Note that if the amplifier, in real life, breaks into parasitic oscillation, SPICE sims of the amp won't always show that happening. I found out the hard way.
It's a good idea to view the Bode plot for the amp to get a bead on its stability.
Try feeding an input signal to the model that would ask for an output voltage higher than the specified rail voltage(s). Unless LT spice updated to correct for that, you could swing the output to the high heavens in spite of a small rail voltage.
The sound of the amp is quite impressive... sometimes I feel it can bet my Yamaha receiver... hahaha...
(The circuit from #4) How does that even work? There's no DC path to the base of Q4, so it'll be switched off with only leakage currents flowing I think?
Low distortion usually involves a current-mirror load to the input pair, and an emitter follower added to the VAS, both exemplified in the Doug Self "Blameless Amplifier" design(s).
I'd go for a phase margin of at least 60 degrees as the gain approaches unity at high frequencies.
I'm afraid its not working 🙁 You can see that from the output amplitudes. Nothing there really. No amplitude of signal.
There is a lot wrong with the basic configuration unfortunately. Try this, all with default models. I'll let you tidy it up and add suitable output stage bias.
There is a lot wrong with the basic configuration unfortunately. Try this, all with default models. I'll let you tidy it up and add suitable output stage bias.
Load the amp with an 8- or 4-Ohm resistor; drive it into clipping; plot V(out)*I(load). Start with plotting V(out) and edit it by right clicking the plot label. Left click and drag to the peak to measure the power trace. The rms power will be half the clipping power. Expect he positive and negative clip to be slightly different. The power may be "negative" if I(load) is backwards, flip the load resistor.
Simulating clipping is also an important step to detect rail sticking and stability problems.
Simulating clipping is also an important step to detect rail sticking and stability problems.
Attachments
Apologies but I'm a bit curious, and you are in the simulation so perhaps it's easy - I was wondering what it would do if that connection was removed 🙂
The idea is that the driver section uses the feedback loop, but the output followers just chat to the speaker without being pulled about by the feedback loop.
Attachments
The driver Q8 requires current through D3 from the speaker to drive U2. U2 is not a follower and requires feedback from Q8 to find the correct current. What you suggest is possible with a fully complementary output NPN+PNP, but there will be distortion in the "followers" so it's not useful. Perhaps you are thinking about current dumping, but that involves a tricky mix of both stages.