Power tube can be the only explanation then. I checked the driver tube (2 brands) and it clips before the power tube. So it is fine here.
I do not have other power tubes at the moment but I made a selector for the cathode resistor to be able to use also EL34 and KT88. I need just to find some at a reasonable price (since it is just a test).
BTW, the distortion is (strongly) dominated by 2nd harmonics (as expected) and stays so up until clipping. And both channels perform identical.
Any other issue I should check before trying different tubes (currently using these https://www.tungsol.com/html/6l6g-tung-sol.html)?
I do not have other power tubes at the moment but I made a selector for the cathode resistor to be able to use also EL34 and KT88. I need just to find some at a reasonable price (since it is just a test).
BTW, the distortion is (strongly) dominated by 2nd harmonics (as expected) and stays so up until clipping. And both channels perform identical.
Any other issue I should check before trying different tubes (currently using these https://www.tungsol.com/html/6l6g-tung-sol.html)?
Attachments
Right now I have 3 SSE operating in my house. One for 6L6GC's, one built for EL-34's and the last one for KT-88's. No doubt in my mind the KT-88 is the best sounding of the 3 I have with horns. With both Electro Harmonics KT-88's and JJ KT-88's on Arta the figures were much lower both in the 2nd and 3rd harmonics and nothing much else after the 3rd. THD + noise was .4% and .9% as previously posted at 1 watt. I do not know what relationship clipping of the input tube means in total distortion. It could well be the choice of pre tube contributing to the distortion numbers you are seeing. George probably has done a lot of research on this subject and could contribute his opinions better than myself. As you know any noise or distortion in the pre stage is only going to be amplified in the power tubes. My advice is bite the bullet on some KT-88 tubes. Re-bias the output stage for those tubes and see what you think. I think a cathode resistor between 470 and 560 ohms will bias a KT-88 nicely with the rail voltages you now have. I think you will prefer the KT-88 sound and the numbers are going to be much better as well.
I should add though, disregarding the numbers, the 6L6gc tube is one good sounding tube. Nerds should not get caught up too much in numbers if you like what you hear. What the screen shot you have shows is there is nothing nasty going on just higher harmonic numbers which in SET amplifiers many do not consider a bad thing. SET's are not the best looking on the bench, just sound good or better than the SS counterparts. What do your ears tell you?
I should add though, disregarding the numbers, the 6L6gc tube is one good sounding tube. Nerds should not get caught up too much in numbers if you like what you hear. What the screen shot you have shows is there is nothing nasty going on just higher harmonic numbers which in SET amplifiers many do not consider a bad thing. SET's are not the best looking on the bench, just sound good or better than the SS counterparts. What do your ears tell you?
I appreciate the feedback. A 470R cathode resistor is already in place (switchable). I guess Santa should know which KT88 pair to choose for me 
However, next week I can borrow a pair of EL34 and will make a test with those. And indeed I do not care of the absolute numbers of THD. I am only checking that everything is working properly. THD is the only one I could not find measurements about here on the forum. Hence I asked.
I am very happy of the 2nd harmonics behaviour (at different power levels) and the low noise floor (100 Hz peak at about -80 dB and cannot hear any hum or buzz from the speakers). Critical listening will occur after my working trip in 10 day or so.
However, next week I can borrow a pair of EL34 and will make a test with those. And indeed I do not care of the absolute numbers of THD. I am only checking that everything is working properly. THD is the only one I could not find measurements about here on the forum. Hence I asked.
I am very happy of the 2nd harmonics behaviour (at different power levels) and the low noise floor (100 Hz peak at about -80 dB and cannot hear any hum or buzz from the speakers). Critical listening will occur after my working trip in 10 day or so.
With a 470R cathode resistor you are pushing the 6L6gc tube pretty hard, at least for me. If my math is correct you are running your tubes at 26.35watts. Some of the newer 6L6gc's will red plate at 26watts. If you want to play some replace the 470r with a larger value and see what your numbers are then.
Math is close enough. Might try playing with the K to P voltages on your preamp tube. Changing cathode resistor. Could possible better the numbers you are seeing but why should be a question if it sounds good to your ears. All that really matters unless you just want to play. Playing with electronics is fun for me. Word is when you start trying to better your distortion numbers you are getting into another class of amplifier builder. Just remember to have fun, hobby.
I would jump straight to the KT-88 tube myself. Just from my experience with the EL-34 family of tubes. Kt-88 sounds best, certainly when it comes to the bass department.
I would jump straight to the KT-88 tube myself. Just from my experience with the EL-34 family of tubes. Kt-88 sounds best, certainly when it comes to the bass department.
I do not want to improve numbers. Just understand how it works. One of the reason why I decided to build electronics. 
I will swap tubes, do some more tests and report here if there is interest. Then I will consider this project done and start some critical listening... or just enjoy some music.
I will swap tubes, do some more tests and report here if there is interest. Then I will consider this project done and start some critical listening... or just enjoy some music.
Yup, the 6L6 family are a sweet sounding family of tubes. I've tried 807@360v, 5933@390v, 6L6GC@360|390v & 6L6WGB@360|390v. They all sound fantastic in UL. The 5933 & 6L6WGB have the best low end of the lot. I've tried them with 680 & 560R. The midrange & highs suit just about any style of music from Rock to Jazz & classical.
The EL34s sounded best at 360|390v @560R. Btw, is there any easy way to calculate the power out numbers without measuring gear, even ball-park figures would be fine.
I've heard 1/3 the dissipation of the tube is a good starting goal. I've never gotten that much out but for some it is possible.
I got SSE board number 1 out of the "box of dead circuits" for some investigating of alternatives to the hard to find IXYX ICP10M45S chip. This board was the first ever SSE board produced by a commercial PC board house, as all SSE's before it were test boards DIYed in my kitchen in 2005. It was built in 2006 and after some testing and use in an amplifier it was swapped out and became my test board for experiments that might not end well. Several traces have been cut and resoldered, the cathode bypass caps on the output tubes both exploded in an experiment gone wrong, and both of the power supply caps have been replaced but I don't remember why, as have many of the other parts. It was last used to test some large transmitter tubes, but that experiment didn't work out, so the board slept in the box since 2009.
I dug the board out of its box, popped in some tubes, wired up a pair of Transcendar 3K ohm OPTs for 6000 ohms in triode mode, connected it to a bench power supply and fired it up. With (expendable) guitar amp quality "Audio Glassic Blues Tubes" 6L6WGB's and a NOS 12AT7WC I get 1.8% THD at 1 watt and around 4% at 4.5 watts. I do not have FFT testing capability at this time. I listened to the old board cranked up to the edge of clipping on the scope (not audible) for 3 CD's just to wake up the old parts before testing. Everything sounded as expected. The OPT's in use do not have UL taps since they were intended for 300B's, so UL was not tested.
Some pictures and data are here, and yes there are alternatives to the 10M45 that are in stock.....for now.
https://www.diyaudio.com/community/threads/10m45s.390771/page-2
I dug the board out of its box, popped in some tubes, wired up a pair of Transcendar 3K ohm OPTs for 6000 ohms in triode mode, connected it to a bench power supply and fired it up. With (expendable) guitar amp quality "Audio Glassic Blues Tubes" 6L6WGB's and a NOS 12AT7WC I get 1.8% THD at 1 watt and around 4% at 4.5 watts. I do not have FFT testing capability at this time. I listened to the old board cranked up to the edge of clipping on the scope (not audible) for 3 CD's just to wake up the old parts before testing. Everything sounded as expected. The OPT's in use do not have UL taps since they were intended for 300B's, so UL was not tested.
Some pictures and data are here, and yes there are alternatives to the 10M45 that are in stock.....for now.
https://www.diyaudio.com/community/threads/10m45s.390771/page-2
Thank you George for chiming in.
I am back home and made a couple of tests with a 680R cathode resistor. With 6L6G the THD at 1W increased (as expected running them a bit cooler) by a small margin (from 2.8% to 3%). Then tested a new pair of JJ EL34II at 80% plate dissipation and THD at 1W got lower-ish at 2.4%.
Numbers do not matter. I like what I am hearing but a test with a pair of KT88 will happen at some point.
On a side note, however, the driver tube plate-to-cathode voltage is 255V and the current across the 10K resistors around 9.5 mA. This puts the 12AT7 at 2.45 W dissipation. Is this something to be worried about?
I am back home and made a couple of tests with a 680R cathode resistor. With 6L6G the THD at 1W increased (as expected running them a bit cooler) by a small margin (from 2.8% to 3%). Then tested a new pair of JJ EL34II at 80% plate dissipation and THD at 1W got lower-ish at 2.4%.
Numbers do not matter. I like what I am hearing but a test with a pair of KT88 will happen at some point.
On a side note, however, the driver tube plate-to-cathode voltage is 255V and the current across the 10K resistors around 9.5 mA. This puts the 12AT7 at 2.45 W dissipation. Is this something to be worried about?
R14/R24 will not affect the voltage or current through the tube until it's made large enough to mess things up. Leave the 10K in place for an amp running 400+ volts of B+ It's only there to take some heat off of the 10M45, and it should be replaced with a piece of wire in low voltage applications.
R13 / R23 sets the plate CURRENT and the cathode resistor R10 / R20 sets the plate VOLTAGE. I recently did some experimenting on an SSE and I simply put a 500 ohm pot in place of each resistor and tuned for lowest THD. The tube current can be monitored by placing a voltmeter across R14 / R24. Easch 10 volts equals 1 mA, IE 85 volts is 8.5 mA. The two pots interact to some degree. The plate voltage can be measured directly. Best measured THD does not always coincide with best sound quality, so listen before soldering the final choices in place. Every tube has a different sweet spot, and often the two halves are different. You could leave the trim pots in the amp, but use something of higher quality than the surplus Chinese Bourns clones from MPJA. I find that often best measured or perceived sound quality comes at or over the maximum dissipation rating. I have run some 12AT7's pretty hard at 2.7 to 3 watts each with no issues, but I have lots of spares, so they don't fail. If you only have one, it will die.
Attachments
Recently I found a lot differences between tube brands when it came to plate voltage. The voltage drop of the tube can vary a lot. Before changing anything try some other brands, preferable NOS one, of 12AT7's. The GE 6679 I presently am using has a K-P voltage of around 235V's each side fairly well matched. Some 12AT7's I tried had K-P volts as high as yours some lower than 235v's.