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

[regenfreak]

I have found the calibration menu rather clunky and not intuitive compared with all of the NanoVNA models. For S21 measurement, the calibration step with a 50 ohm load can be skipped. I always thought it is necessarily to do 50 ohm load calibration in S21? Now I can do Open, Short and Response Through with or without doing the "Port 1 Cal that requires the use of 50 ohm load". When I turned the Port 1 Cal on and off, I get slightly different responses. Also I have compared the results with my NanoVNA SAA-2 V2(see attachments),  while the S11 and S21 are fairly close, but the Smith charts are very different. The attached photo is the sweep of a homebrew filter. Also the * symbol that shows the completion of calibration is inconsistent. Sometimes * does not appear even I complete and confirm the calibration on the touch screen prompt. It is so easy to display four traces with the NanoVNA while it takes me a couple of steps to display multiple traces with the SVA1015X.

[regenfreak]

OK, for S11 measurement, we have to use port 1 Cal and then save the trace. This requires open, short and load. The calibration menu looks confusing because it has open and short beside the port 1 cal (which also has open, short and load in the sub menu).

For S21 measurement, only through calibration is needed. This raises the question why Siglent introduces two sets of open and short menu buttons? Now I don't know if S11 and S21 can be measured simultaneously with SVA1015X like the NanoVNAs. It may need independent calibrations and measurements for S11,S21, Smith chart and SWR separately like the Rigol VNA in the link below. I have got this information from the Tek VNA and HP manuals:

https://download.tek.com/document/70W_60918_0_Tek_VNA_PR.pdf

http://web.ecs.baylor.edu/faculty/baylis/eel4421/Procedure7.pdf

and also Rigol VNA demo:

https://youtu.be/ptXTblLFUBc

My Smith charts look different because there are many Smith chart axis options in the SVA1015X. In the NanoVNA, it shows real and imag.


There are actually many powerfully features in the SVA1015X. From now on, I will have to develop a new mindset adapting to the more advanced features.

[hendorog]

The Open and Short top level options are actual calibrations in their own right.
They are not related to the Open and Short steps in an OSL calibration.

The idea is that you can do a rough cal with just one standard.

[regenfreak]

yes i have just noticed that. If i do port 1 calibration first and repeat open cal, it will overwrite port 1 calibration. So does short cal, it will overwrite open cal. I dont see the point of having this menu design. A half-arsed calibration like that has no real value in a proper characterisation of DUT.

[hendorog]

It is just a way of providing a fast calibration.
Saves on cal kit wear and is also implemented on Keysight gear.
https://na.support.keysight.com/pna/help/latest/S3_Cals/Select_Cal.htm#openshort

I just turned my SSA on and did an open cal with the port open, and then put on a load. It was < 30dB up to almost 7GHz.

So certainly useable for measuring an antenna for example.

[regenfreak]

I see. It makes more sense now.
It will take some time for me getting used to the menu. It is interesting to read how about old, mega expensive VNAs (one of those still uses floppy disc) are calibrated.

[regenfreak]

I am flabbergasted and depressed. The new replacement unit has developed intermittent faults after two days of testing (on top of the black screen problem when i first powered it up .)  It was working fine until today that I have encountered repeated occurrence of these weird faults many times (all random):

1. The Smith Chart is very noisy before after and calibration. The traces are jumping around across the screen after calibration when it is open or with a 50 ohm load (see screen shot of the video). The noise is bad sweeping from 10MHz to 1.5G. It looks more horrible at lower frequency.  I switched it on and off, sometimes it is very noisy and sometimes it is normal. I  checked the connectors and calibration standards with NanoVNAs. The SMA adaptors  are all fully tightened.

2. The TG and VNA are intermittently not working, completely unresponsive ,only showing the noise floor (this was the same issue with the first unit)

I made a number of videos that show the faults as evidence, in case Telonic think I make it up.

The attached images show the results when the unit was working:
 (a) Wes Hayward's di-directional amplifier from 100k to 30MHz vs NanoVNA V2, it shows correct response down to about 2.5MHz which is OK.
(b) 467Mhz cavity bandpass filter
(c) A chinese 20db low noise amplifier

When it works, it is great. When it plays up, I am pulling my hair out.

[tautech]

When it works, it is great. When it plays up, I am pulling my hair out.

Please don't think I'm taking the piss but a good USA friend with new SNA5014A had similar problems until he got on top of connector cleanliness as outlined in this Siglent video:

[hendorog]

Start from basics to confirm:

Do a factory reset.
Disconnect everything
Do an Open Cal with nothing on the port. Dont use your open standard.
Attach a load and show the S11 trace screenshot.
If that is inconsistant then either your load standard is faulty or its dirty, or the device is faulty.

[regenfreak]

The noisy traces happen when there is NOTHING attached to the VNA before and after calibration. I have to switch the unit on and off whenever it happens.  I have three NanoVNAs and have tested three sets of calibration standards. This problem has happened even I use different sets of calibration standards. I am 100% it is nothing to do with the calibration standards because I have checked them with my NanoVNAs doing side by side comparison. The SMA-N adaptors are new, and I have looked at the connectors under microscope, they are perfectly fine. I have a few videos proof showing this random fault happening so no one can blame the  calibration standards. I feel very stupid when the second unit seems to be a lemon too. Do they have a bad batch?

If you look carefully at the screenshot of the video, there is nothing connected to the port. The noisy traces are like flying sparks from a tesla coil!

[regenfreak]

i have checked all new N-SMA connectors under a microscope; no bent pin. So they can't blame the connectors.

[hendorog]

The response at the low freq end of the SVA devices was poor, and so you could just be seeing the low freq 'tail' of the smith chart jumping around.
You have already noted it is fine above 2.5MHz. Below that it will bounce around on the Smith chart. So start the sweep at 2.5Mhz and see what happens.
This spec was improved in the newer units, but I don't believe it is great. I don't have one to try and replicate.

Other points:
You need to simplify things and start from a known base. This is troubleshooting 101.
Switch to LogMag images only for now, it is much easier to look at and interpret.
Factory reset is not the same as turning it off and on. You really need to factory reset before reporting a fault so that others can replicate the issue.
You're testing active devices, so perhaps you reduced the TG output level? This could be making the S11 low freq response worse.
It is hard work to follow your pictures, there is too much going on, too many traces. 
Your DUT's are too complex. If you want to demonstrate a fault with the unit then test something simple with a known response - like an attenuator or a load.
The marker colours are not matching either - which appears to be a screenshot bug worth reporting.
Have you checked to see if you are using the latest firmware?
Also, I have a set of new N-SMA adapters which I think I got from MiniCircuits which look clean and perfect but cause problems. But agree, not likely a problem at 1MHz.

Factory reset. One trace per picture. Start with the simplest calibration possible and then measure some VERY simple device.
Simple device would be an attenuator if you have one, or a load.
Ideally make a video showing these steps, if it shows an issue then stick it up on Youtube for us to review. Then if we can't solve it we can send it direct to the Siglent engineers.

You may well be correct and the device could well be faulty.
But you could also be suffering from confirmation bias caused to the frustration of having so many problems.

[regenfreak]

The response at the low freq end of the SVA devices was poor, and so you could just be seeing the low freq 'tail' of the smith chart jumping around.
You have already noted it is fine above 2.5MHz. Below that it will bounce around on the Smith chart. So start the sweep at 2.5Mhz and see what happens.
This spec was improved in the newer units, but I don't believe it is great. I don't have one to try and replicate.

Other points:
You need to simplify things and start from a known base. This is troubleshooting 101.
Switch to LogMag images only for now, it is much easier to look at and interpret.
Factory reset is not the same as turning it off and on. You really need to factory reset before reporting a fault so that others can replicate the issue.
You're testing active devices, so perhaps you reduced the TG output level? This could be making the S11 low freq response worse.
It is hard work to follow your pictures, there is too much going on, too many traces.
Your DUT's are too complex. If you want to demonstrate a fault with the unit then test something simple with a known response - like an attenuator or a load.
The marker colours are not matching either - which appears to be a screenshot bug worth reporting.
Have you checked to see if you are using the latest firmware?
Also, I have a set of new N-SMA adapters which I think I got from MiniCircuits which look clean and perfect but cause problems. But agree, not likely a problem at 1MHz.

Factory reset. One trace per picture. Start with the simplest calibration possible and then measure some VERY simple device.
Simple device would be an attenuator if you have one, or a load.
Ideally make a video showing these steps, if it shows an issue then stick it up on Youtube for us to review. Then if we can't solve it we can send it direct to the Siglent engineers.

You may well be correct and the device could well be faulty.
But you could also be suffering from confirmation bias caused to the frustration of having so many problems.

You misunderstood my post. It is easier to show the problems with the youtube videos:

Noisy Smith chart:
https://youtu.be/t-wb9Cz4c1s

Black screen problem:
https://youtu.be/xJvOhKnAI1U


The unit is working normally and correctly most of the time except. I encountered the intermittent problems with noisy Smith chart even i don't attach any DUT to it before and after calibration. The examples I posted are correct. I have not complained about the measurements! They are not complicated, just S11, S21, Smith chart and SWR. It is very useful and meaningful  to look at them by-to-side. They all make sense to me but maybe most people avoid Smith chart.  Of course it has the latest firmware and i did preset many times. Remember i have had the unit only two days.

I think the TG and VNA are OK.

[hendorog]

Thanks, it helps to have the video. The text is unreadable, so your words are very important.
Your measurements make sense to you, but if you were on my side looking at your screenshots you would understand.
The confusing thing is some of the information in your screenshots (S21) is correct and matches the NanoVNA.

You state in the video the default start freq is 5MHz - is that correct? I thought it was 100kHz from your screenshots?
This is a big difference and does make it look like a fault with S11 only.
Can you show an image or video of it working correctly for comparison?

Would you please show a video of the following:
Factory Reset.
One port cal.
Then take the load off the port.
If you can do it so the text is readable, and the screen and ports are visible then that would be most helpful to the engineers.

No information on firmware version?

[hendorog]

Ok I watched the video more closely.

The first time you did the 1 port cal you used the 50 ohm load for the short and the load.
Short/Load:
https://youtu.be/t-wb9Cz4c1s?t=181
The second time you did it correctly.

That is why it worked the second time and not the first.

[regenfreak]

Ok I watched the video more closely.

The first time you did the 1 port cal you used the 50 ohm load for the short and the load.
Short/Load:
https://youtu.be/t-wb9Cz4c1s?t=181
The second time you did it correctly.

That is why it worked the second time and not the first.

It is nothing to do with the calibration.  You cannot explain why the Smith chart is so noisy before I attached anything to unit! Can't you see how quiet the NanVNA is?  I did it three times in the video. It only worked after i switched it off in the third time.

I have the black screen problem again this morning:

https://youtu.be/egP5dX0NdG4

It is not supposed to do that!

This is another video showing the noisy traces (sorry for the handshake, I filmed with my phone):

https://youtu.be/9yDo8Nj0wc8

[regenfreak]

Another demo of the problem. This time I tightened the 50 ohm load with an expensive SMA torque wrench:

https://youtu.be/inXVlpaLY4E

The use of torque wrench is an overkill for quick test like this.

[hendorog]

Well I can only comment on what you show me, and that was a clear cut operator error which explained the results perfectly.

The calibration is one problem. It _does not mean_ the device is completely fine and I did not say that!
Here are the other issues that I can see:
The black screen is a problem.
The default calibration after factory reset should be much better than what your video shows. This could be because the device is set to Last or to a User config on Power On/Preset, and a bad cal has been saved.
And there appears to be a bug in the screenshot trace colours.
- There still may be a random tail flicking around if you set the start freq low enough though. I can't compare as I don't have a device.
The NanoVNA will be better in this regard as it does not have magnetics in its bridge.

So the first two things I would do are a Factory reset to eliminate any possibility of corrupt settings and remove any saved calibration (which will be restored after preset/power on if set by the user)
And a check of the firmware version and an upgrade if necessary.
There have been many bugs found and fixed.
The torque wrench is not required and is not the problem, but FWIW you are not using it correctly. The calibrated torque will only apply if your hand is at the line on the handle.

Aside from that I don't think there is anything else I can do to help.

[regenfreak]

Obviously I did not do any of what you described.  You keep making assumptions that i created the problems myself. If you never own the unit, you are simply making assumptions i am clueless and did stupid things to mess it up. 

[Bicurico]

I would return the second unit and ask for a full refund.

I would then buy the SSA3021X+ and read this thread from the beginning.

If the second unit is still working weird with black screens and strange behaviour, you are certainly entitled to just send it back for a full refund. No point trying this and that, doing a factory reset, etc. This device should work fine out of the box.

[regenfreak]

I would return the second unit and ask for a full refund.

I would then buy the SSA3021X+ and read this thread from the beginning.

If the second unit is still working weird with black screens and strange behaviour, you are certainly entitled to just send it back for a full refund. No point trying this and that, doing a factory reset, etc. This device should work fine out of the box.

Yes, finally I have got a sympathetic answer. I have not changed the factory settings or messed around with the default display or power settings. I always press preset which resets it to factory settings. It should not happen, full stop. This video shows another black frozen screen problem, I have to cut the power and switch it off. I will return it and ask for a refund. I was going to buy the SSA3021X+ but I made a wrong decision going for the VNA and SA hybrid. I really like the VNA when is working but I am completely put off by the intermittent problems.

https://youtu.be/VHqUKjxeupc

I own Siglent signal generator and oscilloscope, Rigol signal generator and oscilloscope, all purchased from Telonic in the UK.

[regenfreak]

Ok I watched the video more closely.

The first time you did the 1 port cal you used the 50 ohm load for the short and the load.
Short/Load:
https://youtu.be/t-wb9Cz4c1s?t=181
The second time you did it correctly.

That is why it worked the second time and not the first.

you are wrong I am afraid. Please watch 55s. At 1:00 I transferred the load from NanoVNA to Siglent VNA. At 1.25s I incorrectly said it is registered open, it should be short.  Also you ignore all evidence presented to you . There are lots of noisy traces flying across  the Smith chart. You are wrong to assume the load is longer than others. My load is quite short.

You try to tell me I cannot tell what is a 50 ohm load?  Come on. I have both long and short 50 ohm loads. Anyone who has used a VNA would notice that  the traces should not fly across  the screen like that without or with calibration.

[Bicurico]

The hint was that you can hack a SSA3021X+ (the plus model) into a SVA1032.
It will cost you little more money than the SVA1015X. I guess this model is not sold much anymore...

[regenfreak]

The hint was that you can hack a SSA3021X+ (the plus model) into a SVA1032.
It will cost you little more money than the SVA1015X. I guess this model is not sold much anymore...

I would never do that. I do not want to waive my warranty. So far those cheap NanoVNAs can do the same measurements as SVA1015x. I must say I really like SVA1015X when it is working but I have packed it up.

 I dont want to spend another minute thinking about it or try to argue about it.

[hendorog]

Ok I watched the video more closely.

The first time you did the 1 port cal you used the 50 ohm load for the short and the load.
Short/Load:
https://youtu.be/t-wb9Cz4c1s?t=181
The second time you did it correctly.

That is why it worked the second time and not the first.

you are wrong I am afraid. Please watch 55s. At 1:00 I transferred the load from NanoVNA to Siglent VNA. At 1.25s I incorrectly said it is registered open, it should be short.  Also you ignore all evidence presented to you . There are lots of noisy traces flying across  the Smith chart. You are wrong to assume the load is longer than others. My load is quite short.

You try to tell me I cannot tell what is a 50 ohm load?  Come on. I have both long and short 50 ohm loads. Anyone who has used a VNA would notice that  the traces should not fly across  the screen like that without or with calibration.

Actually I am correct. You did confuse them, and i simply pointed it out, because that will cause the problem you are seeing.

Then you posted another video later where you do demonstrate the problem. This is the one you are talking about.

I'm actually not arguing with you, i am trying to help you to be sure it is faulty before you go through the hassle of sending it back again.

Anyway. Whatever happens, all the best, I have done all I can.