPL1007 is more expensive but don't forget that my SSA3032X-R has the same hardware as the 3075X-R. Then you no longer talk about entry-level models. Especially not if you buy a few expensive options.

While we are comparing numbers, let's put thing in perspective...

Even SSA3075X-R is still half the price of FPL1007... And that is for FPL1007 base price without any options, that are, by the way, even more expensive on R&S. For instance: 40 MHz BW realtime option on R&S cost 4000 € alone! While it is actually free on SSA3075X-R now...

For 4000€ (only for realtime option on R&S) you can buy ALL the options for SSA3075X-R, that includes a set of nearfield EMC probes...

So my comment on this is : " 2-3x times more expensive R&S SA actually has slightly better performance than SSA3075X-R?
Really? Well it effing better have better performance for the price you pay for it..."

[colorado.rob]

Let's consider for a moment the consequences of bad phase noise.

Is there a reason you use a different RBW when comparing the R&S to the Siglent?

[IM3]

Is there a reason you use a different RBW when comparing the R&S to the Siglent?

The only reason is for display convenience. If I had  the bandwidth of the FPL the same as  with the Siglent, the noise floor would have dropped below the screen. It was a quick choice to solve it this way. There was no ulterior motive behind it.

[nctnico]

Yet you still compared apples with oranges. 

That's quite a bold statement, so to say.

But I have done a new measurement for you in which you will see that the noise values with the sample detector are  better than with a peak measurement. The choice of the detector is therefore important.

In my post where I used a peak measurement (by mistake) I actually put the SSA3032X-R at a disadvantage and yet it was even better than the SVA1032X.

There is some extra noise going on at the left few divisions of the display. Is this still not fixed?

[IM3]

There is some extra noise going on at the left few divisions of the display. Is this still not fixed?

Looks like it indeed. I made a new screendump for you.

This analyser has productiondate january 2024

Maybey Tautech can remind a few people (again) to fix this bug.

[IM3]

Of course the FPL1007 is more expensive but don't forget that my SSA3032X-R has the same hardware as the 3075X-R. Then you no longer talk about entry-level models. Especially not if you buy a few expensive options.

While we are comparing numbers, let's put thing in perspective...

Even SSA3075X-R is still half the price of FPL1007... And that is for FPL1007 base price without any options, that are, by the way, even more expensive on R&S. For instance: 40 MHz BW realtime option on R&S cost 4000 € alone! While it is actually free on SSA3075X-R now...

For 4000€ (only for realtime option on R&S) you can buy ALL the options for SSA3075X-R, that includes a set of nearfield EMC probes...

So my comment on this is : " 2-3x times more expensive R&S SA actually has slightly better performance than SSA3075X-R?
Really? Well it effing better have better performance for the price you pay for it..."

Perhaps you do not understand the meaning of my previous post, because you mainly respond on using the expensive FPL.

What I am trying to explain is the importance of good phase noise properties of a Spectrum Analyzer. And as far as I'm concerned, that's not entirely clear if you only rely on the specifications that Siglent provides!

I may be treading on thin ice because there are many 'Siglent believers' on this blog.

But there is a bit of psychology involved. If someone buys a nice car after considering everything and thinks he has made the right decision, he will not easily accept it if someone explains to him that it is a bad brand and that he has made the wrong choice.

I'm not here to bash Siglent and I'm not saying it's a bad brand, but attention should be paid to the things they prefer not to mention.

It is still a Chinese company whose name is suspiciously similar to Agilent and they do their best to copy expensive Western items. There are so many examples that show that Siglent is far from having things in order. You notice it in the documentation and you notice it in the errors in the firmware.

[IM3]

Let's get back on topic:

Screendump_1:
First an addition to measuring phase noise:
As stated earlier, you measure this with a sample detector and NOT with a peak detector. Signent also writes this in his service doc.

As a consequence, you may lose the carrier on the spectrum, but that does not matter for the measurement.

Screendump_2:
Siglent stated that the phasenoise is -98dBc@10 kHz and -97dBc@100kHz
What they don't tell you is that the phase noise increases beyond 300kHz to -97dBc and then drops to -114dBc@1MHz. This makes the skirt larger than expected.   (there is hardly a carrier to be seen as stated because of sample detector)

Screendump_3:
 You can see the difference with the FPL of R&S. What matters now is not that the phase noise is lower, but rather that the skirt is narrower.

Signent could make a good improvement there.

So I'm not complaining about the -98dBc, but I am complaining about its width.

Screen dump 4 & 5:
I tell you ones again that if you spend €11,000 on, for example, an SSA3075X-R (without options) you may be disappointed if your phase noise looks like screendump 4 or 5.

At €11,000 they are really not looking for the hobbyist, so they have to make improvement on that.

[2N3055]

It is still a Chinese company whose name is suspiciously similar to Agilent and they do their best to copy expensive Western items. There are so many examples that show that Siglent is far from having things in order. You notice it in the documentation and you notice it in the errors in the firmware.

[EE-digger]

This is OT from the phase noise but I guess this is a MT (multi topic) thread 

In the process of downsizing and eliminating several higher end gear, I'm headed toward either an SVA1075X or possibly a -R of the SSA series.  One thing I have not seen on this blog, flog me if wrong, is the SOLT performance of these units.  In general, it looks like the -R may be either a superior design or given more attention, possibly a few better components.

Can you post an image of your SOLT cal with the load performed last, in LOG form.  How stable is it over an hour or two (assuming analyzer has been warmed for an hour)?

Higher end VNAs are fairly rock solid.  The early FieldFox wanders like crazy.  I'm scared sh-less over the four leaf clovers out of the Siglents Smith chart prior to cal.  I am seeing better than -60dB on the bench VNA and the FF.

I would like to think that Siglent's converters and dynamic range are good enough to get down there but have seen NO plots of such.

[tautech]

In general, it looks like the -R may be either a superior design or given more attention, possibly a few better components.

It's not quite that simple.

X-R models certainly use different HW to most SSA/SVA models but not all.
We can clearly see evidence of this in SSA+, SVA and SSA X-R datasheets.

With some study we can identify 3 HW versions that cover all models from 1.5 - 7.5 GHz.

[EE-digger]

In general, it looks like the -R may be either a superior design or given more attention, possibly a few better components.

It's not quite that simple.

X-R models certainly use different HW to most SSA/SVA models but not all.
We can clearly see evidence of this in SSA+, SVA and SSA X-R datasheets.

With some study we can identify 3 HW versions that cover all models from 1.5 - 7.5 GHz.

I'm sure you're right but what I'm really after is visibility to how well it calibrates and how stable that calibration is.  In spite of videos and web reviews, there isn't much at the VNA end and nothing at the cal end.  A few more looks at how well it behaves up to 7+ GHz would also be helpful.

Hey, you could be an industry leader.  Just get Siglent to support you and your US and worldwide counterparts with a virtual try out online.  You select the instrument which is then connected to a matrix of sources, loads, reflectances, etc.  Our internet bandwidths are high enough to support that fairly well today.

[tautech]

In general, it looks like the -R may be either a superior design or given more attention, possibly a few better components.

It's not quite that simple.

X-R models certainly use different HW to most SSA/SVA models but not all.
We can clearly see evidence of this in SSA+, SVA and SSA X-R datasheets.

With some study we can identify 3 HW versions that cover all models from 1.5 - 7.5 GHz.

I'm sure you're right but what I'm really after is visibility to how well it calibrates and how stable that calibration is.  In spite of videos and web reviews, there isn't much at the VNA end and nothing at the cal end.  A few more looks at how well it behaves up to 7+ GHz would also be helpful.

Currently only have my SNA5004A 4.5 GHz unit to test.....

Hey, you could be an industry leader.  Just get Siglent to support you and your US and worldwide counterparts with a virtual try out online.  You select the instrument which is then connected to a matrix of sources, loads, reflectances, etc.  Our internet bandwidths are high enough to support that fairly well today.

Similar has been asked for before as we have used other equipment via the inbuilt webserver of a beta tester in Oregon.
For Siglent to support such online trial of equipment bookings would be necessary where of course other than using equipment as it has previously been connected and specific port settings for remote connections ....something that I wouldn't know where to start.

Many of us resellers have some demo units so ask around if some close by have something you can visit and test.

[IM3]

This is OT from the phase noise but I guess this is a MT (multi topic) thread 

In the process of downsizing and eliminating several higher end gear, I'm headed toward either an SVA1075X or possibly a -R of the SSA series.  One thing I have not seen on this blog, flog me if wrong, is the SOLT performance of these units.  In general, it looks like the -R may be either a superior design or given more attention, possibly a few better components.

Can you post an image of your SOLT cal with the load performed last, in LOG form.  How stable is it over an hour or two (assuming analyzer has been warmed for an hour)?

Higher end VNAs are fairly rock solid.  The early FieldFox wanders like crazy.  I'm scared sh-less over the four leaf clovers out of the Siglents Smith chart prior to cal.  I am seeing better than -60dB on the bench VNA and the FF.

I would like to think that Siglent's converters and dynamic range are good enough to get down there but have seen NO plots of such.

Maybe I can help you partially with the following measurement;

My SSA3032X-R doesn't go beyond 3.2 GHz (because I didn't enter a secret key) but it might still give you some useful information. I think the SSA3032X-R is equivalent to the SVA1075X on this subject.

I ran a SOL on my SSA3032X-R and used a SOL  (R&S) specified to 4GHz straight onto the N-connector of the VNA. The tracking generator at 0 dBm

Screen dump_1:
From 1 MHz to 3.2 GHz the reflection attenuation is slightly better than -50dB.

Screendump_2:
I tried a narrower span to see if I could get a little deeper into the noise floor. Span 100 MHz @ fc 3 GHz
The reflection attenuation appears a little more towards -55 dB.

It is difficult to place a cursor because the noise floor varies too much. Remarkably, an average of 100x does not change that...?
It seems to me that averaging does not work in VNA mode.

Unfortunately, the problem is that this VNA have a fixed bandwidth setting of 10 kHz in VNA mode. That causes the return signal to disappear into its noise.

It would be nice if Siglent solved that in the near future. It's not clear to me why they don't implement this.

By the way, a reflection attenuation of -50 dB is a very nice value, but that mainly depends on the job you are doing.

Screendump_3:
An overview of my setup.

Screendump_4:
A quick check whether your VNA is sufficiently calibrated by using an open end rigged coax.

[EE-digger]

Thanks IM3 !  That's exactly what I was looking for.  I'm not asking you to run it again, but have you noticed that it remains stable for some time?

Yes, -50dB is a decent number, especially if it's repeatable and stable.  For small chip and pcb antenna work, the final result is usually well above -30dB and can be a lot worse but acceptable.

My other units are PNA and early FieldFox, both of which may be gone with retirement downsizing.  They are about 20dB (at 10kHz IFBW) lower but the FieldFox drifts a bit, while the PNA is rock solid (should be at 55lbs / 25kg).  The difference must be in the directional couplers.

And of course, this is where Siglent's 38dB to 40dB corrected directivity comes from.

[IM3]

Thanks IM3 !  That's exactly what I was looking for.  I'm not asking you to run it again, but have you noticed that it remains stable for some time?

Yes, -50dB is a decent number, especially if it's repeatable and stable.  For small chip and pcb antenna work, the final result is usually well above -30dB and can be a lot worse but acceptable.

My other units are PNA and early FieldFox, both of which may be gone with retirement downsizing.  They are about 20dB (at 10kHz IFBW) lower but the FieldFox drifts a bit, while the PNA is rock solid (should be at 55lbs / 25kg).  The difference must be in the directional couplers.

And of course, this is where Siglent's 38dB to 40dB corrected directivity comes from.

Glad I could help you.

Stability is not a problem. That is more than sufficient, but that is actually no longer a problem nowadays. When I think about what we had to do in the past to achieve 0.1 ppm, that is no problem nowadays.

Downgrading from high-end equipment such as KeySight, R&S, TEK, etc., takes time to get used to.

The choice not to opt for the big brands but for Siglent or Rigol, etc., is always motivated by money!

You really get less.

And that's fine. But you will notice that many things are not as you are used to.

Measuring equipment is never cheap. If you want to pay less, you have to realize that you will also compromise. That's almost a law of nature.

I think that Siglent makes reasonably good stuff for the money, but with Spectrum Analysers (and VNAs) it really comes down to good hardware engineering and there is still room for improvement.

In addition, there are so many errors in the user interface, which is just software.

A very small example that seems meaningless.
For example, with the SSA or SVA (X, plus, X-R) from Siglent you can change the trace brightness but not the overall screen brightness. It seems small but gives a less good look and feel. It's also annoying.

You also often cannot easily take a step back in the screen menu because there is simply no return or back button. So you have to start going through the menu at the very beginning again. Very annoying. But that's what you get included for free.

If I may advise, And you would go for Siglent than I would buy an SSA3032X-R as I did after I got rid of my SVA1032X.
This device has the best hardware of the series and can easily be upgraded for free with all options  to 7.5 GHz and will cost you around €5000.
(kudos for the hard work some did  to realise this )

That is the big advantage of buying Chinese stuff.   

[EE-digger]

You've made many good points.  And it does come down to minor annoyances, minor errors and minor omissions.

Fortunately, Siglent seems fairly responsive.  If you could not access a critical or basic function, I'm sure they would fix it quickly (I hope).

You hit on one of my minor annoyances.  The Agilent PNA and all of their 2000, 3000A/T/G series scopes have no screen brightness control.  Fortunately, you do have trace brightness for those at low waveform rates.  The FieldFox has good menu back function and also brightness control, primarily because of the wide environmental variation it is used in.

I see you have an R&S unit and know the differences.  I don't dare compare the PNA at $65,000 (2012) to the SSA or SVA series.  But if the Siglent gets the job done at low cost, at 10lbs vs 110lbs for my PNA and SA (PSA series), then that's good.

On the SA side, the SSA is somewhat better than the old, big ticket analyzers.  It has around 10dB over the PSA on DANL.  Phase noise is about equal and a bit better on some SSA model variants.

[WT4DX]

I've recently been lurking the blog doing my due diligence in search for a viable SA/ VNA for amateur radio use, and use as a hobbyist. I was about to pull the trigger on a SVA1015X when IM3's post came up. I would like to mention that I do appreciate the level of expertise in this forum.

I had considered the SSA3032X-R, but do I really need a RTSA, probably not. Is it a a really cool function, yes it is! Looking at the data provided by IM3 the 3032X-R is back in the radar.

I'd hate to invest the extra capital just for the RTSA feature if all of the other performance characteristics are relatively similar between the SVA1015X, and the SSA3032X-R.

From the most recent information in the thread, I want to verify by consensus that the SSA3032X-R does indeed have better phase noise characteristics and better HW than the SVA-1015X?

Thank you for the consideration of my question.

[IM3]

I've recently been lurking the blog doing my due diligence in search for a viable SA/ VNA for amateur radio use, and use as a hobbyist. I was about to pull the trigger on a SVA1015X when IM3's post came up. I would like to mention that I do appreciate the level of expertise in this forum.

I had considered the SSA3032X-R, but do I really need a RTSA, probably not. Is it a a really cool function, yes it is! Looking at the data provided by IM3 the 3032X-R is back in the radar.

I'd hate to invest the extra capital just for the RTSA feature if all of the other performance characteristics are relatively similar between the SVA1015X, and the SSA3032X-R.

From the most recent information in the thread, I want to verify by consensus that the SSA3032X-R does indeed have better phase noise characteristics and better HW than the SVA-1015X?

Thank you for the consideration of my question.

As always a lot of things depends on budget, so I can't help with that, but I can explain how I came to my choice.

For now there are 2 main RF designs of interest within the Siglent spectrum analyzers. (The 3rd choice is the 5000 series and too expensive)

The better RF design of the 2 can be recognized in the documentation by a DNAL of -165 dBm.
The simpler RF designs have a more worse DNAL value. More like -156 dBm and -161dBm.

I think all beter ones are the SVA1075X, SSA3075X-PLUS, SSA3032X-R, SSA3050X-R, SSA3075X-R.

Now what I had already described in an earlier post is that the simpler RF design of my SVA1032X has led to poor phase noise in a large part of the spectrum. In any case, worse than -98dBc that Siglent specifies at 1GHz.
The better RF design doesn't have that problem and have the same phase noise across the entire spectrum.

To see how physically different the designs are, I advise you to view 2 links.

For the more simple RF design, watch a video from EEVBlog here:


Click here for a look at the better RF design.  (Scroll to reply 53 & 54)
https://www.eevblog.com/forum/testgear/siglent-ssa3000x-r-57-5ghz-real-time-spectrum-analyzer/?all

As far as I know the only photos on the internet of this version. (Thanks to TV84 on this blog)

How I made a choice which analyzer to buy:
Of the SVA series, only the SVA1075X is suitable, but then you pay €9200.

Of the SSA X-PLUS series, only the SSA3075X-PLUS is suitable, but then you pay €7875. But it has no VNA function. I thought that could be upgraded, but it would require some tinkering.

The SSA3032X-R is the cheapest of the SSA X-R series and costs €4830, but you can expand it to 7.5 GHz for free, including all options with the key generator that you can download here.
http://tpcg.io/_SV2SRE  (Many thanks to the guys who made this script)

In my opinion, the SSA3032X-R is the device that gives you the most for the least money.

And real time is nice to have, really.

[WT4DX]

I've recently been lurking the blog doing my due diligence in search for a viable SA/ VNA for amateur radio use, and use as a hobbyist. I was about to pull the trigger on a SVA1015X when IM3's post came up. I would like to mention that I do appreciate the level of expertise in this forum.

I had considered the SSA3032X-R, but do I really need a RTSA, probably not. Is it a a really cool function, yes it is! Looking at the data provided by IM3 the 3032X-R is back in the radar.

I'd hate to invest the extra capital just for the RTSA feature if all of the other performance characteristics are relatively similar between the SVA1015X, and the SSA3032X-R.

From the most recent information in the thread, I want to verify by consensus that the SSA3032X-R does indeed have better phase noise characteristics and better HW than the SVA-1015X?

Thank you for the consideration of my question.

As always a lot of things depends on budget, so I can't help with that, but I can explain how I came to my choice.

For now there are 2 main RF designs of interest within the Siglent spectrum analyzers. (The 3rd choice is the 5000 series and too expensive)

The better RF design of the 2 can be recognized in the documentation by a DNAL of -165 dBm.
The simpler RF designs have a more worse DNAL value. More like -156 dBm and -161dBm.

I think all beter ones are the SVA1075X, SSA3075X-PLUS, SSA3032X-R, SSA3050X-R, SSA3075X-R.

Now what I had already described in an earlier post is that the simpler RF design of my SVA1032X has led to poor phase noise in a large part of the spectrum. In any case, worse than -98dBc that Siglent specifies at 1GHz.
The better RF design doesn't have that problem and have the same phase noise across the entire spectrum.

To see how physically different the designs are, I advise you to view 2 links.

For the more simple RF design, watch a video from EEVBlog here:


Click here for a look at the better RF design.  (Scroll to reply 53 & 54)
https://www.eevblog.com/forum/testgear/siglent-ssa3000x-r-57-5ghz-real-time-spectrum-analyzer/?all

As far as I know the only photos on the internet of this version. (Thanks to TV84 on this blog)

How I made a choice which analyzer to buy:
Of the SVA series, only the SVA1075X is suitable, but then you pay €9200.

Of the SSA X-PLUS series, only the SSA3075X-PLUS is suitable, but then you pay €7875. But it has no VNA function. I thought that could be upgraded, but it would require some tinkering.

The SSA3032X-R is the cheapest of the SSA X-R series and costs €4830, but you can expand it to 7.5 GHz for free, including all options with the key generator that you can download here.
http://tpcg.io/_SV2SRE  (Many thanks to the guys who made this script)

In my opinion, the SSA3032X-R is the device that gives you the most for the least money.

And real time is nice to have, really.

There's no question that there are differences between the two different RF designs, and thank you for sharing as well as sharing your decision making process.

Also, thank you for sharing the information on the keygen. I'll admit that I'm not 100% sure how the compiler works, but I'm sure that I can figure it out.

[maxi]

My SVA1032X : Firmware_V3.2.2.6.0R7 original Version with all Options

yellow: 10MHz, blue: 200MHz, red: 500MHZ, green: 1GHz

[WT4DX]

My SVA1032X : Firmware_V3.2.2.6.0R7 original Version with all Options

yellow: 10MHz, blue: 200MHz, red: 500MHZ, green: 1GHz

(Attachment Link)

What has been your overall impression of the SVA1032X,  and what signal generator where you using for the posted test?

Thanks