Ok, so I got the amplifier together without burning myself too badly, and I must say I really like it!
I picked this kit because it had gotten great reviews as a good example of single ended tube amplifiers. Unfortunately most of those reviews are of the 'flowery' subjective kind.
If anything probably a bit worse, because it's a tube amp.
Luckily Ken Rockwell posted some numbers and plots which confirm that this is indeed not one of those 116 dB SINAD amplifiers, but should still have serviceable sub-percent noise and distortion unless driven into its (few Watt) power limit.
Listening, I'm very pleased that I can't hear any audible deficiencies (I have what you might call about 66 dB or 11 bit ears, i.e. I can discern 0.05% THD in a 125 Hz test signal). The amplifier has plenty of power to drive my KEF R3s to my listening levels and beyond - I typically listen at 60 to 75 dB, with peaks around 77 dB and 91 dB respectively according to REW (the average to peak ratio probably depending quite a bit on the music). A few Watt of power should be enough for that.
But I also wanted to do at least a very rough check that I'm not fooling myself into hearing what I want to hear, so here are a few very rudimentary and low quality but hopefully still somewhat meaningful measurements.
I'm comparing acoustical measurements of frequency response from 20 Hz to 20 kHz and distortion at 100 Hz and 1 kHz for the Elekit TU-8200R against a Loxjie A30, with the rest of the audio chain exactly identical.
I'm running REW on my Microsoft Surface 7 tablet, I use the DAC and headphone output of the tablet, feed that into the line input of either the TU-8200R or the A30, and either amp drives one KEF R3 on the left channel. The speaker gets recorded by REW using a UMIK-2 microphone at a distance of 1 meter, coaxial with the tweeter.
The TU-8200R is set to 'ultra linear' mode, output tap switch on high impedance. It draws 64 Watt from the mains at 122 Volt, if I can believe my trusty Etekcity WiFi plug/power monitor, and runs with the case 13K (above transformer hump) to 22K (center between tubes) warmer than ambient (38C to 47C for 25C ambient, measured after a few hours).
The first plot shows frequency response with the same speaker in the same room, same microphone, same positions, for two SPL levels; so please ignore the wiggles, what matters is the differences in each of the two sets of curves.
I'm relieved that there doesn't seem to be any 'warm' tube amp roll off at high frequencies, and the output transformers are not doing anything horrible at low frequencies either. The TU-8200R response (driving the KEF R3) is a bit (1 dB?) higher around 2 to 4 kHz, and maybe a dB or two lower around 150 - 250 Hz, but that is in a region of pretty wiggly room response, so maybe one should refrain from overinterpreting that.
Distortion: these are particularly bad measurements. There is quite a bit of mains 2nd (and 4th) harmonic coming through, and the noise floor is overall pretty horrible, but the measurements should be good enough to exclude or confirm multi-percent second harmonic, should it be present.
1kHz 75 dB amp setting (less than 75 dB for the single frequency in this measurement):
1 kHz 85 dB:
100 Hz 75 dB:
100 Hz 85 dB setting:
For comparison, the solid state amp driving the same speaker:
1kHz 75 dB setting:
10
0 Hz 75 dB:
100Hz 85 dB:
So there is definitely some added distortion due to the TU-8200R at 85 dB noise level setting (80-81 dB single frequency SPL), but we are at the limits of this measurement setup:
frequency / SPL | THD solid state A30 | THD tube TU-8200R |
100 Hz 75 dB | 0.45% | 0.36% |
100 Hz 85 dB | 0.38% | 0.67% |
1 kHz 75 dB | 0.067% | 0.41% |
1 kHz 85 dB | 0.14% | 1.33% |