Output device for SE Hybrid Headphone Amp
 

Hi,

I'm looking at two different headphone amps to build. One is Erno Borbely's hybrid headphone amp, and the other is Pete Millet's "Starving Student". Both are Class A, SE, and run on the same rail voltage. The former uses 2SK2013 output devices with feedback; the latter uses IRF510 without feedback.

- Is there likely to be a audible difference between these two output devices, which appear to be operated as source followers?

- Can I use the 2SK2013 in the "Starving Student"? I understand that the threshold voltages for the 2SK1013 (depending on the Vgs classification) is much lower than for the IRF510

- I may want to parallel 2 2SK2013's. Are there any estimates on the variance of Vgs vs. Idss for the 2SK2013? If I wanted to match pairs, I'd like to get a sense of how many devices I need to buy in order to get a reasonably close match, say <5% from the same lot.

Thanks and regards,

Rob

Why go single-ended and discrete? You end up with more distortion, more cost (this was the main good reason to go SE, when semiconductors cost more), and more work to do.

If you're really intent on using such outdated topologies I'd suggest the starving student. At least it uses feedback. Even on that one the output caps are way too small and I'd be a bit weary of trusting that point to point design -- something rattles around in the wrong way and you can say bye-bye to your best headphones. It looks like they payed no attention to grounding but since they use tubes I guess noise performance is a moot point.

Honestly you're better off using a BUF634 and an op amp all with global feedback, or even a plain high-current op amp. At least this will be rather predictable. That starving student design has a THD and frequency response that I'd expect from a cheap MP3 player, and god knows how that other one without any feedback measures.