"Phil Allison" wrote in message
...
"Alex Pogossov"
In another thread Phil Allison commended Quad 405.
Its typical schematic is he
http://schematic.narod.ru/Audio/semi/Amp/Quad405-2.gif
What is the purpose of diodes D5 and D6 connected in series?
But seeing as you asked, here is my take: D5 and D6 ensure output
transistor TR9 remains biased off at idle and at low output current while
permitting some current flow through R38 into the load.
=== Alex:
Yes, at idle pre-driver Tr8 will be biased to conducting about 15mA. Power
transistor Tr10 will not even be conducting -- R33 is too small.
But why did they decide to get the output stage run in class A at very low
levels? Probably to reduce distortion at low levels. Running at least Tr8
allows the bootstrap C10R31 to function increasing open-loop gain. Probably
this amp would sound screechy if at low levels the outpud would have been
completely cut off. There is in fact a cross-over kink. Below output stage
operation the feed-through bridge resistor attenuates the signal 6 times
(with 8R load). Plus on top of that the bootstrap not working would made
things even worse.
Facts to keep in mind about the 405:
1. The output transistors operate in pure class B, ie with zero bias
current.
=== Alex:
I think one silicon diode or perhaps even one silicon + one germanium
between the bases of Tr8 and Tr9 would not hurt. The size of the kink would
be reduced while still staying in class B without any risk of thermal
runaway.
2. TR7, the "class A" stage transistor drives the upper output transistor
TR9 direct while TR8 is the driver for the lower one, TR10.
=== Alex:
I noticed that, but I believe another "darlingtonising" transistor in front
of Tr9 would make load on the class A stage more symmetrical, would increase
loop gain on positive waves and overall reduce distortion.
3. The bridge network allows the class A stage to drive the load direct
when the output transistors are biased off, so virtually eliminating
crossover distortion.
=== Alex:
As I said, the crossover exists: at low levels below the output stage
activation the lop gain is at least 6 times smaller. It causes kinks and
distortion which then is marvelously almost (but not completely) compensated
by the bridge. In a conventional class AB output stages we often have kinks
the other way -- at low levels the gain is higher (closer to unity) because
both (idle current taming) emitter resistors work in parallel (class A),
while at high levels -- one at a time.
4. It really worked - Quad were able to use slow, rugged NPN output
transistors in the original 405, have no bias adjustment or thermal
compensation for same and yet get vanishingly low THD and crossover
distortion.
=== Alex:
Indeed, in those days it was quite revolutionary.