I'll just add my 2 cents, though ksc pretty much laid out my feelings. The 12bh7a at the proper resistance is not as bright as the e80cc, and definitely has the same level of imaging and dynamics, which was lacking from 12au7s. I couldn't handle the obvious sibilance in lesser recordings that came with the e80cc + T1s. The e80cc matched the 650s very well though.
I just finished installing a 3 way switch to let me run the 12bh7a, e80cc, or 12au7s, so I can try to do a direct A/B with the former 2 tubes soon.
On another note, the MJE5731As are in the mail, too! I've been looking at the 5687 pinout, and here are the differences I see between that and the 12bh7/au7 pinout:
12au/bh7
1 - Plate (2)
2 - Grid (2)
3 - Cathode (2)
4 - Heater (2)
5 - Heater (1)
6 - Plate (1)
7 - Grid (1)
8 - Cathode (1)
9 - Heater (2/Midtap)
5687
1 - Plate (2)
2 - Grid (2)
3 - Cathode (2)
4 - Heater (2)
5 - Heater (1)
6 - Cathode (1) [Redirected from stock pin 8]
7 - Grid (1)
8 - Heater (Midtap) [Redirected from stock pin 9]
9 - Plate (1) [Redirected from stock pin 6]
So pin 8 must go to 6, 9 must go to 8, and 6 must go to 9.
I realize I may have displayed that poorly, but does it seem to be correct? If these are the only changes necessary for the 5687, couldn't these three leads go through some 2-way on-on switches, with the biasing LED on it's own switch between pins 6 & 8? It seems previous 5687 modders used blue LEDs to achieve a different bias, which I'm guessing would eliminate 12au7s and the lot for tube rolling - but it seems the 5687 could run just fine with the stock HLMP6000s? Sorry for all the questions!