How do the features in old VNAs and SAs differ from new ones?

[G0HZU]

So in this respect there isn't even a contest. The HP is a different class of VNA because it is a full 2 port VNA. The Siglent SA/VNA doesn't qualify for the contest because it's just a T/R VNA.

The Siglent SSA3000X-R can do S11 and S21 measurement using two ports. It will not do S22 or S12 as only the one port can transmit. You will therefore have to switch the ports manually to get those done.

In other words, the Siglent SSA3000X-R is just a T/R VNA. So it doesn't qualify as a full 2 port VNA.

[TimFox]

Two useful reference books about the previous generation of swept-spectrum SAs with digital control and analog processing (ca. 1990) with examples from Tektronix.
(1) M Engelson Modern Spectrum Analyzer Measurements JMS, 1991;  Practical uses.
(2) M Engelson Modern Spectrum Analyzer Theory and Applications Artech House, 1984;  More mathematical background than (1).

[baldurn]

So in this respect there isn't even a contest. The HP is a different class of VNA because it is a full 2 port VNA. The Siglent SA/VNA doesn't qualify for the contest because it's just a T/R VNA.

The Siglent SSA3000X-R can do S11 and S21 measurement using two ports. It will not do S22 or S12 as only the one port can transmit. You will therefore have to switch the ports manually to get those done.

In other words, the Siglent SSA3000X-R is just a T/R VNA. So it doesn't qualify as a full 2 port VNA.

The OP said he needed it to build an antenna. Besides you can do most things by switching the cables. It is just not as convenient. You can get all four S parameters.

I am not sure the claim that it will do worse measurements just because of only having one transmit port can be correct. But I have read that spectrum analyzers are build as SA first and the VNA part is tacked on trying to reuse circuits made for SA and therefore typically has worse performance. Unless OP can specify some requirements, I am going to assume the performance is more than good enough for what was described.

Siglent does have a device for 2 and 4 ports VNA with all ports equal. It is the SNA5000A but it is over budget even for the cheapest option. This does include a SA option and in this case it is the opposite of the SSA3000X-R: The SNA5000A is VNA first and SA second. Does seem to lack the protocol decodes and the RT option, which they are selling, does seem to be missing in the manual.

[joeqsmith]

It is just not as convenient.

Missing the more advanced error corrections.

[G0HZU]

baldurn: I am not sure the claim that it will do worse measurements just because of only having one transmit port can be correct. But I have read that spectrum analyzers are build as SA first and the VNA part is tacked on trying to reuse circuits made for SA and therefore typically has worse performance. Unless OP can specify some requirements, I am going to assume the performance is more than good enough for what was described.

Have a look through this doc

https://www.kirkbymicrowave.co.uk/Support/Links/application-notes/HP-Agilent-Keysight/Keysight-Applying-Error-Correction-to-Network-Analyzer-Measurements_5965-7709E.pdf

The first couple of pages show how much better a full two port calibration is. See figure 1 that shows a full two port cal vs the response cal. The response cal has a lot of ripple on it. However, the ripple can sometimes be a lot worse than this.

It is possible to use external attenuators on the ports to improve a T/R VNA when measuring S21 and S12, I did this for many years using an old HP 8714B T/R VNA. It is very clunky to do this, especially if trying for all four s-parameters. I did this for many years until I upgraded to a more capable VNA.

[baldurn]

The first couple of pages show how much better a full two port calibration is. See figure 1 that shows a full two port cal vs the response cal. The response cal has a lot of ripple on it. However, the ripple can sometimes be a lot worse than this.

But the Siglent SSA3000X-R will be doing the full two port calibration named "Short-Open-Load-Through (SOLT)" above figure 1?

Maybe we have a misunderstanding. It can only transmit on one port but it will receive on two ports. It will therefore do a two port through calibration just fine. Naturally in addition to the short, open and load calibration.

[joeqsmith]

Not SOLT but rather some of the more advanced methods.  Look up unknown thru for example. 

[G0HZU]

The first couple of pages show how much better a full two port calibration is. See figure 1 that shows a full two port cal vs the response cal. The response cal has a lot of ripple on it. However, the ripple can sometimes be a lot worse than this.

But the Siglent SSA3000X-R will be doing the full two port calibration named "Short-Open-Load-Through (SOLT)" above figure 1?

Maybe we have a misunderstanding. It can only transmit on one port but it will receive on two ports. It will therefore do a two port through calibration just fine. Naturally in addition to the short, open and load calibration.

OK, I can try and explain my point. I think the  Siglent SSA3000X-R is a T/R VNA and it can do something called an enhanced through calibration. This corrects for the port 1 mismatch but this isn't as good as a full 2 port calibration because it can't correct for the port 2 mismatch. Port 2 won't be a perfect 50R port. It will have some mismatch.

By contrast a genuine full 2 port VNA can typically do a 12 term correction that includes correcting for port 2 mismatch.

If it helps, the red trace in the plot below shows what a perfect VNA could measure (in terms of S21) for a fairly short 25 ohm transmission line between port 1 and port 2.

The red trace is close to what a proper lab VNA would show if it supported a full 2 port calibration. This is the correct response.

If the T/R VNA had an uncorrected port 2 VSWR of 1.3:1 (but everything else was perfectly calibrated correctly) then the best you can hope for is an uncertainty window represented by the blue trace.

Can you see how grim the T/R VNA performance is in this comparison? It is possible to add an attenuator at port 2 of the T/R VNA to improve things but this comes with the penalty of reduced dynamic range.