Determining Er of PCB by moving small copper foil piece along microstrip line!

[Terminator3]

While moving small copper foil piece along microstrip line I observed how noise/power levels depend on chosen position, and how their maximum repeat each few millimeters. And then i come to conclusion, that signal noise/power pattern repeats each half wavelength. As i understand it is because of half-wavelength impedance repeating, so adding stub at half wavelength apart gives similar noise levels. So i measured three maximums in noise/power and marked their position on PCB. For given line width and substrate Er calculations i have wavelength in dielectric of 15.08mm. Distance between first and third marker marker position is 15.3 mm. Looks like real wavelength in dielectric is measured! Correct me if i am wrong. Also making longer line can give more precise result, as we can measure, for example 5-10 wavelengths. I must also say that line is well-terminated. Please comment.

 

[volker@muehlhaus]

Sure, this gives an estimate of the effective permittivity.

For microstrip, the effective permittivity then depends on substrate material permittivity and also on line width. Wider lines have higher effective permittivity than narrow lines because the percentage of field in the PCB substrate is higher. For narrow lines, more of the field is in air, and we get a lower effective permittivity.

 

[vfone]

Microstrip resonators (line, ring, patch, etc.) are very often used for determination of the dielectric constant.
This approach is based on the difference of the estimated and the measured resonant frequency. Frequency dependent value for dielectric constant can be calculated at each resonant frequency.
Ring resonators are the best in this regards, because in contrast to distributed-line resonators, no end-effects need to be considered.

 

[Terminator3]

ring resonator is good, but it needs some equipment, at least vco.

 

[vfone]

To find the approximate resonant frequency of a ring resonator you don't need mandatory an oscillator, and just a VNA.

 

[Terminator3]

yes. but vna is too expensive at x-band. All good prices seems ending above cellular and wifi frequencies. VCO is cheaper, just measure uncalibrated power on the output of DUT circuit

 

[volker@muehlhaus]

To find the approximate resonant frequency of a ring resonator you don't need mandatory an oscillator, and just a VNA.

I think Terminator3 means that his method works with a simple fixed frequency source (no sweep required).