Front Stage oscillating problem

HI guys,

I'm building a front stage for a mosfet amplifier and i'm having trouble stabilizing it.

First i use POT 1 to set 4mA current through J3.
After that i use POT 2 to set the output DC offset to 0. It works ok till i'll approach 30mv at the OUT ( 0V IN). At that point the system oscillates at 2-3 MHZ at an amplitude of 30mV.

For the moment i use a test breadboard.
What i have already tried out:
  • have shielded the feedback trace and the one that goes to M11's gate.
  • have lower the gain to 4.7,
  • put an 10nF capacitor over R15
  • tried both differential and single input

any other ideas?


  • driver-stage-bjt-cascode.pdf
    26.5 KB · Views: 124
  • IMG_7161.jpg
    1 MB · Views: 260
Last edited:


Disabled Account
2015-12-16 2:00 pm
At this stage it is a two stage amplifier to end up as three. You must have a stage which will carry the dominant pole . Generally, the VAS is assigned by adding a Miller capacitor from drain to gate . First the input differential 's pole needs to pushed as high as possible for high NFB . Reduce the feedback resistors by 10 fold . Apply a capacitor grounding the gate of J2 , that you can brave it to be HF filter . You can add a resistor on the collector of Q4 same vale as pott2 in order to apply a series RC between collectors .
Today i had some time to work on the prototype again. And i started to try your suggestions from top to bottom.

So, first i checked again that the amp does not oscillate without feedback just to confirm what rayma asked me to check first time. To my surprise i found that it does oscillate ; it just that without feedback the offset voltage is around 55V DC, so the first time i checked i failed to see the other 30mV AC on top of it .

So the Op Amp went out the next second, and things get better immediately.

The offset still wonder 10 to 20 mV in either direction but i'm pretty sure this will stop when i'll connect it to earth or put it in a solid aluminium case.

i say this because the offset start wabelling if i move my hand above the board or in the immediate vicinity and it stops when i put one finger on the power supply ground).

thank you very much all!
i have attached the final scheme that tested ok.


  • driver-stage-bjt-cascode - no opamp.pdf
    26.3 KB · Views: 61
FYI, I had comparable issues during the design of my latest amp.
I remember, increasing VAS degeneration helped a lot in solving the stability issues.

I also did compare the 2sk216/2sj79 with the irf9610/710 combo. The irf sounded better, no matter what I changed regarding bias, degeneration...

Yesterday I had another audio lover visiting me, he really was impressed by the sound... Goosebumps

More info somewhere here:
Yes, the suggested method of CCS as above is very stable with temperature changing. Note that Q1 & Q2 should be matched Vgs to reduce offset.
You can try adding Cc to M11. Phase lead compensation maybe is not enough in this case due IFN146 is a high gain Gm JFET. A combination of phase lead compensation and ordinary Miller compensation usually more stable.

Cuong Nguyen