I’ve been thinking about an application where I want to provide transformer-isolated outputs (a 10MHz reference distribution amplifier). But I really find working with enamel coated (“magnet”) wire hard - stripping the enamel to solder to the wire is hard, and it is hard to keep track of which wire ends belong together. So I thought, I have a spool of single strand CAT-5 cable for installation, what if I used a twisted pair from this to wind the transformer? I probably couldn’t get as many turns on to a small core, but I decided to try. I managed to put 6 turns onto a T37-43 core, and soldered it to a small stripboard fixture with SMA edge-mount connectors:
Transformer in test fixture.
This is not the prettiest of transformers, and it is large compared to what is commercially available (such as the Mini-Circuits ADT1-1+).
I used to nanoVNA to measure the return loss and the attenuation of this transformer over 1 to 100 MHz (but I’m most interested in what happens around 10MHz).
Looking at the graph, we see that from 3-4 MHz, the return loss is above 20 dB, and it stays above 15 dB until 50 MHz. The insertion loss is minimal throughout the this range. For my use case, 10 MHz, I measure the return loss to be 22.2 dB, and the insertion loss to be 0.26 dB. This is not much worse than the Mini-Circuits part I referenced above (which has other nice features).
I conclude that transformers of this type, even with only 6 turns, will work just fine for the output of a 10 MHz distribution amplifier. Note to self: I really should get the serial interface to the nanoVNA working such that I don’t have to punch the numbers myself.