amplifier oscilates in LTspice

perfectly fine (sorry for the shorter runtime here)
1730579049159.png
 
No, R16 should simply be moved to the emitter. It is then the most symmetrical arrangement possible, with the exception of doing that and then adding a Baxandall diode in series with Q6 emitter. A CFP with Q6 emitter tied to Q8 collector, then both run thru the resistor is the “right” way to do a CFP, but it is less symmetrical than the other proposal. Do them that way if they are complementary, of course.
 
No, R16 should simply be moved to the emitter. It is then the most symmetrical arrangement possible, with the exception of doing that and then adding a Baxandall diode in series with Q6 emitter. A CFP with Q6 emitter tied to Q8 collector, then both run thru the resistor is the “right” way to do a CFP, but it is less symmetrical than the other proposal. Do them that way if they are complementary, of course.
i will do that too but could i replace the op amp with something like the lm393?
 
Like that, Drive the diode string center, with Bootstraps which gets max swing from opamp. P1 adjust bias
Dont change the currents with wild guessing. Using modern 2n3055 models.
5534QUASI.JPG


Square wave 20 kHz 8 ohm No ripple, dam good slew for 3055. More than stable
1730635940031.png
Gain 25dB 2 Hz to 1.6 MHz
1730636121211.png
10watts 8 ohm
1kHz .006% distortion

10watts 4 ohm
1kHz .006% distortion

Max Clean Power 4 ohm 15.76 Watts
1kHz .007% Distortion
20 kHz .06% Distortion
1730637027766.png1730637255237.png


10 kHz 5volt P 10volt PP 90% and 10% 4 volt
.287 us 8 volts 27 V/us Slew rate
1730638899068.png
1730638175994.png
 

Attachments

  • 1730636861926.png
    1730636861926.png
    248.6 KB · Views: 40
Complementary pair is Way The Hell Better. The PNP side won’t do annoying unexpected unpredictable things like oscillate. Someone playing with 2N3055’s might not have modern ones.

And then you will always get the emitter resistor in the right place.

Also, my experience is that 2N3055’s, even modern ones can demand more than 200 mA of base current. 2N2222/MPS2907 solve that little problem nicely.
 
it works it seems that the bucherott cell/zobel network creates these socilations
2.2R is a very low value for a Zobel network / Becherott cell load - usually ~10 ohms is used. Have you tried such a value?

The normal reason for instability is a lack of phase margin due to too many RC stages in series within the global loop (each active device can provide the equivalent of an RC stage), and a simple if crude fix is to provide a bit more feedback from the opamp output direct to its inverting input via a small capacitor, i.e. a "Cdom". The downside to this is the feedback loop becomes less effective at linearizing due to the greater fall off of open-loop gain at higher frequencies.

Another approach is use faster transistors (or FETs) so that the phase delays are smaller in the loop.