Ok, thanks, that's what I thought you were doing, more or less.
Why does one side of the input need a ferrite, while the other is ok with air core?
Wouldn't you get better balance with three windings on the same transformer?
Clifford Heath.
No common mode voltage drop. :)
I'll sometimes put on a little FB, just for feels. Doubt it does anything. (Again -- not pushing the envelope on bandwidth here, so it doesn't much matter that the length of transmission line is balanced, yet being used for an unbalanced signal. Not much radiation given the length and bandwidth. Some shielding would be better, really.)
If you are pushing bandwidth, such that the transformer's winding length is a sizable fraction (or multiple!) of the minimum wavelength of interest (i.e., at max F, the phase shift across the transformer is >= 90 degrees), then you'll want to be much more careful about common mode (loss, radiation, susceptibility).
Howso? Y'mean the input inverting transformer? There's no way to improve balance further, by doing just that -- if I added another winding to ground, to induce a known reference voltage in the core, well, that would just be equivalent to the one winding that's already shorted to ground on the far side!
Or to put it another way: I can't have an instantaneous "induces voltage in the other windings" winding -- in the ideal case, that would violate causality(!).
The ideal arrangement, to illustrate that, would be: using a straight transmission line segment, with ferrite beads stacked all along it. No wave, between any pairs of wires in that TL, can propagate any faster than the speed of light does. (Which is locally fixed, because, TL, dielectrics and all that.)
Another way to put it: the waves all need to end up, in the same place, at the same time. Preempting it would be /worse/!
Or -- since I do have the TL bunched up on a core, I could apply a more "instantaneous" wave, by looping just one turn around the core. But then it wouldn't be the right ratio, and after a few TL delays, the signal would be all but shorted out. (A little blip would still get through, but this is like putting a speed-up cap across a voltage divider. You're still losing all your signal! :) )
There are better ways to enforce balance -- suppose I had a straight length of TL here. If I start the first 1/3 length as coax (it'll be special 100 ohm coax, in this case..), and that goes through a tall stack of ferrite beads, then it transitions to 100 ohm twisted pair for the middle 1/3rd, with more beads, then for the last 1/3rd as coax connected backwards, and the rest of the beads, then I'll have about the most balance I can get.
Well okay, I could come up with some TL sections that have different amounts of E-field exposure (which is basically what we're balancing here, by going between coax or twist or whatever). The first and last 1/5th could be coax, the middle could be twisted pair, and the intermediate (2/5 and 4/5) could be asymmetrical twist, like 2/1 twist or something.
You can take this into the continuous limit: Suppose you start with coax, then you slit the shield and taper it back, so it gradually covers a smaller and smaller arc segment over the core, as you go along the length. Meanwhile, the core gradually thickens and widens, and becomes more distant from the shield (or, what used to be the shield!). In the middle, the cross section is that of parallel round wires; at the two ends, it is of coax; but at either end, the core and shield have swapped positions because of this slit-and-taper geometry. And, at each point along the length of this transition, the axisymmetric impedance remains constant (i.e., you plug any cross-section into ATLC and it always says it's 100 ohms).
Now you wrap this with an insulator, and slide ferrite beads over it. The individual ferrite beads shouldn't be very tall, and they should be insulated from one another with low-K dielectric (because ferrite has a high K, and acts as an AC conductor itself). This structure is then suspended relatively distant from the walls of a metal box, which should maybe be lined with absorptive foam or something.
This is the most ideal possible structure inside, say, a Picosecond Labs (now Tektronix, isn't it?) inverter transformer. :)
Natch, it's hardly the most manufacturable possible structure, but it would be good for frequencies up to the TE/TM mode limit of the coax used (so, some GHz).
It would be pretty to 3D-print though.
Tim
-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: http://seventransistorlabs.com
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.