As a side project I'm designing a little headphone amplifier, using discrete components. The design is pretty straightforward - it runs off a single supply and is single ended, consisting of 2 voltage amplification stages and a output buffer. In the simulation it looks good, distortion appears to be very low across the frequencies of interest and the frequency response is where I'd like it to be.
The problem is that when the output level is pushed to clipping at certain frequencies, say around 1-5 kHz, the amplifier breaks into high frequency oscillation. In the sim the oscillation "rides" on top of the clipped output signal, at a level of about 0.75V p2p of oscillation for around 10V p2p of output. The amplifier does use substantial negative feedback. I don't intend on listening to the amplifier when it's clipping, of course, but I would think a well designed circuit should at least behave itself better when it occurs.
I'd appreciate any guesses as to what could cause this sort of issue. Thanks!