Classic Amp Updated Board IV

Just when you (meaning I) thought it couldn't get any smaller or cleaner, I spy another opportunity for optimization.

• I moved Q6 & Q7 in line with the drivers (Q10 & Q11), which shortened the traces considerably, and made it possible to shrink the whole design once again.
• The size of the board being determined by the big filter caps, I decided to drop two of them, and increase the value of the remaining ones. Four 2200uf are now three 3300uf (which actually increased the total from 8800uf to 9900uf per side). The Panasonic data sheet says 3300uf fit in this footprint.
• I also added a power ground pad next to each power input pad. Both ground lines will go back to the transformer center tap, but this way, the ripple current will go to the associated filter caps, and not have to flow across the ground plane. 
• While I was moving grounds around, I decided to move the speaker ground where it would be centered on the board, so the current flow would be more symmetrical through the ground plane. 

Great Grandson of Updated PCB - Click to Enlarge

Anyone want to wager on how many more iterations there are?

Updated Schematic - Click to Enlarge

One final thought: There might be a bit of ripple at Q12 & Q13, or R14 & R15. I don't think this will be a problem, since the output voltage will be governed by Q10 & Q11. Any perturbations in the supply to the output devices would be compensated by the local feedback inherent in the Sziklai configuration. Still, I would have preferred to have the raw rectifier output hit a capacitor before it hits my output transistors. Even with the big fat power supply rail, there is still some resistance between P+ and C6, et. al. For this reason, there might be some justification for a bit of off-board filtration. I'm just sayin'.