I'm interested in how you find those 1620 valves stack up. Glancing at the data they seem pretty much the same as our EF37a valves from the same era
I did this test some time back, but it's worth another look:-
These were only single tube tests but the most important thing to notice is the ratio between the plate current and the g2 current. This is because it directly affects the level of partition noise.
The 1620/6J7 tube's ratio was pretty good for such an early date and the later tubes gradually increased the g2 wire spacing to reach a maximum with the EF86. I think the EF37A has a very high internal impedance rp so it draws the least current of all.
I think it's telling that RCA and others, carried on using the 6J7/1620 long after the 6SJ7 was introduced.
On page 510 of RDH4 there is a comparison of intermodulation distortion between a 6J7 and 6SJ7, guess which is best!
DaveP