Self-made dielectric constant measurement device (rf laminte Er). Please comment.

i decided to create separate topic on this idea, with some updates. So, will it work?
I guess it will work. The idea is the same as measurement of doppler shift. But here we do not measure doppler shift, we measure shift in laminate. As wavelength is relative big, so Er can be measurement very well using phase information.
If knowing people say it is real, then I try to implement it and report the results. It is a little expensive at first, as we need to produce main board of well good known laminate, like rogers. After it is produced, measurement of any other laminate becomes possible.

So, here it is - Cheap and effective Er-meter design:
1. First we buy good laminate with known Er and etch the following microstrip PCB
- 5 000 000 Hz oscillator using DRO
- Mixer for generating 5GHz+1000Hz = 5001000 Hz frequency (AS I KNOW, it can awesomely easy be implemented using balanced mixer by direct injecting 1000Hz audio frequency in IF port, so mixer will work in other direction).
- Two Wilkinson power dividers for 5GHz signal and for 5.001GHz signal
- Signals divided as follows: LO1=5GHz, LO2=5GHz, RF1=5.001GHz, RF2=5.001GHz
- Two mixers for generating IF1 and IF2 from LOn and RFn. (SO, here is the second mixer, now for downconverting to IF)
- It looks like balanced mixer, BUT there is no 90 degree phase shift (read: microstrip lines are equal in length for LOn and RFn)

How it works:
- because microstrip lengths are equal then IF1 and IF2 signals are equal (phase and frequency of 1000Hz).
- we use knife and extract small rectangular part about 1x1cm of LO2 microstrip area
- then we take UNKNOWN laminate of 1x1 cm with etched microstrip and put it in 1x1cm hole. Maybe a little welding or connectors.

If we put original 1x1cm peace of laminate, then IF1 and IF2 phases are equal
If we put UNKNOWN 1x1cm peace of laminate, IF2 phase will change according to Er costant of dielectric: for example, phase can be "faster" or "slower" in this dielectric.

So we can use coplex FFT transform on IF1 and IF2 signals, calculate phase difference between this two signals, and convert it to Er constant. Because we know size of the 1x1cm peace, microstrip length is 1cm. All we need to examine UNKNOWN rf laminate samples is 1x1 or bigger peace of laminate with etched microstrip. The higher frequency.

It is cheap, easy to produce, we only need:
- 1 cheap RF transisto
- some DRO
- 3 or 4 Schottky diodes for mixers
- computer sound card to record signal
- the only expensive thing: piece of good laminate for device.

And also it is good: we can make this device for some frequency we use, and get real results for laminate.

Is it good idea or i reinvented some old trick?

ALSO i still have a question about Er measurement without spectrum analyzer and/or network analyzer: Can i do it with two antennas and fully etched copper (PCB without copper between antennas, measure power with and without dielectric plate)?

Thanks!

This sounds like it is similar to the same technique used in electronic stud finders. The kind that we use to detect wood in back of plasterboard.

When Zircon put the first one on the market about 30 years ago, I read a description saying it 'measures the dielectric constant.'

Since I knew capacitors contain a dielectric, this suggested it had something to do with measuring capacitance. Rather than spend 30 dollars (the initial price), I wondered if I could make my own.

I cut a large square of aluminum foil and attached it to the wall. I thought I could measure a change in capacitance (in the same way I was told a proximity detector does).

I tried applying DC volts through a high resistance to the foil, to see if any kind of charge developed, and how long it took.

I was unsuccessful.

I ended up buying one and sure enough it is sensitive to the densities of materials under it.

If you look into how such devices work, it may help with your own project.

Now I am dreamng about having network analyzer or even spectrum analyzer. What i can say they are too expensive, few thousand of dollars ever for second hand ones.

I did not buy rogers laminate yet, still experiment with cheap unknown brand.
What is good: i etched transmission line of few cm, and feed oscillator output to it. All is still working up to 5GHz
What is bad: i used RC method again to get Er, just measured capacity of unetched plate and used some formulas. BUT i do not trust the result.

I want to do some test on 5 and 10ghz, because i got few transistors for free. So now i want use other approach, please tell me if i am wrong:
My approach is to etch many band pass filters with step of 0.25mm. Then take fmcw sensor, feed it oscillator output to antenna through each of my band pass filters. Get analog low frequency signal for each filter, choose one that gives maximum amplitude, then use formulas to get Er. Can i success with this approach? I think that i can ever etch one band pass filter, then use knife to cut it step by step, and find length that gives best amplitude ..

Thanks people for your attention!

I personally don't know enough to comment on what you've described. I know sometimes you have to invest a lot to get a lot.

Since you speak of frequency shifts, mixers, phase differences, and even an audio signal...

I am reminded of how a metal detector uses two coils, which carry high frequency oscillations, and which detune, or intermodulate (or whatever), when close to dense substances.

There is a project called the 'poor man's spectrum analyzer' which is described at several websites.

It's supposed to be almost as good as a commercial unit.

Your point is cristal clear, he-he. That's the truth. Sometimes invest money, sometimes time.
My first big investiment was usb oscilloscope, then a soldering station. I thought it is expensive... at that point i did not realize prices on spectrum things. Now i see that such big investiment open door to interesting field of high frequencies. he-he.

oh thanks, i will check it out. I saw this topic few times, but did not read it. Interesting reading i guess, even if do not implement it.
Thanks!

Hello again! Peoples, what if i do the following trick of S11 measurement:

1. take FMCW device VCO for 5GHz, frequency is controlled with voltage. Replace it's horn antenna with self-made patch antenna on rf-laminate with unknown Er.
2. for this, i etch 10 simple patch antennas (each consist of 1 patch). Simple patch with WxH, where W=H and simple feed line. Feed line goes to the hole where the horn antenna was. Size of W varies, W=W+1mm. So i get various patches.
3. do measurements with each of antenna. I think that this gives some kind of S11 measurement.

So using this approach I can measure return loss of patch antenna with different sizes, choose one with biggest return loss. FMCW works as narrow-band spectrometer. After finding best sizes of patch, we can make new 10 patches with W=W+0.1mm, and find best one. He-he, hope my ideas will help someone too.