Wanted open-source code for dielectric measurements

[drkirkby]

There are a number of methods of measuring the dielectric properties (both permittivity and loss) of materials with a vector network analyzer. These include

* An open-ended coaxial probe.
* Measure the resonate frequency of a cavity, put the material in, then look at the change in resonate frequency, and Q of the cavity to find the complex permittivity.
* Free space measurement, putting the material between two antennas.
* Inserting the material in a waveguide or coaxial cable.
* Make an air-spaced capacitor, then replace it with the dielectric.

Making up the hardware for any of these methods does not require a big effort, but converting VNA measurements to dielectric properties does.

The Keysight (formally Agilent) 85071E Materials Measurement Software does some/all of these methods, but starts at over $7000

https://www.keysight.com/en/pd-2174...nt-software?nid=-32987.536883268&cc=US&lc=eng

with options to add features like resonate cavity method (more than $1000). The software is not only expensive, but will assume you have the particular Keysight parts.

There are tons of papers on these methods in IEEE journals and similar, but I have not seen any software for it. The maths is usually impenetrable for me anyway.

So I am just wondering if anyone is aware of any software on these methods, or even a paper that does not require a degree in maths to understand so I can write the software myself.

Dave

 

[volker@muehlhaus]

Hi Dave,

I guess this isn't cheaper than the Agilent/Keysight solution, but wanted to mention that SPEA showed a similar product at EuMW Rome:
https://www.speag.com/products/dak/dielectric-measurements/

Best regards
Volker

 

[drkirkby]

Hi Dave,

I guess this isn't cheaper than the Agilent/Keysight solution, but wanted to mention that SPEA showed a similar product at EuMW Rome:
https://www.speag.com/products/dak/dielectric-measurements/

Best regards
Volker

Thank you Volker,
I must admit I was not aware of that, and it does support my HP 8720D VNA, but not my more modern Agilent PNA, as the PNA wont take new enough firmware! But cost is going to be the issue.

I don't blame Keysight keeping their source closed, as they have invested money in the R&D. But there must be numerous government bodies that have funded research in universities doing this sort of thing. So both you and I have probably paid for this research. It is about time the grant giving bodies made more of an effort to only fund work where the authors agree to make source code available. I believe there are some grant giving bodies that do this, but I see a research report where the author contacted the grant holders to get the source code for work that had been developed under such contracts. Basically he got either

* Refusal
* Ignored
* A small percentage did share the code.

When I was a young lad at school (35-40 years ago), we were told that experiments should be written in a log book in a way that others could repeat them. Basically for a large fraction of research published today, that is just not possible.

Dave

 

[tony_lth]

If you have paper for that, then using MATLAB is very convenient.

 

[FvM]

I believe there are test setups that expose a simple relation between dielectric properties and measured s-parameters, so can you calculate it applying text book formulas.

 

[volker@muehlhaus]

The guys from SPEAG told me that they used their 3D EM simulators to develop the evaluation method for their probes (post #2). This way, they can include field distribution in the dielectric media in a more accurate way than textbook equations.

 

[drkirkby]

I believe there are test setups that expose a simple relation between dielectric properties and measured s-parameters, so can you calculate it applying text book formulas.

Yes, there are. One I suspect is particularly easy for that is to insert a piece of the material across the cross section of a waveguide. But there are a number of different techniques, and all have their relative advantages and disadvantages. Some work for solids, some for liquids. Some work well for high loss materials, others for low loss materials. Some, like the free space measurement require very large samples. Some like the parallel plate capacitor technique only work at low frequencies.

There is non one method that is suitable in all cases.

Perhaps a 3D EM simulator is the only way to go for some setups.

Dave

 

[drkirkby]

The guys from SPEAG told me that they used their 3D EM simulators to develop the evaluation method for their probes (post #2). This way, they can include field distribution in the dielectric media in a more accurate way than textbook equations.

Hi,
I thought there was an option on this site to give a "thumbs up" for helpful posts. I'd like to do it for your post, but can't seem to.

Anyway, what you said there is interesting & helpful. I assume once you have used the EM simulator to determine S11 from the material properties Er' and Er'', some sort of "lookup table" or empirical equation will allow one to determine Er' and Er'' from the S11 measurements.

The next problem I would have is how does one generate an empirical equation based on the data? I guess that is a job for a mathematician to answer. One might reasonably assume

* Phase of S11 is 0 at low frequency for a coaxial probe.
* Phase of S11 increases with frequency (for simplicity ignoring wrap around at 180 degrees)
* Phase of S11 increases with permittivity (for simplicity ignoring wrap around at 180 degrees)
* The amplitude of S11 decreases for increasing loss of the material.

All based on a fixed size of probe.

Oh well, it will give me something to think about!!

Dave

 

[FvM]

Don't you see this line at the bottom of each post?

 

[volker@muehlhaus]

The next problem I would have is how does one generate an empirical equation based on the data?

Hi Dave,

here's a software that I found useful for curve fitting in similar cases:
https://www.curveexpert.net/products/curveexpert-professional/

Best regards
Volker

 

[drkirkby]

Don't you see this line at the bottom of each post?

I do now, and have seen it in the past. I just could not find it when I wanted it!

Anyway, I have hit the "Helped Me" for Volker's post, as it was genuinely useful.

 

[biff44]

i fully agree dave. While time goes on, some basic microwave problems, like converting S11 measurements into dielectric constant, never seem to show up on shareware! I too would love to see someone show, in simple language, how to do it. Sure, we can all download papers, research it, and throw it on matlab/excell and probably after a few weeks of effort find the answer ourseves.

BUT, why has someone not published that excersise for us to say "BRAVO!' to??

 

[drkirkby]

i fully agree dave. While time goes on, some basic microwave problems, like converting S11 measurements into dielectric constant, never seem to show up on shareware! I too would love to see someone show, in simple language, how to do it. Sure, we can all download papers, research it, and throw it on matlab/excell and probably after a few weeks of effort find the answer ourseves.

BUT, why has someone not published that excersise for us to say "BRAVO!' to??

Don't confuse shareware with open-source. In shareware, the source is kept closed, so you never understand what is happening. Normally one has to pay to keep the software working beyond a trial period, and/or get the full functionality. It is not much better than buying stuff from Agilent, although shareware prices have tended to be a lot less than anything Agilent would produce.

With open-source, one can actually see what is being done, which makes it far more useful.

Dave