Hi all...
Since Ron.Id has posted the schematics here, I might as well explain it a bit...
Yes, probably it's not optimal, but given the purpose of the design is:
"To build a multiloop amp, with parts available only on local stores around me"
Now, if you live in my country, well, to get a good op-amp is really difficult, the same also with buffer chip like the popular BUF634 is almost impossible to get in local stores around, the best I can find is only LM4562 and OPA2134...
Yes we can get parts from internet, but I usually just use what available around me...
So why does the L/R have two stages, with the 2nd stage just acting as follower and yes, it's nowhere more efficient in terms of current...
Because to use the multiloop topology we need buffer (or in this case follower), and since I need bigger current output to drive low impedance headphones, so a quick hack is just to parallel the 2nd stage L/R (the follower) with another op-amp
~ a l a A47
So why don't I just use single op-amp and follow it with discrete diamond buffer? Since it's easy to get general purpose BJTs around?
Simple, because I want it to be able to swap buffer (as long as it works inside the feedback loop)
first the use of dual op-amp (for both stage) on the L/R makes it easy for me to just use the U1 DIP8 socket as an "expansion" socket, just like Ron.Id do with his Diamond buffer module which is plugged in directly to the U1 socket
and second, because the U2 op-amp is in parallel, whenever there's an "expansion" module inserted to U1 socket, I can just ignore it completely by not plugging any op-amp on the U2 socket...
And the last, the use of INA134 as rail splitter, it's just simply because I can't get TLE2426 around, and this is the only chip I found that states inside the application note that it can be use as a virtual ground...
I hope it explains well about the schematics
Regards...