Hello, I am currently finishing the build of a portable UHF two elements cubical quad. I modeled the antenna and the simulation shows that it should have a 90 ohm feed point impedance.
I used Jerry Sevick dual-stage design of a 50 to 90 ohm balun. (a trifilar unun followed by a 4:1 guanella balun)

The balun enclosure and the radiating elements are not close one to another (mainly to improve mechanical / center of gravity specs of the antenna), so there will be some balanced transmission line from the balun up to the radiating element.

The easy choice would have been to use some 100 ohm differential pair or better yet, some 100 ohm twinax. The problem is that twinax sockets and coax are quite expensive. whereas 100 ohm differential pair nowadays seems to be mostly used for data and low power signals, and i don't think using a RJ45 socket for more than a few watts PEP would be good engineering.
And its UHF, so balanced terminals should be made carefully because in terms of wavelength even small irregularities start to be noticeable.

So, I devised to use two 50 ohm coax, with the braid of both coaxial cables connected to the center tap (leads 10 and 12 in the schematic) of the guanella balun. On the radiating side of the antenna, the coax braid would remain open, only the center conductors would be connected to each loop leg.

The question is : Is this setup ok in term of transmission line impedance matching ? (2x50 ohm is close to 2*45ohms) however I have certain doubts about this setup, and also the preferable coax lengths i should use if this setup is viable.

in terms of impedance, each coax is V/2 relative to the center tap, while the current is the same on the center conductors; which should make Zcoax = Zdiff/2 = 45 ohms

I have attached a schematic URL to make the setup clear.

I plan to test the setup with a nanoVNA box when i have it, I am still waiting for the shipment.

Advice is appreciated !