A query was raised on the RSGBTech forum about an odd-looking balun. It consists of two pieces of coax, one a quarterwave and the other three quarterwaves long. The inners are joined at one end to the feeder, the sheaths are strapped and connected to the feeder sheath.
At the other end, the sheaths are also strapped. Each of the inners goes to one side of a 50 ohm load. I think it might also be worth strapping the output sheaths to the input sheaths and folding the whole balun up.
The original notes by I0QM from 26 Feb 1977 are based on a sketch by I4BBF from the early 1970s and are here: http://www.iw5edi.com/ham-radio/files/I0QM_BALUN.PDF
The mode of operation is a combination of the classic 4:1 balun using a half-wave line to provide a 180 degree phase shift, and two quarterwave transformers, each terminated with half of the 50 ohm balanced load. A 25 ohm resistive load on the end of a quarterwave of 50 ohms coax is transformed to 100 ohms. Those two are in parallel, so the feedpoint sees 50 ohms. The clever trick is using longer bits of coax to do the transformation. Neat.
I made one to test things out. I used RG213 (VF=0.66) with one leg 2115mm (0.75λ at 70.2MHz) and the other 705mm (0.25λ at 70.2MHz).
I connected an isolated 50 ohm 250W flange-mount resistor between the two ends and stuck 10W at 70.200 MHz into it. I used my Bird 4410A with a 50-200MHz x10 slug. The reflected power was so low even on the 0.1Wx10 range that I reversed the connections to be sure it was really that low.
Between 68 and 72 MHz, the reflected power was too low to measure, certainly under 50mW. The load resistor is a Florida RF flange mount which is good to several GHz.
Looks like a useful rediscovery for feeding balanced 50 ohm loads.
Soldering in a hurry….
10 watts forward at 70.2MHz
Nil reflected power at 10W forward at 70.200MHz
You can of course also extend the coax after the output terminals, but you would need to use 25 ohm coax or two identical-length 50 ohm coaxes in parallel on each leg to maintain the match to 50 ohms balanced. If you don’t mind the SWR on the line being 2:1, you could just use an integer number of halfwaves of coax on each output of ANY impedance, as usual.
As the outer shield of the coax is only 0.66 of a wavelength in circumference, there should be no weird effects with asymmetric interaction with the antenna.
It would be quite possible to bond the shields at the input and output, and even coil the balun up inside a metal box or mount it inside the boom of a yagi if the coax loss was low enough to ensure the cable temp didn’t get excessive (says the op who burnt his hand on a temporary length of Chinese RG402 carrying 140 watts on 2.3GHz the other day…)