What VNA to get?

[Mechatrommer]

Application contexts would include PCB-level RF design

If you set the source as low as it can go (-10dBm) and them put a 15dB gain LNA inline for test it will overload the VNA receiver as the receiver max level appears to be 0dBm. So it is not really suitable for full s2p tests on

use attenuator (such as 6 - 20dB attenuator) on the output of LNA. recalibrate beforehand with the attenuator attached.

[hendorog]

I know you don't like PC instruments, but...

If it was my money, I'd squeeze the boss a bit more and get one of the Copper Mountain VNA's, an eCal kit and the new SignalHound 20GHz SM200A.
That is a heck of a lot of capability for your money.

Or get one of the bench VNA's mentioned, a cal kit, and a Siglent SSA3000X series Spectrum Analyser with the change.

[G0HZU]

Application contexts would include PCB-level RF design

If you set the source as low as it can go (-10dBm) and them put a 15dB gain LNA inline for test it will overload the VNA receiver as the receiver max level appears to be 0dBm. So it is not really suitable for full s2p tests on

use attenuator (such as 6 - 20dB attenuator) on the output of LNA. recalibrate beforehand with the attenuator attached.

The problem with adding attenuators inline is that it degrades the quality of reflection measurements when making full s2p measurements after a full SOLT calibration (with the attenuators in place). So including attenuators like this is a bit of a fudge at best. It gets worse if you try and do something as mundane as trying to measure a classic s2p model of a simple BJT in common emitter. The -10dBm drive level from the VNA will be too big by maybe 20dB and will cause distortion effects in a typical small signal BJT biased at a few mA.

The other limitation of the ZNLE-6 is the linearity of the receiver. It appears to have a spec of 0.3dB (compression?) between input levels of -5 to 0dBm as the max input level of 0dBm is approached. The typical spec here is 0.2dB which is slightly better but all this points to a VNA that is really meant to be used for basic testing of passive parts with the drive level set at -10dBm.

It will be OK for some users but most RF designers will want something more versatile. I couldn't live with it.

[Mechatrommer]

Application contexts would include PCB-level RF design

If you set the source as low as it can go (-10dBm) and them put a 15dB gain LNA inline for test it will overload the VNA receiver as the receiver max level appears to be 0dBm. So it is not really suitable for full s2p tests on

use attenuator (such as 6 - 20dB attenuator) on the output of LNA. recalibrate beforehand with the attenuator attached.

...The -10dBm drive level from the VNA will be too big by maybe 20dB and will cause distortion effects in a typical small signal BJT biased at a few mA.

use 20dB attenuator on the input of small signal bjt. recalibrate beforehand with the attenuator attached. i've done this (attenuators at both input and output) to get a descent S21 measurement, and as workaround of VNA IO impedance mismatches... well... i'm not arguing your points i'm just suggesting workarounds. you maybe right, if OP can spend a little bit more, he can find better VNA. ymmv.

[Omicron]

If you're doing amplifier design then the ZNL is probably the one to buy if shopping for an entry level R&S instrument. The ZNLE is a ZNL without any of the options and without the possibility to install any after purchase. The ZNL offers an option to control the source down to -40dBm along with various other things.

For myself the ZNLE is pretty much the ideal device. I don't do transistor level RF design. My uses for a VNA are antenna matching for various wireless modules (ISM, Bluetooth, ...), basic evaluation of antenna gain and some PDN related measurements for evaluating decoupling performance. Since RF is not my main focus I need something that is easy to use and works reliably without much fuss. When I do dabble with active circuits it's more for self education than professional use and the workarounds for the source limitations are acceptable.

The PicoVNA works well for me but I hate the user interface. The ZNLE is a stretch in terms of price but in comparison it is a pleasure to use.

[Mechatrommer]

Rohde & Schwarz ZNLE6 2 Port 6GHz Vector Network Analyzer $17,720.00

Rohde & Schwarz ZNL6 2 Port 6GHz Vector Network Analyzer $25,565.00
for additional $8000, you get few more dB dynamic range, and... more attenuation on the output (add another $2,470 for Rohde & Schwarz ZNL6-B22.02 Extended Power Range for ZNL6), maximum input port is still 0dB. ymmv..

[G0HZU]

In a business environment it's worth the extra money to get the control of the source power and to have the ability to fit options to it. It looks like the ZNLE-6 is deliberately crippled and that was the impression I got from reading the specs. It will be OK for some users but it will never be a versatile VNA unless there is a way to 'uncripple' it.

When I do dabble with active circuits it's more for self education than professional use and the workarounds for the source limitations are acceptable.

OK, but I can only offer advice in general. Assuming the instrument would be used for semi serious RF design the 'workarounds' suggested by Mechatrommer just won't deliver good performance. The suggestion to use a 20dB attenuator inline with port 1 when doing a full 2 port calibration isn't realistic. You can get away with a 10dB attenuator if noisy/degraded data is acceptable but how is the VNA going to do a valid SOLT calibration when looking through a 20dB attenuator? The unterminated return loss of a typical 20dB attenuator (with a short or open) could be similar or better than the reference 50R load at some frequencies in the GHz region. How is the VNA going to cope with that?

You ideally want to be able to measure S11 S12 S21 and S22 in one go (after a valid calibration) and then generate a valid s2p (S11 S12 S21 S22) data model of the device under test and that is the main function of a full 2 port VNA.

[PnpNpn]

Dear G0HZU,

let me ask you, I have some technical questions, write me private messages.
I do not want to clog a topic. Thanks!

In a business environment it's worth the extra money to get the control of the source power and to have the ability to fit options to it. It looks like the ZNLE-6 is deliberately crippled and that was the impression I got from reading the specs. It will be OK for some users but it will never be a versatile VNA unless there is a way to 'uncripple' it.

[G0HZU]

Hi PnpNpn,
I don't wish to appear rude but I turned my PM feature off a long time ago. Sorry about this...

Besides there is no 'one size fits all' answer to this thread in case anyone doesn't agree with my view. For some people the ZNLE-6 will be a good choice. Others will want the more compact Pico VNA or a Copper Mountain VNA etc etc.

I chipped in to help people make a more informed choice because for users like me, the ZNLE-6 is a non starter because of the limitations it has with the source power and the linearity of the receiver. These limitations limit it mainly to the testing of passive networks in my opinion.

[Omicron]

The other limitation of the ZNLE-6 is the linearity of the receiver. It appears to have a spec of 0.3dB (compression?) between input levels of -5 to 0dBm as the max input level of 0dBm is approached. The typical spec here is 0.2dB which is slightly better but all this points to a VNA that is really meant to be used for basic testing of passive parts with the drive level set at -10dBm.

Where did you find that number? I see one mention of 0.3dBm in the section "Test Port Input" but it refers to +10dB to +5dB input level, not -5dBm to 0dBm. Am I misinterpreting something?

[G0HZU]

The +5dB and +10dB are power ratios referenced to a source power of -10dBm so that equates to input power levels at the receiver of -5dBm and 0dBm.

[Omicron]

Ah, I see!

To get a sense of what these design compromises in the ZNLE mean in the real world, I made a measurement of a Crystek amplifier I had laying around. The data-sheet for the device can be found here:

http://www.crystek.com/microwave/spec-sheets/amplifier/CRBAMP-100-6000.pdf

I used a 20dB attenuator in the source path (port 1) and for the rest pretty much the default settings of the instrument. I would say that certainly below about 4GHz there is good agreement with the graphs in the data-sheet.

[G0HZU]

You've selected a fairly meaty amplifier there that can cope fairly well with -10dBm drive at its ports. Try doing it with something like a low power, high gain LNA running at 3V biased at a couple of mA.

A really simple and classic 'active' task would be to measure a small signal s2p model of a jellybean BJT transistor biased at a few mA. i.e. produce a data file for S11 S12 S21 S22. A VNA designed for active circuits can spit out a small signal s2p file in a few seconds with the source power set down low enough to preserve linearity in the DUT and the VNA. I don't see how you can easily do this with the ZNLE-6 even if you do it in stages with attenuators inline for some measurements.

Note that even R&S state in their brochure that the ZNLE-6 is a VNA designed to be able to make bidirectional measurements on passive components. The specs scream out that this VNA is crippled in this respect.

Can it be hacked to 'decripple' it?

[Mechatrommer]

A really simple and classic 'active' task would be to measure a small signal s2p model of a jellybean BJT transistor biased at a few mA. i.e. produce a data file for S11 S12 S21 S22. A VNA designed for active circuits can spit out a small signal s2p file in a few seconds with the source power set down low enough to preserve linearity in the DUT and the VNA.

any recommendation on VNA model that can do the job?

[KE5FX]

A really simple and classic 'active' task would be to measure a small signal s2p model of a jellybean BJT transistor biased at a few mA. i.e. produce a data file for S11 S12 S21 S22. A VNA designed for active circuits can spit out a small signal s2p file in a few seconds with the source power set down low enough to preserve linearity in the DUT and the VNA.

any recommendation on VNA model that can do the job?

What I'm seeing in this thread is a lot of people talking about $30K instruments that can't do what a 20-year-old HP 8753C can do at 1/10 the price.

So I'd say that's a pretty strong hint, especially for those who prefer non-PC-based instrumentation.  HP VNAs are what all those other companies wish they were building.

[JohnG]

The PicoVNA certainly bears a strong resemblance to this: http://latechniques.net/wp-content/uploads/Dw97049_Brochure_LA19-13-03.pdf.

Is it's software really that bad?

Thanks,
John

[G0HZU]

Yes, the full 2 port 8753 VNAs are fabulous instruments, even today. We still have quite a few 8753ES models at work in the main design labs even though they must be about 18 years old. If I wanted a general purpose 6GHz VNA and also wanted best bang for buck I'd go for the older HP8753D model. The prices for these are tumbling and I'd much rather have a £1500- £2000 HP8753D than the ZNLE-6.

But the HP8753C can be had for even less money.  However, the OP did seem to want to spend a lot of money...  $25k?

[Mechatrommer]

I wouldnt dare investing big money on obsolete model unless there is strong assurance it will run ok for a long time. But i dont mind buying it at small money with slight damage. If i can fix it i saved alot of money, if not, i lose small amount of money. But that is me, ymmv. I think mathematical model of current technology is good enough to 'de embed' system's respond or characteristic. So when attenuators are included in calibration, they are considered in-system elements and de -embedded from final dut result, hence not necessitating of low signal vna output and high input signal level for the purpose mentioned. I read briefly on this 'de embedding' in app notes, agilent/keysight iirc, but not really 100% sure understand since i have no intention at designing a vna, and the math is a bit daunting. Ymmv.

[KE5FX]

Yes, the full 2 port 8753 VNAs are fabulous instruments, even today. We still have quite a few 8753ES models at work in the main design labs even though they must be about 18 years old. If I wanted a general purpose 6GHz VNA and also wanted best bang for buck I'd go for the older HP8753D model. The prices for these are tumbling and I'd much rather have a £1500- £2000 HP8753D than the ZNLE-6.

But the HP8753C can be had for even less money.  However, the OP did seem to want to spend a lot of money...  $25k?

I would be only too happy to sell him mine for that. 

[Omicron]

The ZNLE is not a 30K instrument. Not even half that price with the discount R&S gives you (at least here in Europe). 20 year old second hand gear might be great for hobby but for professional use it is simply not an option. At least not in my shop. I want equipment that has support and warranty.

I don’t dispute that the ZNLE isn’t meant to measure active devices, R&S is pretty clear about that in their documentation. I just wanted to provide a feeling of what that limitation means. I don’t do transistor level design so I can’t easile do a sample measurement in that respect unless someone is willing to send me a DUT.

Don’t discount practical issues either. When I do antenna measurements I take the entire setup outside since I do not have an anechoic chamber. With an instrument like the ZNLE that is trivial, you can carry it in one hand. Try that a few times with a classic HP and let us know how much fun that is :-)

[palpurul]

How about second-hand equipements? Keysight's ebay store has some good VNAs. Most of them are pretty old have still floppy disk drivers, but this one looks pretty modern.

[hagster]

I agree, The software looks identical and LA Technologies sells an OEM module.

Looking in more detail at this PicoVNA specs and features, it does seem to offer plenty of bang for your buck. There is some useful discussion of this device here https://www.eevblog.com/forum/testgear/picovna/25/ It sounds like the software really is that bad though (VB6 in 2018), but like the Copper Mountain it has a demo mode.

The PicoVNA certainly bears a strong resemblance to this: http://latechniques.net/wp-content/uploads/Dw97049_Brochure_LA19-13-03.pdf.

Is it's software really that bad?

Thanks,
John

[Omicron]

The PicoVNA certainly bears a strong resemblance to this: http://latechniques.net/wp-content/uploads/Dw97049_Brochure_LA19-13-03.pdf.

Is it's software really that bad?

Thanks,
John

Looks like the same unit and software, although it seems to have more of the intermediate connections available on the front panel offering some additional flexibility.

You can download the software from the Pico website and try it out for yourself. You can certainly make do with it but the product could have been so much better if they had invested a bit more in the software.

[PnpNpn]

Hi G0HZU,
how can I contact you and hear some tips.

Hi PnpNpn,
I don't wish to appear rude but I turned my PM feature off a long time ago. Sorry about this...

[G0HZU]

The ZNLE is not a 30K instrument. Not even half that price with the discount R&S gives you (at least here in Europe). 20 year old second hand gear might be great for hobby but for professional use it is simply not an option.

I had a look at one of our HP8753ES VNAs at work and it has a 1999 date code. I still feel a bit of a tingle when I use one of these old VNAs because I know they are so special. I think we have three or four of them in the main design labs and I think we have owned them all from new. They aren't as popular these days because of the grainy and gloomy VGA display and the lack of modern I/O but for users like me this doesn't really matter as I mainly use a VNA as a means to grab 'n port' data models to export to a PC.

One thing to watch out for with the ENA E507x series of VNAs from Agilent is that they generally don't have bias tee ports at the back of the instrument. We have a few of them in the ATE/production area and I don't think any of them have bias tee connections. I think it's an option and this tends to push up the used prices of something like an E5071C that has the bias tee option. They do seem to hold their prices really well if they have the desirable options.