This is a follow-up on an article published in Croatia several years ago about a high-end and yet reasonably priced OTL headphone amplifier.
Let me first share a few pictures:
The amplifier was constructed to meet the following specifications:
no output transformers
zero negative feedback
single ECC88 and one 6H30 per channel
dual mono configuration
class A throughout the operating range
has to conform with the mehahertz criteria
AC tube heating
32 Ohm load capability
short signal path
The result was the amplifier shown in the pictures. Only prototype boards were used, there are no printed circuit boards inside. Signal path components were soldered by point-to-point technique by hand, arguably the best way to do it when tubes are in question. The signal path cannot be any shorter than it is so both the output and input connectors are on the front. To satisfy the megahertz criteria, the amplifier has five-times RC filtering per channel and the tube heating is done by AC rather than DC. Despite the slightly higher noise floor, the AC provides for higher linearity. The response extends beyond the 1.2 MHz mark and all high frequency modulation is well beyond the audio range as a result. More or less perfect linearity is achieved between 12 Hz and 80 kHz and slightly less between 8 Hz and 120 kHz at -3 dB roll-off.
Components used were mostly European and Russian (old Soviet) componentry acquired in large quantities, measured and selected by hand to ensure minimum tolerance between the channels. Russian paper-in-oil capacitors were used as well as Wima (red) for bypass. I like Holco military grade resistors so that is what was used. The tubes are Russian military specified tubes manufactured in a certain hangar and under a certain serial number. The 6H30 tubes were, of course, matched.
The amplifier was built to allow for usage of 32 Ohm headphones. While it was possible to achieve a sufficiently low output impedance, the amplifier still has more control over headphones with higher impedance. This is then something of a proof that manufacturers of tube amplifiers employ massive amounts of feedback to counter for this problem. As such, those amplifier cannot, and that is a mathematical certainty, be linear.