Looking for the bestest UcD buffer

[IVX]

Hah! Some error in the my tube buffer setup is detected- mosfet current is too small for the 2Kohm load! I'm sorry. Probably this design is not so bad yet, simulation with the same error show 0.025% vs 0.037% achieved by me in saturday! So i hope new measurement will be close to the audigy THD limits! Coming soon..

[Jan-Peter]

IVX,

Nice work!

Regards,

Jan-Peter

[IVX]

Well, now is much better. About 3Vp-p output voltage, but load is only ~20Kom, because mosfet current still too small (i'm delayed for my day job for 1.5 hour again :-).
Edited:
each anodes current are 2.8mA, and unbalance isn't so hard affected now, PS voltage 130V. BTW, this tube i've bought for 0.75 euro, probably it is will be equal cheap sound??? Audigy have own THD=0.002%, so i can't resist to subtract it from 0.0059%..

[ackcheng]

Very nice!

Congratulations

[IVX]

Thanks guys, however i haven't tried 2Kohm load so far, and probably it's wouldn't be so lucky attempt, there is moot point because mosfet gate charge inherently nonlinear. Seems that not each of the spice model note it? When i'll have a time enough (next weekend if noone have an idea about), 2Kohm load will be checked with irf7103, bs108 and irlml2803, whatever the outcome i'll do it by complicating follower, that isn't preferable yet.
Actually any tube it's a most linear active device around, so our congratulations is rather be adressed to the tube inventors, but not for me.

[IVX]

As i'd promised, the new measurement is done. FFT plot captured at next conditions- 6Vp-p@2Kohm load@2.48mA each anode@16mA ir7103 source current. At the 8.2Vp-p output THD about 0.016%, IRLML2803TR (SOT23) allowed better THD performance vs bs108 or ir7103, but too hot, at least in the P2P setup. THD at >1Khz a bit less(-10%-20%). Warm settling time about 2 min, in this time from 0.008% to 0.02% can be seen.. Cathode resistors was 17k one vs 12k other!! BTW, i'd measured only one output of the circuit.. Now we can subtract 0.002%(audigy selfTHD) from the presented THD figures, and conclude that we haven't chance to obtain such figures without fine adjusting of the anode's resistors. 0.05% in the best case and if happy day, on the other hand, who of us equipped even less then i am? After all it must be listened!
edited:
BTW, Koldby has suggested servo circuit to keep close to the zero voltage at outputs.

[IVX]

Well, i know that's probably too late to being interesting, but i've done discrete opamp, measured it, and we'll listen it soon. Short form: 140V/Us, 300ns settling, 70db open loop gain with pole about 45khz so THD/Freq plot will be flat up to 45khz also.
Slowly but smoothly Tube buffer PCB is finished today too. Short form: DC servo (<4hz), delayed outputs connection (relay), onboard 150v (anode U) and +/-15V regulators.

[subwo1]

Ivan, you are the greatest.

[IVX]

Nice joke, subwo!THX!

[subwo1]