Best VNA for around or under $2000?

[mark432]

Does LiteVNA64 actually allow you to define cal standards?

From looking at the manual it seems that it just assumes the SOLT are perfect devices.
This seems a bit bonkers to me for something that is going up to 6 GHz.

I doubt that the short and open standards they provide line up perfectly with the reference plane.  You would think you could at least set the parameters of the thru.
But the even bigger point is cables/adapters.  Sometimes you need to measure things at the other end of cables and adapters. 

I have a regular nanoVNA, and I didn't worry much about it because of lower frequency limit.

I suppose you just have to measure a reference standard and then correct your data after the fact.
Still, I wonder what the phase error after calibration is.

Regarding the LibreVNA.  I see that an electronic calibration module is hitting the market.  I wonder why they made it a 4 port cal module, when the only librevnas I can find are 2 port?
Is there a 4 port model in the works?  I doubt it, but I would love a dual-source 4 port LibreVNA.

[knudch]

You can do 4 port by using 2 LibreVNA
Supported in SW, called compound device

As understand it, 4 port can be done with 2 of shelf LibreVNA
8 if you some additional PCB's

But you will not have phase information between each unit...as I understand it

Have never tried it as I don't have 2 LibreVNA's

There also something called Mixed Mode....please read the support mail list or ask the designer

Knud

[joeqsmith]

Does LiteVNA64 actually allow you to define cal standards?

Going thought the menus of a fairly recent version of firmware,  it does not appear to.

From looking at the manual it seems that it just assumes the SOLT are perfect devices.
This seems a bit bonkers to me for something that is going up to 6 GHz.

Well, it was designed for a certain price.   I am not sure if the hardware has enough resources to support it (code space, processing power...).       

I doubt that the short and open standards they provide line up perfectly with the reference plane.  You would think you could at least set the parameters of the thru.
But the even bigger point is cables/adapters.  Sometimes you need to measure things at the other end of cables and adapters. 

Bigger point?  Normally I would cal these out.   I normally have cables attached and measure on the other end. 

I have a regular nanoVNA, and I didn't worry much about it because of lower frequency limit.

I suppose you just have to measure a reference standard and then correct your data after the fact.
Still, I wonder what the phase error after calibration is.

Like how the LibreVNA requires a PC,  if you want to perform more advanced math with the LiteVNA, you also would need a PC.   

***
Same for any of these low cost VNAs, not just the LiteVNA. 

***
Link showing LiteVNA HW rev 3.2 + PC + Solver to perform T-Check (after sorting out a few problems).  The only way to get that level of performance was to characterize the cheap standards and enter that data into Solver.   The LiteVNA is the limiting factor, not the standards. 
   
https://www.eevblog.com/forum/rf-microwave/nanovna-custom-software/msg5685217/#msg5685217

Someone had attempted a similar test using the LibreVNA and I think they achieved 3GHz.  Similar the the 4GHz I saw with the LiteVNA.

[joeqsmith]

I put together some quick software to try out that old Weinschel 63dB step attenuator and all of the stages are fine.

Using an IFBW of 500Hz, I stepped from 0 to 60dB, then inserted that 40dB fixed attenuator. 

[joeqsmith]

Showing 80, 90 & 100dB with 4kHz IFBW and 1000 averages.   

[mark432]

Well, it was designed for a certain price.   I am not sure if the hardware has enough resources to support it (code space, processing power...).       

It's not really a "horsepower" limitation as the calibration only needs to be calculated once when you are doing it.
You're solving for a matrix, that you're going to multiply by your measured data matrix, to give the actual result.
There's definitely more code involved in allowing definable cal sets, but it's all math that you only calculate once at cal.
( I wouldn't expect code space to be a problem these days.)

Once it's done, you're doing the same number of calculations every sweep regardless of the cal type.

Bigger point?  Normally I would cal these out.   I normally have cables attached and measure on the other end. 

Yeah, but often at the other end of your cable you might have other sex or other gender connectors than the supplied cal standards.
Like when measuring at the end of a several foot SMA to N cable.  Best case, you're using a decent adapter than only adds half an inch of phase shift and has a 30 dB RL.
Quite likely you're doing worse.

VNAs are about the best example of "garbage in, garbage out" for test equipment.

I don't see anything else better for the price, but I'm leaning towards LibreVNA if I ever want upgrade from my NanoVNA.
It seems to have much closer to a normal VNA functionality. 
The NanoVNA reminds me more of something like an Anritsu Sitemaster than a full VNA.

I would think the best way to cover cal for super low cost VNAs would be to allow known s-parameter cal for the datasets.
The manufacturer can take 1 minute and measure the standards on a traceable VNA and then provide those files with the standards.  No need to try and make them perfect SOLT.  Just use decent repeatable connectors and stable resistors.

Many of us will have access to a lab-grade VNA and it would be easy to measure a handful of parts as home standards.

[joeqsmith]

Well, it was designed for a certain price.   I am not sure if the hardware has enough resources to support it (code space, processing power...).       

It's not really a "horsepower" limitation as the calibration only needs to be calculated once when you are doing it.
You're solving for a matrix, that you're going to multiply by your measured data matrix, to give the actual result.
There's definitely more code involved in allowing definable cal sets, but it's all math that you only calculate once at cal.
( I wouldn't expect code space to be a problem these days.)


Once it's done, you're doing the same number of calculations every sweep regardless of the cal type.

Bigger point?  Normally I would cal these out.   I normally have cables attached and measure on the other end. 

Yeah, but often at the other end of your cable you might have other sex or other gender connectors than the supplied cal standards.
Like when measuring at the end of a several foot SMA to N cable.  Best case, you're using a decent adapter than only adds half an inch of phase shift and has a 30 dB RL.
Quite likely you're doing worse.

VNAs are about the best example of "garbage in, garbage out" for test equipment.

I don't see anything else better for the price, but I'm leaning towards LibreVNA if I ever want upgrade from my NanoVNA.
It seems to have much closer to a normal VNA functionality. 
The NanoVNA reminds me more of something like an Anritsu Sitemaster than a full VNA.

I would think the best way to cover cal for super low cost VNAs would be to allow known s-parameter cal for the datasets.
The manufacturer can take 1 minute and measure the standards on a traceable VNA and then provide those files with the standards.  No need to try and make them perfect SOLT.  Just use decent repeatable connectors and stable resistors.

Many of us will have access to a lab-grade VNA and it would be easy to measure a handful of parts as home standards.

The files I used for the T-Check experiment are 512kB, male only, and limited to 1601 points.  LiteVNA support more points and we would expect to support both M/F.   Trivial for the PC.   

[knudch]

I would think the best way to cover cal for super low cost VNAs would be to allow known s-parameter cal for the datasets.
The manufacturer can take 1 minute and measure the standards on a traceable VNA and then provide those files with the standards.  No need to try and make them perfect SOLT.  Just use decent repeatable connectors and stable resistors.

Many of us will have access to a lab-grade VNA and it would be easy to measure a handful of parts as home standards.

There is a Guy Kurt Poulsen , search for him in the LibreVNA forum, he has written a article about "transfer" all CAL parametre from a lab-grade to home standards in the LibreVNA.

Knud

[Miek]

The LiteVNA64 does allow you to define cal standards, it's under Calibrate -> Calibrate -> Calibration Standard. I checked on v1.3.31 and the latest v.1.3.36

[joeqsmith]

The LiteVNA64 does allow you to define cal standards, it's under Calibrate -> Calibrate -> Calibration Standard. I checked on v1.3.31 and the latest v.1.3.36

Nice find.  Mine is much older (07) and appears to at least have some limited settings.   They do not allow a loss term for any of the standards.  The only standard that allows you to define Z0 is the load.    The load also allows an L & C term which at least METAS does not.   It assumes everything is coaxial. 

mark432 suggested ditching the polynomial for the database.  My friends new H4 and the LiteVNA do have that card reader which would easily hold what ever files I would come up with.  May be an option if there was a need for it.

[theblinkingman]

I ran a T-check on a LibreVNA and it seems to only make it up to 1.25 GHz under 15%, which doesn't seem that great. 

I used the same T-check on a higher quality VNA and got under +/-4% up to 8 GHz. I used the built in calibration procedure on each, so it's possible something went wrong with that on the LibreVNA. 

Also did a S21 measurement with both ports terminated. 

[DiSlord]

Does LiteVNA64 actually allow you to define cal standards?

Yes, LiteVNA allow define calibration standard:
Calibrate->Calibrate->Calibtation standard

Remove Use Ideal flag and define
Open:  Z0, Delay, Offset Loss, C0, C1, C2, C3
Short:  Z0, Delay, Offset Loss, L0, L1, L2, L3
Load: R, Z0, Delay, offset Loss, L, C
Thru: Z0, Delay, Offset Loss

You can change values and in real time see changes

Additional possible save and load standard settings to/from sd card.

PS calibration standard settings equal as on LibreVNA / NanoVNA Saver (NanoVNA saver not use Offset loss settings)
PSS  my tests cheap China calibration kit, define calibration standard settings and improve it
https://groups.io/g/nanovna-beta-test/message/4538

[virtualparticles]

The TR1300 has a dynamic range of 130 dB. You won't find that kind of performance with a low-cost box.

https://coppermountaintech.com/vna/tr1300-1-2-port-1-3-ghz-analyzer/

Best of Luck!

Doesn't quite meet the mark of the topic of the post though does it?

Doesn't meet the price target, but it's an actual VNA with metrology. I wouldn't spend $2k for a random number generator myself.

Great point.

What about the CMT single port VNA's for the price point. I sometimes see those on eBay pretty cheap.
Edit: My bad, I thought this thread was about testing antennas, clearly the OP would need two ports.

Also, Brian, please (for the love of god) speak to your web dev about the menus on the CMT site.
The menu pop's up on mouse over and then often disappear when the mouse is moved to select something in the menu. So the site is impossible to use at times.
Very frustrating.
Keep it simple: Show the menu on click, and hide the menu when an item is clicked, or there is a click elsewhere on the page.

I passed this comment along to our fine webdev people. Feedback appreciated. Full transparency, added a disclaimer to my signature as someone suggested as well. 

[hendorog]

The TR1300 has a dynamic range of 130 dB. You won't find that kind of performance with a low-cost box.

https://coppermountaintech.com/vna/tr1300-1-2-port-1-3-ghz-analyzer/

Best of Luck!

Doesn't quite meet the mark of the topic of the post though does it?

Doesn't meet the price target, but it's an actual VNA with metrology. I wouldn't spend $2k for a random number generator myself.

Great point.

What about the CMT single port VNA's for the price point. I sometimes see those on eBay pretty cheap.
Edit: My bad, I thought this thread was about testing antennas, clearly the OP would need two ports.

Also, Brian, please (for the love of god) speak to your web dev about the menus on the CMT site.
The menu pop's up on mouse over and then often disappear when the mouse is moved to select something in the menu. So the site is impossible to use at times.
Very frustrating.
Keep it simple: Show the menu on click, and hide the menu when an item is clicked, or there is a click elsewhere on the page.

I passed this comment along to our fine webdev people. Feedback appreciated. Full transparency, added a disclaimer to my signature as someone suggested as well. 

Great stuff - they will work it out but the problem is that there is a gap between the menu header and the drop down.
There must be an event which is hiding the menu when the mouse is moved outside the "menu area"

So, if the mouse is moved quickly enough over the gap, then the browser doesn't see a mouse event in the gap and no issue.
Otherwise it does and the menu is closed. Changing the "top" attribute of the ul.sub-menu class from 52px to 45px fixes it for me.

First image shows the gap, second shows the change, third shows the ul.sub-menu object being changed.

[TELCO]

Hello. I own a Copper Mountain Technologies TR1300/1 2-Port 1.3 GHz Vector Network Analyzer for our filter (duplexer) tuning needs, antenna characterization and any other passive RF testing need we may have. When we decided to make an investment in this equipment, I did the research based on the requirements needed and the major concerns were that of dynamic range, the usual application for professional duplexer tuning is around 80 dB to 100 dB. Simply put a “low cost” VNA does not have the sufficient dynamic range required for this to be accomplished, these are better suited for hobby use. I would also stay away from the legacy big box VNA because that will tie you to something with no technical support whatsoever. I looked at the Siglent and RIGOL options also and was not 100% sold either. CMT now has available the new VTR0102 2-Port 1.5 GHz Vector Network Analyzer for $2,995 with 120 dB typical dynamic range, so if you can “pony up” a bit on your original 2k budget you could buy a truly professional VNA that you will use for years to come. This is just my honest personal opinion and I am in no way related to Copper Mountain Technologies, their people and tech support are of the highest quality. I am just another satisfied customer. Hope this helps you in your search. 

[pdenisowski]

I did the research based on the requirements needed and the major concerns were that of dynamic range, the usual application for professional duplexer tuning is around 80 dB to 100 dB.  Simply put a “low cost” VNA does not have the sufficient dynamic range required for this to be accomplished, these are better suited for hobby use.

Yes.  I also did a video on this topic where I mention the need for high dynamic range

I would also stay away from the legacy big box VNA because that will tie you to something with no technical support whatsoever.

 

Can't speak to the other "legacy big box VNA" manufacturers, but we (Rohde & Schwarz) have hundreds of applications engineers that support our products worldwide.  Trust me, I used to be one of them   We even make videos showing how to test duplexers with our VNAs



Granted, most of our VNA's aren't in the price range you specified, but -- for the record -- to say that "legacy big box VNA" manufacturers have "no technical support whatsoever" is definitely not accurate, at least in the case of Rohde & Schwarz (who, incidentally, invented the modern VNA 75 years ago).

[KungFuJosh]

modern VNA 75 years ago

That's hilarious. 

[pdenisowski]

modern VNA 75 years ago

That's hilarious. 

Here's a picture of the unit itself - the Z-g Diagraph.  I got to play with this one during a trip to our Munich headquarters last year.  It's bigger and heavier than it looks 

[KE5FX]

Can't speak to the other "legacy big box VNA" manufacturers, but we (Rohde & Schwarz) have hundreds of applications engineers that support our products worldwide.  Trust me, I used to be one of them   We even make videos showing how to test duplexers with our VNAs

How long before all those videos are yanked off of YouTube and put behind a login wall, requiring credentials that might or might not be granted depending on whether I can convince R&S that I'm a major corporate customer?

I know that's not your call, and I don't mean to imply that you'd do anything like that if it were.  But in this day and age it's absolutely a consideration when it comes to choosing a T&M vendor, for those of us who operate with five- and six-digit budgets rather than seven or eight.

[pdenisowski]

Can't speak to the other "legacy big box VNA" manufacturers, but we (Rohde & Schwarz) have hundreds of applications engineers that support our products worldwide.  Trust me, I used to be one of them   We even make videos showing how to test duplexers with our VNAs

How long before all those videos are yanked off of YouTube and put behind a login wall, requiring credentials that might or might not be granted depending on whether I can convince R&S that I'm a major corporate customer?

I can assure you that's not going to happen   Did I mention that my 300th video went on our YouTube channel this morning? 

I know that's not your call, and I don't mean to imply that you'd do anything like that if it were.  But in this day and age it's absolutely a consideration when it comes to choosing a T&M vendor, for those of us who operate with five- and six-digit budgets rather than seven or eight.

No, it's definitely not my call and it's also not my place to comment on corporate policy (I'm here as a hobbyist who happens to work for R&S, not as an R&S representative).  That said, we don't differentiate business customers by budgets - I've gotten on an airplane many, many times to go help a customer who was operating with a "five-digit budget". 

And unlike some of our competitors (who also differentiate between business and non-business customers, BTW), we don't charge for support, software updates, etc.   
 
73! Paul, KO4LZ

[kmoonwalker]

modern VNA 75 years ago

That's hilarious. 

Here's a picture of the unit itself - the Z-g Diagraph.  I got to play with this one during a trip to our Munich headquarters last year.  It's bigger and heavier than it looks 

My precious ^^