We left off in the previous post trying to figure out a slightly more rigorous approach, to connect together two antennas, resonating in adjacent frequencies. Let’s recall:
The designer may (or may not) choose a point with higher impedance. This can be done in one of two forms:
- Design the patch differently. I know this an opaque statement, but there is more than one way…
- Choose a different feeding distance. This is actually a better idea, most of the times, as it is just simpler. I’ll leave the readers to try it out for themselves.
The experienced reader should recall that every half-wavelength, a point completes a rotation around the smith-chart. However, more rotations means more dispersion. Not super relevant right now, but good to keep in mind. Having said that, maybe a shorter distance can do? On paper, yes. However phyiscally, the antennas are intersecting
This is the reason, if you wondered, that this antenna arrangement exists…
Ok, now let’s expand to a third antenna!
Wow, much bandwidth. However, let’s look at the individual Antenna performance, Vs. the concatenated one.
There is a significant deviation in frequency. Why do you think that is? take a moment, try to think what parameter might align these together. This trait is critical for creating significantly larger arrays.
Looking For The Last Parameter
So we’ve optimized the distance between the Antennas and their size. What else? Truth be told, in this design it’s easy to miss. Want another minute?
Dual-layer microstrip patches are fed in the same layer they are printed. This quite an issue, actually, because the feed chamfers part of the Antenna itself. Besides reducing it’s gain slightly, it also creates a portion of different transmission line, a co-planar waveguide. Without telling anybody here, I already took that into consideration, and set the gaps such that even this small portion of trace is still 
Truth be told, I spoiled the answer earlier. When I designed these, I didn’t pay too much mind to the feed lengths. But good practice would be to take these into account early on. The phase leading up to the transmission line is critical, to maintain high enough impedance at the relevant frequencies. The accute reader is already thinking, it may be possible create a nice design topology in which the antennas are already fed in a different impedance. Well, you are not wrong, but it makes everything just that much harder. Keep it simple!
Now, let’s try to make a real analytical approach out of it.
Don’t Deviate, Formulate
So since everything is scaled with respect to wavelength, you might notice that the following traits:
- Patch dimensions
- Feed length
- Trace length connecting between the patches
increases in a constant ratio. Namely:
and so on, and so forth. Namely it increases exponentially. So if we apply a logarithm on it, we will obtain
Wow, a log periodic Antenna! Now you know why! This is fine and dandy, but realistically speaking, it will work up to several limits:
- The smaller antennas, especially in this feed structure, will be affected significantly by the deformation caused by the feed. This should cause some trouble.
- The parasitics of a close ground will affect the larger antennas much more significantly.
- Once the frequency goes low enough, the ground size itself will be a parasitic on the low-band antennas.
Another annoying trait would be that when the lowest band frequency is an octave below the highest one, the higher one will kick in. But that is not necessarily a bug, rather a feature.
Hope you enjoyed the read. Go design antennas!