FB lowers output impedance of an amplifier, too. This is useful for tube amps. Transistor amps have way too much open loop gain to not use feedback to tame the gain down.
All class A which I sim have monotonic harmonic profile.
Distortion can have pleasing sound or not pleasing, it depend of the kind of distortion.
Not only global feedback but whole design it self. I sim same amplifier with different compensation and it have different profile of distortion. In the implementation, some passive component can add distortion and change the profile.
Just to make sure we are all on the same page, when we say "X level of feedback", "X amount of feedback", "X db of feedback", "X dB of GNFB", etc, what we really mean is loop gain, correct?
Well, physical occurrences cannot be canceled.
It’s all about attenuation. Damping is practically divided into electrical, mechanical and acoustical, but since things happen at an atomic / molecular level, it is fundamentally electrical. Magnetic fields ought to be seen as pure, ungoverned, high velocity acceleration, resistive electric forces holding it in restraint. As energy level rises, rapidly depleting the resistive forces, distortion relentlessly increases regardless of what stupid simulations indicate. In typical systems, there is a glaring scarcity of resistive forces from an initial stage.
It’s all about attenuation. Damping is practically divided into electrical, mechanical and acoustical, but since things happen at an atomic / molecular level, it is fundamentally electrical. Magnetic fields ought to be seen as pure, ungoverned, high velocity acceleration, resistive electric forces holding it in restraint. As energy level rises, rapidly depleting the resistive forces, distortion relentlessly increases regardless of what stupid simulations indicate. In typical systems, there is a glaring scarcity of resistive forces from an initial stage.
Last edited:
Normally not. Normally, the amount of feedback is the difference between open loop gain and closed loop gain.
Let us say an amp has 70dB of open loop gain at some frequency. Then you close the feedback loop so the closed loop gain is 28dB. That amp then is said to have 42dB feedback at that frequency.
Since open loop gain varies with frequency and closed loop gain almost not, the amount of feedback varies with frequency.
Jan
Last edited:
Thats exactly right Damir.
I get the feeling that Jan misread my note and thought I said Open loop gain rather than Loop gain.
Here is the math:
If the forward path has gain of AOL (open loop gain), the feedback path has a gain of f, and hence the closed loop gain ACL is 1/f, then:
Loop gain = AOL x f => Loop gain in dB = 20log(AOL) + 20log(f) = 20log(AOL) - 20log(1/f) = 20log(AOL) - 20log(ACL)
Which is what Jan said: "Normally, the amount of feedback is the difference between open loop gain and closed loop gain."
I get the feeling that Jan misread my note and thought I said Open loop gain rather than Loop gain.
Here is the math:
If the forward path has gain of AOL (open loop gain), the feedback path has a gain of f, and hence the closed loop gain ACL is 1/f, then:
Loop gain = AOL x f => Loop gain in dB = 20log(AOL) + 20log(f) = 20log(AOL) - 20log(1/f) = 20log(AOL) - 20log(ACL)
Which is what Jan said: "Normally, the amount of feedback is the difference between open loop gain and closed loop gain."
atmasphere, can you please cite the paper/page/section where Putzeys makes blanket statements about "'bad area' of feedback" especially "between 4dB up to about 35dB"?
In "The F-word" page 15 he says that from 20dB or 30dB you get an "unambiguous net improvement that goes on forever".
Before that he writes that the first 10dB or 20 dB you get a tradeoff of reducing the 2nd harmonic (where he also confirms what I wrote about the audibility of 2nd order harmonics and the resulting IMD) and slightly increasing the level of higher harmonics.
Though the overall level of the harmonics of his constructed example with >10% THD should be noted.
These are still only rules of thumb, as the effects depend entirely on the inherent linearity of the design.
In "The F-word" page 15 he says that from 20dB or 30dB you get an "unambiguous net improvement that goes on forever".
Before that he writes that the first 10dB or 20 dB you get a tradeoff of reducing the 2nd harmonic (where he also confirms what I wrote about the audibility of 2nd order harmonics and the resulting IMD) and slightly increasing the level of higher harmonics.
Though the overall level of the harmonics of his constructed example with >10% THD should be noted.
These are still only rules of thumb, as the effects depend entirely on the inherent linearity of the design.
Last edited:
Bob Cordell did a similar test and described it in his book. If you start out with a reasonable linear open loop amp, say <1% distortion, adding feedback does NOT show an area where some harmonics increase. In such a case, all distortions decrease from 1dB feedback all the way till the end. There are some wiggles in the curves but none of the harmonics actually rises.
The article where this rising harmonics with increasing feedback comes from is an old article by Baxandall (or Walker, can't remember exactly who), but that was a single FET stage with horrible distortion to begin with. That DID show increasing harmonics with feedback, but that was really a very unrealistic example.
Jan
Negative feedback can change its status, and become positive feedback, leading to sustained unwanted oscillations at frequences which damage speakers and output stages. At every transistor stage, driving current and output voltage do not oscillate in phase with a cumulative effect. The cumulative phase change increases with frequency as more reactive impedances become more dominant. The latter are prevented from becoming dominant by making the amplifier's global gain decrease with frequency with the use of a shunt capacitor across the base and collector in the VAS.
Now, a word for those arguing that negative feedback affects speaker performance.
As they claim a moving coil in a magnetic field, generates a voltage across the coil which is in parallel with the output. However, this voltage cannot affect the output stage and the negative feedback network, as the voltage from the output stage is the driving force, while the induced emf in the voice coil is the reactive force. In short, the status of a reactive force is not an active one but merely reactive.
Now, a word for those arguing that negative feedback affects speaker performance.
As they claim a moving coil in a magnetic field, generates a voltage across the coil which is in parallel with the output. However, this voltage cannot affect the output stage and the negative feedback network, as the voltage from the output stage is the driving force, while the induced emf in the voice coil is the reactive force. In short, the status of a reactive force is not an active one but merely reactive.
Well, the argument is, the EMF generates a voltage that causes a current through the circuit consisting of the voice and the amplifier output. That current thus must necessarily flow through the amp output.
I think that is a valid argument as such.
For the amp it doesn't make a difference. It is a voltage amp, and the output is driven to a certain level, by the feedback, of the input level x gain. And as a result, a current flows which depends on what hangs off the amp output. The amp couldn't care less whatever, as long as the requested load current is within its linear capabilities.
That is also why you can 'measure' output impedance by sending a current into the output and measure the voltage that it generates.
Jan
I think that is a valid argument as such.
For the amp it doesn't make a difference. It is a voltage amp, and the output is driven to a certain level, by the feedback, of the input level x gain. And as a result, a current flows which depends on what hangs off the amp output. The amp couldn't care less whatever, as long as the requested load current is within its linear capabilities.
That is also why you can 'measure' output impedance by sending a current into the output and measure the voltage that it generates.
Jan
His Fig 10 is what I referred to above. But remember: this is a single JFET stage, horribly distorting to begin with! This is not representative to a reasonable linear open loop amp with feedback added.
Jan
And as he said "I performed my own version of the experiment, using a power Mosfet in a single-ended Class A gain stage driving 1 watt into an 8 ohm load:" Fig 11.
It's an interesting article and he highlights the importance of IMD, the "elephant on the dance floor" 😀
It's an interesting article and he highlights the importance of IMD, the "elephant on the dance floor" 😀
Last edited:
Induced EMF is effectively in series, not parallel, otherwise it couldn't have an effect on the current in a voltage-driven coil.
