I don't think I discussed it, but one change I made to the 2A3 amp in the last round of tweaks was chaning the input stage from LED bias to a typical cathode bias with a Jupiter Cosmos electrolytic bypass cap.
In the past, I've been very critical of electrolytic bypass caps, there are numerous electrolytics that sound very bad. However for the Aegis circuit, I had to use an electrolytic bypass given the high plate resistance of the 6SL7; it was a necessity to get good bandwidth out of choke loading that stage. I did a shootout of several "audio grade" electrolytics and the Jupiter Cosmos were my favorite, followed by the Elna Silmic II, well beloved in the DIY community for their sound. Not only were the Jupiter caps the best, but they actually sounded REALLY good, so good in fact that I compared them to LED biasing in a spud amplifier I was prototyping at the time. What I found as by and large I actually preferred the Jupiter Cosmos over my favorite LEDs. So, I made it a point that I should give them a shot in this 2A3 amp before I declare it completed.
I did that the other day. Subjectively though, it was no contest, the Jupiter Cosmos sound better than the LED biasing in this amp. Also, given that the bypassed cathode will lower the effective internal resistance of the input stage tube, relative to the LED, I expected to see improved bandwidth.
Here is an example. The bandwidth is now very impressive, -3dB from 7Hz to 34kHz. This is the benefit of using very well-made transformers.
I also measured the output impedances for the three taps:
120ohm tap: 41ohm output Z
32ohm tap: 8.5ohm output Z
8ohm tap: 2.7ohm output Z