Amp wars

In yesterday's post Listening through another's ears, I mentioned my trip to Arnie Nudell's home to begin voicing the new BHK power amp. Many of you asked for the results. I won't wax enthusiastically about its sound and trust that's ok with you. As my hero Alfred E. Neuman was fond of saying, "What, me worry?" But I will report my findings. This new power amp can be thought of as two separate sections: the input stage and the output stage. The output stage is all MOSFET, running class A up to about 30 watts or so, and fed from a giant analog power supply. That works great and will remain the basis and core of the amp. It's the input stage where all the magic happens. And that's where we focused out attention. The input stage is deceptively simple. A high voltage differential pair with a constant current source and no negative feedback. Bascom designed it such that we can run tubes or transistors to work on voicing it. Mind you that every device we put in winds up measuring the same at the amp's output. These are all sonic differences. It took about a microsecond to eliminate any kind of bipolar transistor. Ugh. Both Bascom and Arnie wanted nothing to do with bipolars but I felt we should be good engineers and try it. Just 'cause. This made it simple and cleared the way to narrow it down to either MOSFETs or tubes. Starting with MOSFETS we devised two topologies: a single high voltage MOSFET pair and a dual Cascode pair. There isn't space in today's post to explain the difference, but perhaps in another post. I spent a great deal of time auditioning the MOSFETS and found them quite good on the IRSV. The tube sounded better: rounded, spacious, layered. But the MOSFETS weren't bad. Then I went to Arnie's house. No one seems to know how to setup a more resolving system than he. We spent the entire day playing music, trying the different topologies. The tubes won. Yes, they just obliterated the MOSFETS, contrary to my initial findings. The cascode being the worst among the two MOSFET versions. The tube we settled on his called a 6H23. Now, I must admit, I do not (for even a moment) understand why the tube sounds so much more like music than the MOSFET does, in the input stage. I haven't a freaking clue. But it does. It makes music. And once you hear the music, how can you do anything other than just run with it? I cannot. MOSFETS in the output stage are far superior to tubes - and this I DO understand - but in the input stage remains a mystery to me. Sigh. I never cease to get amazed at what I learn from day to day, to day.