Photos on construction progress
Tube characteristics using uTracer
Extracted models: 1625 Triode and 1625 Tetrode (description of the models can be found here and here)
Modeling using LTSpice and the extracted models
Notes on cathode resistors and bypassing in push-pull designs
Experiments with power supply
Examples of measurements at the 8Ohm outputs showing the switching diodes in the power supply, without snubbers on the secondary and with a 10nF+1kOhm snubber, respectively.
without snubber 
with snubber
Using slugging 10nF capacitors parallel to UF4007 diodes created more ringing (which agrees with this, saying that fast-switching diodes should not be paralleled with capacitors).
with (red and blue) and without (green and cyan) 4x 10pF caps, parallel to rectifying diodes
A bit more about snubber can be found here. Some extra plots:
Testing the Lead-RC network
Testing the open loop connecting with a RC-lead network for different C, with R=39kOhm. Examples of Left and Right channels in open-loop configuration, and the frequency responses for (blue) no lead-network, (yellow) C=82pf, (green) C=100pf, (red) C=200pf:
with 100pF+39kOhm lead at 2.2kHz 
openl loop, red-without 10kOhm grid-stopper, and green – with
THD measurements
no feedback and no input signal (pink), no feedback with input signal (orange), input signal and 16dB feedback (green) [0dB = 1Vrms]
Experiments with coupling caps (open-loop gain)
the coupling cap between the pream and power states is 0.015uf (red), 0.047uf (blue) and 0.1uf (green).
LTSpice model
Influence of lag-RC network (39kOhm+100pF) on the feedback
(pink) only lead-RC network (27kOhm+82pF) and (yellow) lead and lag-RC network (39kOhm+100pF) at 8kHz square wave.