PCB Layout feedback needed

Hi guys, i would like to get some feedbacks about my Layout, based on the attached schematic.
Thanks in advance

Immagine 2022-04-07 123841.png
Immagine 2022-04-07 123904.png
 
The feedback take-off points should be from the output pad (not some random point on the high current trace between emitter resistors, as that trace carries half-cycle pulses and its non-linear IR voltages are significant). IE directly from the 10ohm 3W.

Otherwise you've introduced a certain amount of unavoidable distortion that cannot be fixed.

Here's a diagram:

feedback_takeoff.jpg


And also the ground reference used by the feedback circuitry has to come from the speaker return ground pad, not somewhere else in the ground wiring.

This is basically just star-grounding applied to the output and feedback paths of the amp.
 
The feedback take-off points should be from the output pad (not some random point on the high current trace between emitter resistors, as that trace carries half-cycle pulses and its non-linear IR voltages are significant). IE directly from the 10ohm 3W.

Otherwise you've introduced a certain amount of unavoidable distortion that cannot be fixed.

Here's a diagram:

View attachment 1042234

And also the ground reference used by the feedback circuitry has to come from the speaker return ground pad, not somewhere else in the ground wiring.

This is basically just star-grounding applied to the output and feedback paths of the amp.
I always star ground.
Keep feedback tracks short as possible. 1mm of track = 1nH of inductance.
Dont mix power supply and audio ground except at supply centre and then only once.

Ok, i did some modifications:

Immagine 2022-04-08 110123.png
 
i would not tee-in the zobel to the decoupling caps tracks.
+1
I'd also beef up the grounding trace for the Zobel network. It likely won't see a lot of current unless the amp oscillates, but a wider trace will have lower inductance than a thinner trace of the same length, so you'll get a better Zobel network that way.
also consider buss and star-buss layouts.
Or a ground plane. The challenge with a plane is that you have to be careful in how you design it to prevent 'dirty' currents and 'clean' currents from interfering with each other. But you could put a ground plane on the top layer (assuming the blue is the bottom layer) and use it only for the signal/'clean' ground.

It may be worthwhile to put the driver transistors an Vbe multiplier between the output devices. The goal here being to keep the loop area of the bias spreader and drivers to a minimum as it tends to be a sensitive node.

Tom