Complementary JFET-JFET cascode input, BJT VAS-Drivers, Lateral Output
Somewhat high performance amplifier.
Complementary input with 2SK170/2SJ74 cascoded by 2SK246/2SJ103.
About 4 mA through each 2Sk170/2SJ74, 8 mA through R2.
Darlington VAS'es with 2SC3503/2SA1381. 9 mA though the VAS section.
Driver section with 10 mA through 2SC3503/2SA1381 pre-drivers and about 50 mA through MJE15032/MJE15033 drivers.
Output consists of 4 pairs of 2SK1058/2Sj162 Lateral FET's biased at 150 mA each.
Curerntly only done in simulation but the results are good enough for me to start building it soon, once I get the layout done.
The tests below were done WITH the input filter and 470 uF cap in the NFB network.
THD-1 and THD-20 simulation results. 8 Ohm load.
1 W : 0.0000375%
30 W : 0.000234%
60 W : 0.000391%
90 W : 0.000495%
1 W : 0.000556%
30 W : 0.005390%
60 W : 0.006296%
90 W : 0.020645%
Attached images shows output frequency response and squarewave performance plots at 1 KHz and 20 KHz.
Gain is about 26.6 dB
Frequency response plot shows 26.2 dB at 1 Hz and a high frequency -3 dB point of 195 KHz.
Below tests were done WITHOUT the input filter and the 470 uF cap in the NFB network.
Attached are Loop gain and Slewrate measurement plots.
Loop gain plot shows about 69 degrees phase margin and about 10 dB gain margin.
Slewrate measurement plot shows a slewrate of about 60V/500nS, which scaled to 1us gives a slewrate of about 120 V/uS.
Nice, but do you need the double emiter follower if you are using a FET output stage? You should be able to drop this to a simple emitter follower buffer between the VAS and the FET's.
Your loop gain plot looks a bit optimistic and in practice you may run into instability issues - I would shoot for a unity gain cross over frequency of around 1-2MHz. You can accomplish this by providing more degeneration on the JFETS, or increasing Cdom - but this latter appproach will cause your slrew rate to decrease unless you increase the LTP current.
The 10pF comp fap across your feedback resistor also looks a bit on the high side. This value should be set during development of the physical prototype, and it needs to be tweaked to give the optimum gain margin. For a feedback resistor values of 5k, this typically means about a 5-10pF capacitor.
You may want to consider using TMC instead of the miller comp currently shown in your designs - this will get your distortion down another 2x or more.
Good luck with your project.
Single EF driver stage is more than enough for lateral FETs, looks like BJT amp is adapted for laterals. :D
A single emitter follower loaded the VAS too much. Going from a single emitter follower to double emitter follower reduced the load on the VAS from 750uA to about 70-80uA. Which resulted in a nice reduction of THD-20.
Distortion wise I am more than happy at the current level, TMC really isnt needed.
So it is NOT a BJT amplifier adapted to FET's. I started out with wanting to design a FET output amp.
Not my fault that it ended up looking like a BJT amp with FET outputs. ;)
I'll take them into consideration.
Just out of curiosity; have you tried or considered a FET driver stage instead of the 3EF?
You are running the drivers very hot I see, personaly I would have there about 20 tot 30mA. Have you seen an improvement somewhere with increasing the bias throught the drivers?
Driver current is something that could be adjusted. 20 mA would be just as reasonable.
I will be trying a FET driver stage tonight or sometime tomorrow.
If it comes close in performance I might go that way.
My goal is less than 0.01% THD-20 at 60 W into an 8 Ohm load as a minimum.
|All times are GMT. The time now is 11:44 AM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright ©1999-2017 diyAudio