There is always something better
I hear you and I'm always leaning into that direction, but...
You're generally right when it comes to simple systems, like connectors, for example.
Connectors have minute parasitic capacitance and inductance, insignificant in comparison with the cable they are terminating that you can always go in any direction you want, use better alloys, plating, shapes, whatever.
With complex systems, improvements in one aspect don't come without diminishing performance in some other aspect.
That comes from the established envelope for the system usage.
A good example is racing.
An LMP can be quicker than F1, but only if it goes outside the rules for LMP class, as shown with Porsche 919 Hybrid, while F1 stays withing it's envelope of rules.
So, the envelope of "rules" we're dealing in headphones are pretty wide open and not very strict, but one stands out as the most important:
-Striving for more sensitivity, in order to use less power.
Now, there's a saying that "every system's performance approaches perfection if its efficiency factor approaches zero".
So, if you allow me to drop the sensitivity of the Interface-cable-ribbon system by 10dB, you will have a much higher bandwidth and perhaps with that, marginally better highs.
That would require 10x higher power to drive the headphones, so it doesn't come for free, as we're adding complexity (and expense) on a higher powered amp.
If that amp would have to be an SET, expect 10x higher price for 10x the power, or deal with the lower SPL.
Now, within the envelope I have set for my product, there is no way to engineer a better transformer, I can assure you of that.
It started with the cable design that pushed the limits in what is physically possible for reduction of parasitic inductance, then designing a transformer for that.
I had no pre-constrains when I designed it, aside from cable parameters that were pushed to the limit, and things were done the way they had to be done.
If it needed to be bigger, it would have been, if it needed to be a C-core with dual bobbins, it would have been, if it needed to have a gapped core, it would have it, but it isn't all that because those actions are not better for bandwidth when you add pros and cons for each of those features. As in all complex systems, you win on one end and loose on the other.
It ended up to be a toroidal core (for windings that cover the whole core, minimizing the parasitic inductance), annealed (for having higher initial permeability, reacting better to low level signal), no gap (for reducing number of turns which reduces resistive losses as well as parasitic inductance), special terminating plate that parallels the secondary windings (without adding parasitic inductance).
Now, we audiophiles like to view things as we like to view races. There has to be a winner, but...2 seconds faster in a 90 minute race of 300 kilometer distance?
That is essentially the same performance. Only 0.037% better, which means nothing, really, it just makes a winner or looser crossing the line.
Finishing a full lap behind is just 1-1.5% worse, (but it costs $100M less for a season).
So, yes, things could be done to improve on the existing transformer, but it would require obtaining custom nanocrystalline cores and Silver wire windings.
And I'm actually thinking about doing that lately, but the price would be really, really high in retail, so I'm keeping that on a back burner for now.
The margin of quality with our existing transformers is almost fully saturated, so it's very hard (expensive) to go up from here.
