Your question ought to be, how much difference will you see between the asy mmetrical and dispersive microstrip structures on the outer surfaces of the board and the more or less symmetric buried stripline structures with grou nd planes above and below them, inside the board.
formatting link
formatting link
formatting link
One practical point is that buried strip-line has to be rather narrower for the same characteristic impedance than a microstrip line on the surface of a board. I couldn't use buried 75R lines because the local printed circuit shop couldn't reliably etch narrow enough tracks.
Forgot. In LTspice you MUST set the inductor's Rser to 0, else LTspice puts 1m ohm in there by default, which obviously makes near DC look 'strange'
You put all the coil's series resistance AND resonant capacitance 'external' to the Current Source and coil.
If you want 'frequency dependence'; wait until you include eddy current losses into your model. Now THAT adds some complexity.
ALTERNATIVE METHOD: Another way to model an aircore current sensor is to make a simple transformer, where the primary is the Conductor wire and the secondary is the sensor coil. THEN everything goes into the Coupling coefficient. Don't worry about the primary having two components of some inductance and some mutual inductance, the coupling coefficient sorts that all out.
Although, you will find out that much becomes obfuscated into that K-term. It works,is straight forward.
That is asymmetric stripline, mathematically a bit nasty. EM simulator territory.
Attenuation will of course be a function of frequency. Also a function of dielectric losses and, at high frequencies, the surface roughness of the copper.
I'm guessing that adding the guard traces will increase attenuation.
Ah, if you really want guard traces, make them wider and farther away and nail them to ground with lots of vias. That has a name, asymmetric coplanar stripline, something exotic like that.
Fab one and measure it! There are too many unknowns (for me) to analyze it.
I bet one of the laminate companies, Rogers or someone, has tools and motivation to do some analysis for you. Ask them which of their wonderful laminates would work best.
One of my co-workers did a 2 GHz stripline ring resonator in Rogers 4350 to test dielectric constant and Q. The manufacturer's loss tangent was assum ed to be close enough. A degraded conductivity can be backed out of the Q measurement, and partially attributed to roughness once the dielectric loss is removed from the calculations. The conductivity was estimated to be be tween 0.4 and 0.5 that of smooth copper.
I am not sure how much the degradation due to roughness we considered for R ogers 4350 compares to FR4, but the test does indicate that it can be signi ficant.
Another note: the 3D simulators really don't do roughness. One has to put in conductivity. While your application is narrow, I just point out that c onductivity due to roughness is frequency dependent, as is dielectric loss. These are in addition to skin effect issues.
Thanks for the 'heads up' on roughness. Once conductivity lower than expected caught me off guard when we plated the copper onto the substrate and I found out that there must be 'air', or crud, or something making the copper porous, because we did NOT get 58MS/m conductivity in that plated on material.
When you said, "...conductivity was estimated to be between 0.4 and 0.5 that of smooth copper." Did you mean eddy current losses were that much, or BOTH eddy current and roughness losses were that much?
Now you've got me curious how to 'add' rounghness effects.
I forget what it is about plated copper (and possible defects, including poor crystalline structure, porosity, etc.), but any kind of structural or crystalline defect impacts resistivity. Defects are much higher, causing more resistance, and built-in strain (deadly when it comes to pure tin platings..!). Essentially, work-hardening in situ. Which is at least in part why basically all platings are harder than their annealed forms.
The strain can also be a challenge to getting a good plating to begin with: many metals want to flake off, so can't be depositied too thick.
More like, as skin depth gets close to scratch depth/width, the surface effectively gets longer. Think "fractal dimension".
(On a sufficiently small scale, all surfaces cease to look like surfaces, and instead consist of foggy beachballs, neatly stacked. But well before those wavelengths are reached by photons, the crystal ceases to behave like a metal, or insulator, anyway, so we're no longer talking about metals as such. The limiting case -- the finest scratches that should/could matter -- is somewhere around the wavelength of the metal's plasma frequency, or the Dubye shielding depth. ~~100nm range in any case.)
In a simplistic tool like TXLINE.EXE (AWR), you can add a conductivity numb er. Presumably, things like skin effect and roughness get lumped together in such a single number, and will change with frequency. It is a bit chick en-egg, as you need empirical data before "simulation."
I am not sure what you mean by eddy currents in the context of transmission lines. There was no attempt in my co-workers experiment to segregate thin gs like skin effect from roughness. It was all lumped together to obtain a net Q. (Actually and again, the net Q was the thing provided first, by me asurement.) Only the dielectric contribution to loss was segregated. The rest was lumped together as "copper loss." I realize that is a bit crude, but it was good enough for what we were doing (printed filters).
The roughness is added to improve adhesion and thus improve peel strength. Rogers has so-called LoPro for less roughness. It does reduce loss. I hav e not used it (yet).
I use the phrase 'eddy currents' to separate out the non-roughness aspect. So far leaving out 'roughness' close enough.
I do see where supplying Q would lump the two losses together. Back in the old HP days we gained over 10% Q just by silver plating [highly polished] silver onto the waveguide resonators.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.