Do I need a VNA for this RF design?

[Valve]

I'm the CEO of a small instrumentation company and would like to get some advice on equipment expenditure. We're currently working on the development of a materials analyser that operates in the 1.5 - 3.0 GHz range, with the objective of producing and selling an industrial grade product. It will do a frequency sweep to determine the material's transmission characteristics (attenuation and group delay) in samples up to 1 m thick. Our final product will be a custom made and packaged hardware design that incorporates an RF front end (TX and RX for the test signal over 1.5 - 3 GHz) and a microprocessor to analyse and deliver up the final data on the material.

NOW, the R&D equipment required! My research engineers say we need a very expensive 3 to 6 GHz VNA to do this work. I say we don't. I've told them they can design, build, test and manufacture the product with our existing equipment (i.e. multimeters, oscilloscopes etc.).

Please anyone, let me have your opinion on this. I get the feeling my engineers are just looking to blast company money on an expensive toy. Am I right? Or should I spend the money? What would be the consequences of not buying the VNA (if any)? Comments on this would be most welcome.

 

[thylacine1975]

You should certainly spend the money to buy one - indeed, what you are proposing to build is essentially a VNA anyway!

Buying one will a) let you build your own system more expediently, and b) provide a truth against which to compare your own measurements.

I'd actually suggest a reason c) though... forget about building your own hardware altogether - buy a (cheap) USB/headless VNA and wrap your signal processing/materials identification algorithm around it. Let someone else do the hard (read: expensive development) work of low phase noise, wideband sweep oscillators and high dynamic range magnitude/phase measurements and you put the effort into the (low capital expenditure) signal processing.

 

[SLK001]

Although I am at a loss to know why your engineers need a 3 to 6 GHz VNA, when their frequency range is 1.5 to 3 GHz, I am appalled that you would question the very people you need to produce a working product. What is your background? Do you have the expertise to question what your engineers say is a requirement? If the equipment that you do have is in line with your "mulimeters, oscilloscopes, etc.", then your company DOES NOT HAVE THE EXPERTISE TO EVEN BEGIN THIS PROJECT.

RF work is difficult and the equipment needed to adequately perform it is very expensive. It is also one of the most difficult curriculums in the electrical engineering field. If you have people qualified to make equipment selections, then LISTEN TO THEM!!! No one here has any idea of the particular needs of the project, given your brief synopsis. When I was doing RF amplifier design, I had $500,000US worth of equipment on my desk and every ounce of it was needed. So, given my statement at the beginning of this posting, stop cheaping out on the equipment requirements and get what your people need.

 

[filip.amator]

Imagine that you want to make a high precision tool used to measure distances - a RULER. And your R&D people want you to buy a precision interferometer to calibrate your rulers because they need to compare product with definition of the metre. What do you do?

And BTW you can hire an expensive lab equipment.

 

[BigBoss]

If the VNA will be used for a short period, rent it.Don't invest for nothing..

 

[G4BCH]

>> What would be the consequences of not buying the VNA (if any)?

You will probably waste many times more than the cost of either buying or hiring the instrument trying to do it without.

I've been on the receiving end comments like this a number of times and had to fight hard for the kit or software that has been needed. On one occasion it needed upsetting our finance director by asking him whether the latest version of a spreadsheet was necessary for him to do his job.

I concur Concur with all the comments, if you don't have the expertise and it sounds like you don't, buy a relatively cheap USB instrument and use that. Depending on the performance needed something like the AD ADLMPLUTO may be able to do the job, but that still won't get you over having to verify and calibrate your instrument.

 

[biff44]

You can NOT use meters and oscilloscopes to measure group delay and attenuation...at least to even the grossest of accuracy.
You need a vector network analyzer.
You CAN design your own network analyzer, and the cost might be very affordable if your requirements are not too bad. I once built a 5.8 Ghz VNA for a consumer project for about $30 in productoin...but it has some very specific specifications that allowed me to get away with such a low cost.

If i were you, i would RENT a good quality two port 6 GHz VNA, make all sorts of measurements of various materials with it, and then decide on exactly what ACCURACY and REPEATABILITY you need. Also pay attention to how critical some frequency drift would be to your results, as you might just use a free running source. Once you know what parameters you need to achieve the deisred materials characterization, THEN you can decide the most cost effective way to go about making the measurement hardware.

another factor is if the materials are very lossy...then you will need a very high dynamic range in the VNA's receiver circuits.

 

[volker@muehlhaus]

<*-¦-¦-¦-¦-¦-><

 

[vfone]

My research engineers say we need a very expensive 3 to 6 GHz VNA to do this work. I say we don't. I've told them they can design, build, test and manufacture the product with our existing equipment (i.e. multimeters, oscilloscopes etc.).

Sorry no offense, but according to above statement I am pretty sure you are what you mentioned. The CEO of the company.
To replace a 6 GHz VNA (bought or hired) with multimeters and oscilloscopes, doesn't make sense to continue this discussion.

 

[FvM]

Before VNA were generally available, RF engineers didn't specifically use oscilloscopes and multimeters, but e.g. slotted lines to measure network parameters, see: https://en.wikipedia.org/wiki/Slotted_line

I fear, few engineers have the skills today to master these traditional measurement methods. Work also proceeds rather slowly with it. Time is money, isn't it?

 

[vfone]

Also years ago people do not have matches, and they used to lit the fire using flint.
I am sure that Valve was not looking for those old and rusted guys that you think about

 

[Valve]

Thanks guys, for all your comments. Much appreciated. Looks like I'll have to reconsider my original decision.

 

[Azulykit]

A VNA is a basic measurement tool for microwave measurements. Even at your relatively low frequency the equipment will be costly. The microwave game is expensive and you will also need a host of additional accessories. Included are calibration standards, interconnections and a torque wrench to make repeatable connections. You probably need to make sure you have a microwave mechanic and technician on your staff. It is not all black magic but it does need someone skilled in the art.

You get what you pay for. A VNA in the $20 K range will be just barely above junk. $100 K is likely a credible instrument and the really good stuff is significantly higher. A set of (GOOD) calibration standards is essential and around $10K. Without a calibration kit a VNA is dumb and using old and questionable pedigree standards is foolish. You will probably end up spending about the same on the collateral support items that you spend on the VNA. That is something you probably do not want to hear. By the way if a connector, adapter, standard or the like is damaged throw it away IMMEDIATELY. Using broken items corrupts your measurements and they can also destroy otherwise working parts.

Renting equipment may be an option and there is the used market to consider.

As a general rule the development tools need to be about 10X better than the result you are trying to measure to be useful. I would avoid equipment that is no longer supported by the particular manufacturer in an effort to minimize expense. That would be a false saving particularly as you are looking to produce an "industrial" grade product.

High end microwave simulators (software) will also be helpful. While also very expensive you can significantly accelerate your development cycle. About 4x to 6x faster than developing with hardware alone is a rule of thumb. I know, the good software is too expensive and someone has to be able to run it as well.

Spend some time on microwaves101.com to get more insight about what is going on and how to make a more informed decision.

From your question, it is apparent that you need to quickly become familiar with the microwave arena. It is an expensive game you are contemplating.



Regards,

Azulykit

 

[FvM]

I have to contradict the previous statement. For the intended 1.5 to 3 GHz frequency range, a used instrument at a few k$ or an even more economic modern USB VNA as mentioned in post #2 plus accessory for a few 100 $ can do the job. A lot depends of course on the required accuracy, many measurements don't need high S parameter precision.

 

[Azulykit]

Good.

Fast.

Cheap.



Pick any two.





Regards,

Azulykit

 

[drkirkby]

Although I am at a loss to know why your engineers need a 3 to 6 GHz VNA, when their frequency range is 1.5 to 3 GHz, I am appalled that you would question the very people you need to produce a working product. What is your background? Do you have the expertise to question what your engineers say is a requirement? If the equipment that you do have is in line with your "mulimeters, oscilloscopes, etc.", then your company DOES NOT HAVE THE EXPERTISE TO EVEN BEGIN THIS PROJECT.

As

1. An RF engineer, who specializes in the design of VNA cal kits (https://www.kirkbymicrowave.co.uk/)
2. A company director

I can see why a CEO would question what is required, and what is not. That seems good business sense to me. The CEO should also budget for a cal kit and phase stable cables, not just the VNA.

Personally I have seen a few of these VNAs that have no front end, but USB connections to a PC, and personally they would not be something I would want. I know machines from Copper Mountain in particular have excellent performance, but the whole idea of such an instrument does not attract me. I have one here (admittedly the low cost VNWA), and have seen more expensive ones, but I'd rather a used 8753D or 8753ES any day.

I would question how much information can be gained about the attenuation and group delay of a sample 1 mm thick using 1.5 to 3 GHz. Unless the material has particularly high permittivity, the phase difference is going to be pretty small. The free space wavelength at 3 GHz is 100 mm, so phase change is only going to be a few degrees. It's not immediately clear the best way to couple to such a sample either.

I'm not saying its impossible to get the information required, but it might not be as easy as one might initially think. I'm guessing the application is proprietry, so I don't suppose the original poster is going to say too much, but I think there are a few challenges there.

But yes, a VNA seems like it would be pretty essential!

 

[volker@muehlhaus]

I would question how much information can be gained about the attenuation and group delay of a sample 1 mm thick using 1.5 to 3 GHz. Unless the material has particularly high permittivity, the phase difference is going to be pretty small. The free space wavelength at 3 GHz is 100 mm, so phase change is only going to be a few degrees.

He wrote "up to 1m thick" and that should give large effect :bsdetector:

 

[SLK001]

I can see why a CEO would question what is required, and what is not. That seems good business sense to me. The CEO should also budget for a cal kit and phase stable cables, not just the VNA.

This is very true and this is probably what happened and I don't fault the CEO for asking these types of questions. However, when told of the requirements, this CEO, who apparently doesn't have the RF background that you or I do, decided that these "things" were too expensive and what was needed to be done could easily be done with multimeters and oscilloscopes. (I guess that this company has slotted lines lying around to use in such a requirement.)

As an engineer with an advanced technical degree (MSEE) and with an advanced business degree (MBA), I have faced this kind of foolishness throughout my career. I have even had to explain to CEOs how to expense the purchase of expensive equipment by financing the purchase, expensing the interest and depreciating the cost over 5 to 7 years. Or, renting the equipment and expensing everything. I always preferred the former, because the CEOs always wanted to have "more capabilities" to bid on contracts and I explained that without an inventory of advanced equipment at their disposal, many companies posting contracts wouldn't even consider a company "who only had multimeters and oscilloscopes" in their inventory.

I would question how much information can be gained about the attenuation and group delay of a sample 1 mm thick using 1.5 to 3 GHz. Unless the material has particularly high permittivity, the phase difference is going to be pretty small. The free space wavelength at 3 GHz is 100 mm, so phase change is only going to be a few degrees. It's not immediately clear the best way to couple to such a sample either.

The test subjects are up to a meter thick, so measuring the phase change shouldn't be impossible and measuring phase angles almost requires a VNA, or at least some type of vector measuring device.

But yes, a VNA seems like it would be pretty essential!

My point exactly.

 

[drkirkby]

He wrote "up to 1m thick" and that should give large effect

Oops, I mis-read it, and thought it was 1 mm. Only a factor of 1000 out!