Jeff Liebermann wrote:
> miso@sushi.com hath wroth:
>
> >I'm trying to visualize how the pcb material interacts with the copper
> >track. It's not like you are doing stripline, i.e. metal traces with
> >the fr4 between the traces. The signal hits the copper from free air.
> >But a reflector would have the dielectric of the pcb in the path.
>
> Oh, that's easy. A non-radiating transmission line has at least one
> ground plane below the trace and is terminated at both ends.
> Sometimes, it's sandwitched between two ground planes but must always
> be terminated. A radiating trace antenna has no ground planes and is
> matched only at only one end by the output impedance of the
> transmitter or the input impedance of the receiver. The other end is
> matched to the impedance of free space or 377 ohms. Think of an
> antenna as an impedance transformer between 50 ohms (or whatever) and
> 377 ohms.
I think you are thinking too much, or you didn't get my original
statement. I understand stripline. [I made a high speed controlled
impedance dut board for an ECL DAC I was evaluating.] In the case of
the loop on a PCB, think of the antenna being used as a transmitter.
Then the element radiates in one direction (i.e. forward) without a
dielectric in the path. The reflector does have a dielectric in the
path, but only for a short distance relative to the signal path. So I
just don't see the pcb effecting the dimensions of the antenna.
I have copper tape, though hell if I know where. I think Excess
Solutions in Milpitas sells it. I could make some fiberglass as a
substrate if it is better than using pcb.
Back to the stripline board I designed, I had a HP TDR on loan at the
time. It was a TDR plus oscilloscope. My traces were close to the
target impedance. When I did the math, I came up with really wide
traces based on the spacing between planes. I was glad went truth
matched math. Then I was looking at a test board designed by HP for
their high speed chip tester. The traces were quite small, so I figured
the dielectric thickness was proportionally small. Still, you need the
thickness to be well controlled. As it turns out, the HP traces were
way off. I don't recall which direction (high or low), but I informed
HP of this error. Needless to say, it didn't look good when I showed
their board was no good based on their own instrumentation.
>
> The easiest way to visualize this by having the dielectric between
> between the antenna elements and the reflector be composed of two
> different materials, air with e=1.0 and G10 with e=5.0 (approx) with
> corresponding variations in thickness. I don't know exactly how to
> model such an antenna. I'm also lazy and think it's time you dig out
> 4NEC2 or other modeling program and try it thyself first.
> http://home.ict.nl/~arivoors/
> Send me you model and I'll try to untangle it. Use this model:
> http://802.11junk.com/jeffl/antennas/Biquad/biquad.nec
> as a starting point. You might also find it interesting to look at
> the web site where I stole the model:
> http://www.pow.za.net
> http://www.pow.za.net/panel_2x2commercial.jpg
>
> If you're going to dive into surface radiating patch or panel
> antennas, most of the NEC2 surface models are marginal approximations.
> Instead, use MSTRIP40:
> http://www.e-technik.fh-kiel.de/~splitt/html/mstrip.htm
> for something more accurate. Incidentally, there's quite a bit on
> strip line and surface radiating calcs in the "lab manual" at:
> http://www.e-technik.fh-kiel.de/~spl...0LabManual.pdf
> which might explain how to visualize the antenna.
>
>
> --
> Jeff Liebermann jeffl@comix.santa-cruz.ca.us
> 150 Felker St #D http://www.LearnByDestroying.com
> Santa Cruz CA 95060 http://802.11junk.com
> Skype: JeffLiebermann AE6KS 831-336-2558