This is actually how it works:
The signal to the headphone circuit IS fed via the tube circuit. The output impedance of the LS 100 tube circuit is 300 ohms. Headphones range from 32-600 ohms generally speaking, with Grados being 32 ohm, Senn's being 600 ohm and others somewhere in between I imagine.
The rule of thumb between low-level (preamp) and high-level (amp) impedance ratio is about 100 x. Thus, our preamp would be ideally suited with amps having a min. input impedance of 3K.
in the case of speakers and amps, there is also a need for a ratio (called damping factor) between output impedance of amplifier and impedance load of speakers. Most SS amps have Zout of MUCH less than an ohm, hence product damping factors of 100-1000+ for SS amps. In the case of tube amps, it is typically much lower and this too attributes to the different control that SS amps have over speakers vs. tube amps.
The headphone is just a speaker, albeit requiring MUCH less power. Our headphone amp produces 1.5W at clipping into a 32 ohm load.
The purpose of the headphone amp itself, is to take the signal from the preamp tube output stage and buffer (amplify current) and provide impedance matching. In our case, we use a discrete FET stage to buffer current and provide the VERY LOW Zout (on par with SS amps) in order to drive headphone loads properly.
The other way that this is accomplished with tube preamps and headphone amps is with output transformers, that allow for impedance matching. I did not experiment with high-quality output transformers for the headphone circuit in our LS 100 because they would have added significant cost to the overall product.
Most mfrs. that add a headphone output, simply add an op-amp to the output of the circuit, which also buffers current and matches impedance, but does so with an inherently feedback-based device. In our case, the circuit is discrete, pure Class A and uses zero feedback.
I believe that this provides a high-quality headphone out as an added feature without significantly raising the product price.
We power the FETs in the headphone amp from the 12V windings of the transformer that power the heaters in the tube circuit. Obviously the headphone amp voltage is separately regulated.
I hope this clears things up.
Thanks,
Dan W.