jFET amp with current NFB

[VladimirK]

I just have assembled a prototype of this amp, and the first listening, before the burning-in procedure, has made me happy. The sound is very-very life-like. Input volume control potentiometer is ALPS 10K (plan to change to DACT). Although I realize that jFET based schematics is rather individual, need specific adjustments, but the final result justifies all the efforts.
Apart of this, I was curious about jFET output stage, is there any listenable effect from its implementation.
For achieving reasonable power, I have assembled a cluster-transistor from 8 pcs of paralleled RF power jFETS KP903A. I enclose the Id vs Vgs curve for this cluster transistor.
In this schematics, an optimum idle current appeared to be 2,4 A.
Since total transconductance is still less than 1S, I had to use some NFB, current NFB via R4, R3 resistors. All measures have been taken for assuring small length of NFB path, and the R4 is non-inductive type for RF applications.
Voltage at the source of VT11 was adjusted by selecting R2-2. Thermal stability of this voltage is good. The schematics shown is “as build”, it is absolutely stable and gives good square wave without any additional correction capacitors.
Now the amp is burning-in, I’ll post some measured characteristics later on.
The only things I have checked out are: clipping is symmetric and soft, 10V RMS undistorted output sine wave, pass band -1dB is around 200kHz.

[ingenieus]

Very interesting. It's always nice to see something new.


You seem to have left out the component values for the diodes and LEDs.

[VladimirK]

Originally Posted by ingenieus

You seem to have left out the component values for the diodes and LEDs.

The Zeners used are 5,6V 1,5W, I believe that at this voltage they have the best voltage-to-noise ratio.
The LEDs are red ones, 1,8V forward voltage drop.
As one can mention, I try to avoid using of electrolytic caps everythere it is possible. But, for the output cap, I use well tested approach - 1000uF low-ESR (or Elna Silmic or Cerafine) cluster cap, plus 100uF decent polypropylele cap (connected first to the schematics). So, lytics shunt the poly at low frequencies.

[VladimirK]

Here I post measurements of output impedance of the amp. Measurements have been done by Immittance Measurement Instrument, at various frequencies from 25Hz to 100kHz. The instrument show Impedance Modulus Ztot, as well as phase angle, resistive Zres and reactive Zreac components of the vector impedance. It is supposed
Ztot**2 = Zres**2 + Zreac**2.
Zreac is negative if it is capacitive type, and positive if inductive type.
Ztot is almost equal to Zres at 100Hz...10kHz frequency range (reactive component is negligible). At frequencies below 100Hz the capacitive impedance component starts to rise, and above 10kHz the inductive component starts to rise.

[apassgear]

Innovation as its best.
Thanks for sharing Vladimir.

Tony

[VladimirK]

Originally Posted by apassgear

Innovation as its best.
Thanks for sharing Vladimir.

Tony

Thanks, Tony
I am so happy with what I have got from this schematics, that I started similar version with 4,4A idle current. Just curious, is it possible to preserve the same perfect sound with higher power version. As usual, I can find definite compromises.

[AKSA]

Vladimir,

Original thinking, unique work. Much similarity with tube circuits, but iron replaced with current sources! Thoughtful, expressive work clearly showing you have considered all the psychoacoustic issues as well as the technical ones.....

Congratulations!

Hugh

[Lazy Cat]

Vladimir, hats-off for inovative design, practice proven I believe, hopefully sounding good.

One suggestion, how about to exchange an eight 2P903A with a single SJEP120R063 fet. Usually one device behave more controlled than eight and there is no parameters matching needed.

[VladimirK]

Originally Posted by AKSA

Vladimir,

Original thinking, unique work. Much similarity with tube circuits, but iron replaced with current sources! Thoughtful, expressive work clearly showing you have considered all the psychoacoustic issues as well as the technical ones.....

Congratulations!

Hugh

Yes, Hugh, I was inspired by a competition with good tube amp, and formulated my task as to win against tubes at their field - in the midrange clarity and sweetness. I confess it is possible, with an additional benefit - bass responce from this transistor schematics can perfectly handle the PMC EB1i speakers (better than No NFB schematics and not worse than Yamaha B2-x amp). The sound is really "big" and extremely sweet and transparent. I guess, it will be the same with modern SITs at the output stage.
The very key points - shunt power supplies and good current sources "everywere".

[VladimirK]

Originally Posted by Lazy Cat

Vladimir, hats-off for inovative design, practice proven I believe, hopefully sounding good.

One suggestion, how about to exchange an eight 2P903A with a single SJEP120R063 fet. Usually one device behave more controlled than eight and there is no parameters matching needed.

Yes, it is a pain to get a lot of 2P903A and match them. However, to substitute them, I would propose 2SK60, or RF MOSFET RD100HHF1, or probably recent Nelson's SIT.
Looking at the SJEP120R063 datasheet, it resembles ordinary MOSFET, to me 2SK1530 would be better.
This SiC transistor has the only benefit - high dissipation power.
With 8 pcs of 2P903A I have total Ciss near 140pF, while SJEP120R063 has 1220pF, and far from linear Id vs Vgs around 2-3A Id current. Second stage in this schematics, will need 20...40mA idle current, instead of 5...6mA with 2P903A.