As I mentioned earlier, you can spend a lifetime studying antennas. There are a lot of abstract conceptual things to understand if you want to go the theory way. Equally, there are an awful lot of rules of thumb to learn if you go the empirical/practical way. Typically most antenna gurus have a hybrid understanding, in its simplest for there's the calculation for a dipole resonant length... and then the 95% end effect thing would be a rule of thumb.
The antenna itself is just one part of the equation. The matching of the antenna is a whole topic in its own right, but it's a key part of any antenna design, so here's a worked example for a simple dipole.
The dipole is a reasonable starting point except that it's a balanced antenna, and most (but not all) presentations are 50 ohm unbalanced (ie, coaxial). Coaxial is generally favoured over balanced twin-lead because it's easier to route physically and its performance isn't affected by other adjacent objects, like other feeders, pipes or walls for example. To terminate (ie connect the coax to the antenna), you could simply put the outer of the coax feed to one side of the dipole and the inner to the other side, and in practical it will usually work apparently reasonably well. The downside is that because you're feeding a balanced antenna with an unbalanced feeder, some of the power will also be radiated back along the coax outer as well as the antenna itself, affecting the radiation pattern. To fix this, you use a balun which is simply a high frequency isolating transformer, with two windings, one you put across the coax and the other goes to your dipole. This can be a simple off the shelf part you can mount on a PCB for low power applications.
All that is great but there's one other thing... a dipole has an impedance at resonance of round about 75 ohms, but your transmitter and coax feeder is 50 ohms, so you won't achieve an optimum match, resulting in some of the power being reflected back to the transmitter, rather than absorbed and radiated by the antenna. This is not black magic, this is all about maximum power transfer, you can do the same experiment at DC with resistors, and maximum power is transferred when the impedance of the load (antenna) is the same as the supply (transmitter).
Practically speaking, just like the balanced/unbalanced thing, you might well have difficulty in telling any difference in most circumstances, as the impedances aren't a million miles away from each other, but frequently that is not the case, and anyway it's bad form from an engineering perspective to throw away power. So to fix the impedance mismatch, you
could use a
25 ohm series resistor couple of resistors to convert that 50 ohms to 75 ohms at the antenna... but then resistors will throw away that power as heat, we want to transfer all the power to the antenna. So a better way is either to use an inductor/capacitor matching network, or as you're already using a balun, use a balun with a different windings ratio, ie 1.5:1 instead or 1:1. Typically to reduce losses, if it's economically effective, I'd rather use a single part than several.
Say, though, you do want to match 50 ohms to 75 ohms in a relatively lossless way at a given frequency with an LC network rather than a balun, how do you do that? The internet is at your service! There are plenty of online calculators for this, for example
http://leleivre.com/rf_lcmatch.htmlI plugged in a frequency of 433MHz, source impedance of 50 ohms and load impedance of 75 ohms. We are only interested at resonance, so in theory there is no reactive component on source or load, so we set the j ohm entries to zero. (Note that the j notation is to do with complex numbers: don't worry about this for now as it'll just confuse the issue, but if you do want to understand antennas in any depth you'll have to engage with the topic at some point.)
The calculator came out with two solutions (13nH & 3.5pF or 39nH & 10.4pF) and two "Nan" (not a number, ie unsolvable) solutions which simply can't be made with the topologies given. For the two solutions provided, you could use either one, but for this I'd choose the one with the most practical of components and easiest to fabricate, which would be the second one. Why? It's easier to fabricate the second one because parasitics (inductance and capacitance from PCB and other construction effects) on those larger value parts will be relatively less than on the first option. Sometimes though, there may be an engineering reason to choose the first option, for example you might want a DC path which you get with the first option. Particularly at UHF and higher, rather than using "lumped" parts frequently you can fabricate the matching on the PCB itself, using those parasitic characteristics to your advantage, but that's yet another topic.
As with many engineering topics, knowing what matters and what doesn't in a practical sense is frequently a matter of experience, and no matter how many theory classes you do, you'll never know what can be discounted and what's really important, how things interact, and how to prioritise things given a set of design criteria.
Home
Help
Search
Login
Register
