Let me offer a few words on matching tubes.......
The GK1 uses its 6ES8 in cathode follower (CF) mode; that is, no voltage gain, but very high current gain. The preceding SS stage gives a small amount of gain, the signal passes to the level control, then to the grid of the output tube. The output of the tube is then capacitively coupled to the output RCA jack.
A CF produces around 5/32 of the distortion of a conventional plate loaded triode, which is commonly used in all tube amplifiers for its voltage and current gain. When the tube is used in plate loaded configuration, it is extremely susceptible to imbalance, particularly in the phase splitter, where technically we like outputs in each phase to match in amplitude. In cathode follower form, this changes however, because there is no voltage gain and thus mismatches do not result in differences in output.
The GK1 uses two triodes wired in identical CF mode for each channel. They operate in antiphase. While technically they should produce identical outputs such that their combined currents sum to a constant, this property is exploited only for the relatively non-critical bass frequencies and removes the influence of the power supply. However, the midrange and top end quite deliberately utilize the interaction of the tube with its power supply. So using tubes with matched sections here is not necessary to the all important mid and high end bands.
However, this is audio, and there are undoubtedly subjective differences. I'm told by DSK (whose killer system is a jaw-dropper) that the cat's meow is the Seimens gold pin 7308, which is around $US80 per matched pair. You can use the following tubes in the GK1:
6ES8 (US) or ECC189 (EU)
7ES8 or PCC189 (7V filament)
6DJ8 (US) or ECC88 (EU)
6922 (US and RU)
7308 (German and US)
Those who know their tubes will recognise that these are all high transconductance frame grid tubes, around 12.5 milliseimens (mA per volt). Frame grid tubes are very quiet and were the penultimate development of this technology, the last being the Nuvistor. Plate impedance of these lies between 2.5-3.2K. Correct plate operating current in the GK1 is 8.2 - 9.5 mA; but it's not particularly critical.
There is no doubt different tubes from the many manufacturers do sound different. However, this is a subjective minefield, because there are only seven or eight factories, and with tubes there is a great deal of re-badging, so you often don't know what you are getting or where it was made. Large tube dealers often had their own 'house' brands; this clouds the issues somewhat. Seimens, Amperex, RCA, AWA, Sylvania, Mullard, Brimar and Tesla are all very good manufacturers, but the nature of the GK1 design plays down differences because the tube is not used as a voltage amplifier.
The variable mu issue is interesting. I originally cast around looking for a suitable tube for the GK1 with a gimlet eye on cost. The 6DJ8/6922 was my original choice, but these were very expensive. The sound was extremely clean, almost transparent, and not quite what I wanted - a little color to aid engagement was needed. Since the intrinsic distortion of the CF is so low, to achieve this color I decided to use the variable mu variant of the 6DJ8 - the 6ES8 - and this has been extremely successful. This tube is readily available, not overly expensive, has a secondary use in compressors for recording studios (one of the best audio compressors in the market is made in the States and called the ES8!!), and was originally designed for Automatic Volume Control elements in AM radios where the variable gain function (changing gain with operating point) in plate load is essential.
The HT bypass simply bypasses the entire GK1 - SS and tube sections combined - for connecting up a HT unit, which of course has its own source selection and volume control. It has often caused confusion but may be omitted if desired.
Hope this clears up the debate!
Cheers,
Hugh
