Current mirrors as collector loads in a symmetric amp

[steph_tsf]

John Linsley Hood
Audio amp symmetry
Wireless World Jan 1985 p.31

What is the practical implementation of the John Linsley Hood idea about using current mirrors as collector loads in a symmetric amp ?

Additional comment from smoking-amp :
From post #77 in Unstable VAS current in amp from Slone book
R8 and R9 provide the VAS bias feedback in a convenient manner. R8 and R9 only need to feedback DC info, so can put caps from Q9,Q12 ends of R8,R9 to respective rails. Now can put another 100 Ohm resistor in each emitter circuit (Q9,Q12) to act as conventional degeneration.

[Joachim Gerhard]

I do not understand. You can build it they way it is or like smokin-amp is saying.
I build circuits like that many times but mostly with Mosfets at the output. It sounds excellent. It can also be configured as a buffer. The Schlotzaur buffer is very similar but single ended. I build symmetrical versions of the Schlotzaur but they are so damn fast that i could make them only stable with a gain of 6dB. You can also use bipolar LTPs at the input and you can also use this stage inverted. It is very versatile.

[jcx]

I simmed a VAS current servo approach just for the fun of it

optoisolator VAS bias for Comp Diff?

I really don't see the reason for dual complementary diff pair inputs except as "eye candy" - it really doesn't do much for the circuit's operation compared to spending the extra Q elsewhere - see Bob's MOSFET amp diffpair VAS, then consider bootstrap cascoding the (single) input diff pair

[steph_tsf]

Am I right, the class AB current bias Q18 Q19 in the output devices, and the VAS Q9 Q10 idle current are all defined by the optocoupler U1 ? And no emitter resistors for Q18 Q19 ? Sound weird ...
How do you implement the differential pair bias ? The current source I2 doesn't sound possible because of the reverse voltage.
Steph

[jcx]

the circuit does contain some "cute" spice hacks, the ideal spice ccs works just fine as shown, in real hardware you will need the usual ccs from each supply

the output bias depends on the Vbe multiplier - in this instance the LED Vf is also part of the Vref - but the LED Vf like the Q11 Vbe are "shunt regulator references" and not too sensitive to the VAS current

Q18,19 would need emitter R in a real amp - I didn't need them to show the common mode VAS current control loop function

likewise feedback would never be tapped off load - but works fine in sim

[steph_tsf]

Hi jcx, now better I understand. Thanks. Interesting approach. But why oh why such an elaborate crcuit ? Have you tried the Linsley Hood proposal with a total capacitive decoupling ?