danlavry wrote on Wed, 25 May 2005 17:01 |
"I was quite clear in stating that:
At this point I am not arguing the merits of a few stage's with localized feedback in each stage against the "opamp approach" which is one overall feedback path. I am just trying to point out, for starters, that a lot of people in audio believe that negative feedback is an undesirable factor, but the circuits used ALL HAVE NEGATIVE FEEDBACK, and plenty of it!
It would be good of you to help me make it clear to the readers, that the statement "No negative feedback" is fundamentally incorrect. It should be substituted by "different circuit implementations of negative feedback". No negative feedback means no resistors in the emitters or sources. I did not see your circuit, but if you took out all the negative feedback mechanisms (including current feedback), what would happen to the performance?
|
HI Dan,
Yes, technically you are correct.
My understanding is that if really correct terminology is
used, then it is impossible to make any amplification circuit
without negative feedback (unless it is atransformer).
I will clarify; even a single triode that has the cathode
grounded (through a capacitor) and voltage input will have a
finite OP impedance due to the tubes internal negative
feedback. Dave Collins sent me a nice technical paper on this
subject (thanks DC).
So I believe there are only various levels of feedback.
The most common is global feedback. A proportion of the OP is
sent back to the IP but out of phase. This is probably the most
lamented form of feedback. Possibly the claim has grounds due
to it being the most invasive circuit wise. Global feedback
needs to have some kind of control on the open loop gain to
prevent oscillation due to phase shift, at some frequency
the signal sent back to the IP will be + phase instead of -
because it has had sufficient phase shift through the device.
I will not get into phase margins, dominant pole, miller
compensation etc... probably too techo for the purpose of this
discussion.
Next down the line is interstage feedback. This works on the
same principle but the OP is only sent back out of phase
between two adjecent stages as opposed to a whole opamp
(usually 3 stages or more).
A classic example of interstage FB is the CFP transistor pair
(complimnetary feedback pair) where a simple 2 transistor
arrangement is such that the OP of one device is fed back to
linearise the first device. These interstage FB arrangements
work very well and can linearise a simple circuit by a factor of
10x to 50x even more.
Next down the line again is a single stage that has
degeneration. This is called local feedback. For a simple
transistor, a resistance is added to the emmitter which has
the effect of making the base to emitter voltage more linear
WRT current and temperature.
I suppose next down the line is a gain stage that has no
degeneration. Generally speaking, this is not workable
with bipolar transistors because they are too non linear,
somewhat workable with fets and quite workable with tubes,
especially triodes.
In fact some triodes show quite incredible linearity with no
degeneration. The 6SN7 comes to mind... from memory it can do
leass than 0.1% THD with 10V RMS and no degeneration.
Getting back to terminology, generally when people speak of
zero feedback it will mean no global or interstage feedback.
But usually some amount of degeneration is used which is
technically local feedback.
Also people will refer to OP emmitter followers as zero
feedback but technically they have 100% local feedback.
Most opemps or gain stages can use combinations of all the
above. Power amplifiers are a good example.
There are many other ways of linearisation, probably too much
to go into here; cascoding, bootstrapping, feedforward
correction (such as Hawksfords OP error correction) and more.
Quote: |
I heard a well know power amplifier maker claims no negative feedeback. All he does is to "convert input voltage wave to a current wave, and steer the current into a high value resistor to build a lot of voltage...". Well, how do you make a low distortion voltage to current conversion? A circuit with negative feedback...
|
Exacty... and it becomes an argument of terminology.
The voltage needs to be buffered numerous times before being
able to drive a speaker which is all emmitter follower type
circuits. See above.
Quote: |
So that is my first objection to the "no feedback" crowed. That "no feedback" and "feedback is bad" should be changed to "feedback is what saves us from the inherit non linearity of cheap pieces of silicone or tubes".
|
The real argument is within various types of FB. When I used
our amp gain stage as an example I was careful to say no global
or intestage FB. This is what is comonly referred to as "zero
feedback" but as can be seen here it is technically a misnomer
and complete threads have been wasted purely trying to establish
a correct terminology for "zero feedback" We don't need that here.
Cheers,
Terry