Dear Odyseus,
I would like to tell two Motorola Application Notes on this topic, as an addition to Chinasky's very good offerings from the Philips side. See url
**broken link removed** and look for AN758/D at the very bottom of the long list. Also see URL **broken link removed** and look for AN721/D.
In building wideband RF baluns, the main difficulty usually is to make a transmission line with the needed characteristic impedance. For up to a few tens watts of RF power these lines can be made from twisted enameled copper wires. The point is how tightly you twist 2 or 3 paralled enameled wires because the characteristic impedance varies with the tightness of the twist. The more tightly you twist, the less characteristic impedance you get and also as the wire diameter increases, the char. impedance also decreases. The practical lower limit that is reachable with a diameter of 0.6-0.7mm wires which are very-very tightly twisted together is around 18-20 ohms. To receive lower than this characteristic impedance for a line, you have to use copper stripes unfortunately. Of course, if the toroidal core (what you have or can afford or have space for a higher size core) takes two identical twisted lines mechanically, you can connect them paralel and this way you can half the characteristic impedance.
These are in nutshell on the practical aspects.
On the lower impedance side of a toroid transformer (or balun) you can measure the inductance of that side with the other side open. (Of course if the balun is 1:1 in ratio, then both sides are equal in inductance.) The rule of thumb here is that the inductive reactance (X-L) of the lower impedance side should be 4-5 times ohms larger than the the lower impedance itself. This gives the lower limit on the permeability of the core. Practical wideband transformers with frequency ranges from 2MHz to a few tens of MHz should have a toroidal core with u(mu)=around 700-800. For higher frequencies the u is around 120-200.
I hope these thoughts are of some further help.
Regards
unkarc