I have my first results for the small signal pentode E83F ..
I'm running it in triode mode with a B+ of 280V regulated, dc filtered heaters, remote psu 2m away
I have a plate resistor of 13K5 at present and a variable bias resistor.
Also, 470K grid leak resistor, 2K2 grid stopper and 2K2 screen stopper.
Driving an edcor xsm 10K-600 output transformer terminated with 2K2 across the secondary
It's running a plate current of around 6mA and a screen current of around 1.3mA for a total cathode current of 7.3mA - less than half the rated power or so.
....
SO ... varying the bias to +4.1V gave me the lowest THD with a nominal 1Vpp input signal, sine 1KHz
THD was very low (no nfb) at 1.15% nearly all H2. Hum floor super low -120dBu. Hum margin -119.35dB
note - this was due to excessive LF rolloff - now corrected for -113dBu hum floor
Which is (still) s really good - especially so given I haven't yet got it humming at it's best .... I need to vary B+ and the plate resistor
The input and output transformers together are adding around 0.1% of thd ....
....
This small signal pentode (2.1W plate) has a high gm at 9000 mA/V and it shows in the very low noise.
Of course, the remote, regulated psu is really important - this one (eby kabota hv reg pcb) is great - easy to vary the Vreg with a trimmer and no issues of any kind so far. I've standardised on these, so a few more on the way!
If I tried this with an 'amp head' type build with just choke-cap filtering it would be at least 12 db worse hum floor, all the time .. bottom line - a case of significantly higher B+ ripple *plus* likely em-induced hum into the signal transformers *and* ground lines ... from the psu.
And anything less than a 47K plate load resistor - in a similar circuit in an 'amp head' - is asking for hum.
....
Gain is reasonable, even in triode mode with this low-ish plate loading. I'm still tweaking on that, so I'll wait for more data!
I definitely haven't noticed any sensitivity or undue microphonics - rock stable hum floor.
Even with my odd-wooden-blocks 'wobbly' test rig

But the wiring is solid .. I mean, the psu is 'way away' !
I haven't even optimised grounds or anything yet, and there's a few things to route better - but this is mostly affecting the 'mids'.
ed. (
so far! - I have to check I'm not 'band -limiting' the bottom end with that quite small coupling cap .. till tomorrow, it's a possibility 
Update : sure enough, my low freq extension *was* being curtailed by too small of a coupling cap from the plate to the edcor xsm 10K-600.
I checked the freq resp of a range of values ... I'm settled on 4.7uF ... a big wima
I now get 20Hz at -2dB or so and 22KHz is less than -1dB - so the response is now good!
My noise floor is now -113.4dBu with +28.3dB gain ... (worsened 6dB from previous low-rolled-off 'prelim' measurements)
....
Early indications are showing this tube performs better than anything yet I've tested, and by a *modest* margin.
Margin over hum is +116dB - previous best (JJ 5751) was 105dB margin with similar distortion ... in an 'amp head' type build
[discounting the 12 dB of worsened hum margin with 'amp head' vs 'remote psu' .. brings things (re. prev best) to within 1dB of today's result - very nice -- but until I do the 'head off' testing between the 5751/6922 historical standouts and newcomers, who knows wth ? ]
When I push it to 3% THD, the hum margin rises to 122dB. Thats for a 2Vpp input, x26 gain, 52Vpp at the plate.
It all amounts to 'less sensitive' than a 5751 stage .. with super low noise .. sparse harm.dist. spectra, albeit with slightly increased thd ... with 'that diff' made up of .. increased h2 .. arguably a good thing 8)
To the tune of only 1dB and maybe 0.3% thd diffs at the end of the day, over the prev. personal best ..
Still it is significant, especially given my 'prototyping' build qualities

... and in it's way, rewarding the right. efforts
....
As to the audible diffs, not so much, really - although I haven't yet 'listened' ... that's another thing altogether.
Ask someone with good hearing!
...
But all very interesting to figure, no doubt. If this world truly is a 'sim' wth does it mean to chase 'tube stuff' seven decades after it's heyday ? Who knows.
....
This tube shines as an single-ended output tube, no doubt .. I suspected as such, given the obvious construction quality and the high gm.
And it performs pretty well as a first-stage tube too! Plus anything in between, I would say.
The onset of distortion should be gradual ... this tube likes to be biased somewhat cool ... +4.1V bias for 1Vpp input signal showed a thd minima.
A good thing too (cooler bias) .. this tube likes to run a decent amount of current (>10mA) given half a chance.
The challenge is to reduce the currents whilst increasing gain without losing the thd or it's spectra .. along the way.
.. it's a 5-way tweaze where the hours fall away as you try to find your current 'best'. (and take a snapshot!)
So I'll spend some time tweaking in triode mode for max gain, min current etc .... and after all that ... with a really high quality output transformer. Much better a one , than I should hope for.
Then continue with some pentode tests - I think they will be very good results.
Finally, I plan to try the 'enhanced triode' or 'screen drive' mode .... where the grid g1 is grounded and signal is applied to grid g2. This tube has a reasonably high gain[g2g1] which is a parameter of interest for doing this.