dp96,
I know what it does in theory, but my ultimate question is:is it as cut and dry as its implied?. My findings says it not, gnfb as a whole is simple, apply said feedback and tailor compensation so that it never exceeds 180 degrees at unity , black and white. But very few delve into the amount of feedback across the band. I'm not into audio to design effects filters, though it's great from the recording end, but gnfb implies that the current drive output is inclusive, however if the open loop output stage distortion is low enough. has enough instantanious current drive on its own, and output impedence is sufficiently low, how does gnfb improve this situation?.
Colin
I know what it does in theory, but my ultimate question is:is it as cut and dry as its implied?. My findings says it not, gnfb as a whole is simple, apply said feedback and tailor compensation so that it never exceeds 180 degrees at unity , black and white. But very few delve into the amount of feedback across the band. I'm not into audio to design effects filters, though it's great from the recording end, but gnfb implies that the current drive output is inclusive, however if the open loop output stage distortion is low enough. has enough instantanious current drive on its own, and output impedence is sufficiently low, how does gnfb improve this situation?.
Colin
Last edited:
Implied by whom?
The theory is solid enough to allow supersonic aircraft, MRI machines, CNC machining, 3D precision printing, radio telescopes, and space probes to function. So, yes, in the hands of an actual engineer, it's cut and dry. The question is still what the goal is- accurate replication of the input signal or effects box? They're both valid goals but, unsurprisingly, require different design approaches.
If all that is sufficient, you're there. Applying feedback in that situation would make the output stage distortion even lower and the output impedance even lower. It cannot increase the output current drive.
As noted by several above: what's your goal? If you can achieve that without NFB and without making it more expensive, prone to drift and aging, and requiring several adjustments, go for it! Nobody forces you to use something you think you can do without!
But if you come with hand-waving and blabla why it is bad etc, we would very much like to know why that is the case.
Jan
Hi Jan !
Here is just one part of 800ohm / 3W FR loudspeakers from mine vintage LS collection , most of them I chop out from junked very old EI-Nis pure tube Blk./Wh. TV-sets .
Manufacturer was Akustika -Svrljig-Serbia , originally they was powered from simple SRPP /OTL low power tube amp , ant they sound amazingly good .
Once I made one quick experiment with four of these LS connected in series(3,2K tot.) and powered directly from 2A3 anode circuit ( GNFB free- OTL- 2W-SET) via one MKP condenser ( virtual parafeed connection) , that combination was spectacular in reproduced sound , unfortunately with too low SPL level for my taste .
Best Regards !
Attachments
After reading 364 posts about global negative feedback, output impedance and some mystical power paradigm i still don't have a clue what this is all about 😡
There is the amplifier - loudspeaker interface: by today standards the amplifier should be an ideal voltage source which can deliver as much current as needed by the loudspeaker.
Obviously this was not so in the dark age of audio. People were experimenting and power waste and CO2 was of no concern. Engineering knowledge was not so widespread and electronic components were precious too.
If the amplifier has 8 ohm of output impedance and the speaker has 8 ohm too, half of the power is wasted in the ouput impedance of the amplifier.
Maybe this is the strongest point against amplifiers with high output impedance.
This has nothing to do with feedback, may it be global or local. Feedback changes the output impedance but does not reduce power consumption!
Feedback can be used to design an amp with 8 ohm output impedance instead of
zero ohm too!
The BIG advantage of using feedback is that the ouput impedance and frequency response is DEFINED and independent of the big tolerances of transistors.
Standardisation is another strong point:
How can a loudspeaker manufacturer design his best loudspeaker if he know nothing about the amplifier interface?
Anyway if a loudspeaker designer thinks that he needs an amp with 4 ohm output impedance he can still add the 4 ohm in his loudspeaker. No problem here if
the amp is a voltage source.
Do you know of any loudspeaker which has an internal series resistor?
Once we have standardized the amplifier - loudspeaker interface we can go on and think about the amp.
What is a good amp?
There is the amplifier - loudspeaker interface: by today standards the amplifier should be an ideal voltage source which can deliver as much current as needed by the loudspeaker.
Obviously this was not so in the dark age of audio. People were experimenting and power waste and CO2 was of no concern. Engineering knowledge was not so widespread and electronic components were precious too.
If the amplifier has 8 ohm of output impedance and the speaker has 8 ohm too, half of the power is wasted in the ouput impedance of the amplifier.
Maybe this is the strongest point against amplifiers with high output impedance.
This has nothing to do with feedback, may it be global or local. Feedback changes the output impedance but does not reduce power consumption!
Feedback can be used to design an amp with 8 ohm output impedance instead of
zero ohm too!
The BIG advantage of using feedback is that the ouput impedance and frequency response is DEFINED and independent of the big tolerances of transistors.
Standardisation is another strong point:
How can a loudspeaker manufacturer design his best loudspeaker if he know nothing about the amplifier interface?
Anyway if a loudspeaker designer thinks that he needs an amp with 4 ohm output impedance he can still add the 4 ohm in his loudspeaker. No problem here if
the amp is a voltage source.
Do you know of any loudspeaker which has an internal series resistor?
Once we have standardized the amplifier - loudspeaker interface we can go on and think about the amp.
What is a good amp?
If the industry weren't standardized like it is today, then anyone who was into hi-fi would have to be a junior engineer in order to match components. In fact, that was what the hi-fi community was like until the early 50s; if you didn't know what you were doing you were stuck with tabletop radios and big living room consoles.
And putting a resistor in series with a woofer changes the bass tuning of the system. You can use this to your advantage; the caveat is of course obvious. And many commercial designs use resistors in the crossover to match driver efficiencies in the system; but none that I know of put a resistor in series with the woofer.
The old high Q, high efficiency speakers of yesteryear sound tinny and thin when driven by modern equipment. Modern speakers sound muddy and muffled when driven by pre-50s amplifiers. Try it.
And I want to point out again that no laws of physics were ever violated by any of these schemes, no matter how odd they may seem today. Voltage is voltage, and Thevenin and Norton theorems still apply.
Here's someone's take on the issue, directly at Bruno Putzeys:
Bruno's statement:
"Hi-fi review is a complete shambles. The few magazines that do measure are capable of reprinting
the most frightening distortion spectra from amplifiers and actually call them good. "Objectivity"
got downgraded from "independent of who's doing the observing" to "not favouring particular
brands". For me personally the affair hit rock bottom when in 2009 two reviewers, one Dutch, one
British, independently remarked of the same amplifier (a reasonably priced product with exemplary
performance) that it sounded surprisingly musical for an amp with such low distortion. In the 21st
century audio engineers build equipment while actively avoiding two of the most powerful tools
available to the whole of science and engineering: measurement and error control. The damage to
the audio industry and its reputation in the wider engineering world will remain immeasurable until
we decide to take control."
Third party comment:
"This tells me that Putzey's is either not aware of or is disregrding what has been found out about human perception and distortion. That the kind of distortion and its PATTERN is far more important than its level. Naturally above a certain level it is coloring the sound but when the pattern is wrong it also colors the sound even if it is THOUGHT that it should not.
Geddes found no significant correlation between THD and IMD and sound quality (in fact it was slightly negative but did not meet 95% confidence).
What he doesn't realize is that the small residual from amps with tons of feedback WILL color the sound in ways that much higher levels of harmonically correct distortion will not. Must be frustrating to him to be fighting with nature that doesn't conform to his "We make it perfect so it must sound perfect and why can't people hear that...dammit!"
Clearly there are many people who, despite the "theory", like ncore just fine....Comments?
Bruno's statement:
"Hi-fi review is a complete shambles. The few magazines that do measure are capable of reprinting
the most frightening distortion spectra from amplifiers and actually call them good. "Objectivity"
got downgraded from "independent of who's doing the observing" to "not favouring particular
brands". For me personally the affair hit rock bottom when in 2009 two reviewers, one Dutch, one
British, independently remarked of the same amplifier (a reasonably priced product with exemplary
performance) that it sounded surprisingly musical for an amp with such low distortion. In the 21st
century audio engineers build equipment while actively avoiding two of the most powerful tools
available to the whole of science and engineering: measurement and error control. The damage to
the audio industry and its reputation in the wider engineering world will remain immeasurable until
we decide to take control."
Third party comment:
"This tells me that Putzey's is either not aware of or is disregrding what has been found out about human perception and distortion. That the kind of distortion and its PATTERN is far more important than its level. Naturally above a certain level it is coloring the sound but when the pattern is wrong it also colors the sound even if it is THOUGHT that it should not.
Geddes found no significant correlation between THD and IMD and sound quality (in fact it was slightly negative but did not meet 95% confidence).
What he doesn't realize is that the small residual from amps with tons of feedback WILL color the sound in ways that much higher levels of harmonically correct distortion will not. Must be frustrating to him to be fighting with nature that doesn't conform to his "We make it perfect so it must sound perfect and why can't people hear that...dammit!"
Clearly there are many people who, despite the "theory", like ncore just fine....Comments?
There are people having their life and reputation depending on such crap. But then I beg to differ: there is ignorance (some deliberately choosing to stay so) and then there is swindle. Not always easy to distinguish.
It doesn't. If something is sufficient to achieve a goal then improving it does not improve the outcome. However, the situation you describe almost certainly arises from strong local feedback and there is nothing whatsoever fundamentally different between local feedback and global feedback.vynuhl.addict said:
That is an interesting view. I thought most people at least think about it. I certainly have. The maths says that the amount of feedback which matters is the amount of feedback at the frequency of the unwanted output. Most people know this and design accordingly.
This is a classic statement of the form "My significant distortion is more accurate to the input signal than your much smaller distortion". It is the standard fall-back of those who don't wish to admit that they prefer some (low order) distortion with their music. Another way of looking at it is "I don't understand electronic engineering but I live in a democracy so I feel free to criticise those who do understand EE".kuribo said:
You can get the most power efficient speakers by utilizing mechanical resonances.
This is still done in hearing aids today to get 120 dB SPL @ 1-2 kHz.
The downside is the bad frequency response.
If you do not like the bad frequency response the loudspeaker designer
has to dampen the resonancies with resistors.
This is a waste of power but will produce a straight line frequency response.
Maybe this is the secret behind the power paradigm?
With tube amps and output transformers it is not possible to get decent low output impedance.
The loudspeaker and the amp with its specific output impednace has to be designed as a pair
and can therefore not be exchanged randomly.
Maybe there was even a standard impedance in these old days?
But i doubt it as amps with more power would have lower output impedance.
Today the amp is not a critical component in the reproduction chain anymore, as long as the speaker power is well below the amp limit.
I doubt than distortion figures below 0.1% have any meaning or produce
any coloration.
When there is strong interest (and time) i setup a simulation to add distortion to a wave file in a user defined way.
Would be fun to see which level of distortion is the most pleasant in a blind test.
If you like you can send me some high quality musical pieces where you think that the effect of distortion could make a difference.
This is still done in hearing aids today to get 120 dB SPL @ 1-2 kHz.
The downside is the bad frequency response.
If you do not like the bad frequency response the loudspeaker designer
has to dampen the resonancies with resistors.
This is a waste of power but will produce a straight line frequency response.
Maybe this is the secret behind the power paradigm?
With tube amps and output transformers it is not possible to get decent low output impedance.
The loudspeaker and the amp with its specific output impednace has to be designed as a pair
and can therefore not be exchanged randomly.
Maybe there was even a standard impedance in these old days?
But i doubt it as amps with more power would have lower output impedance.
Today the amp is not a critical component in the reproduction chain anymore, as long as the speaker power is well below the amp limit.
I doubt than distortion figures below 0.1% have any meaning or produce
any coloration.
When there is strong interest (and time) i setup a simulation to add distortion to a wave file in a user defined way.
Would be fun to see which level of distortion is the most pleasant in a blind test.
If you like you can send me some high quality musical pieces where you think that the effect of distortion could make a difference.
Which always left me wondering why even competent engineers are quoting the loop gain @20KHz, a number which tells squat about the distortion reduction (of course, some extrapolation is possible, but that's beside the point). Quoting the loop gain @60KHz is a number of interest, since it shows how much feedback is available to correct the crossover distortions.
As a matter of interest how much gain is needed at 60KHz to deal with the crossover distortion?
Man that is funny. And it's very true.
There's been a few times that people (salesmen) have tried to run some bullhockey past me. I stop them dead in their tracks. I'm an engineer and I don't believe in pink unicorns.
Short answer: as much as possible.
Long answer: depends on your output stage configuration and design goals.
Hi Kuribo !
Little of topic but related to that yours Ncore-amps serviceability :
I studied structure of that Ncore amps for a while and must admit that yours Ncore amps are Pretty Sophisticated and very very in modern way designed units , but in the same time is Pretty Complicate electronics units too ,
- did you have some service manual supplied from manufacturer in case if those Ncore amps accidentally fail ?
- If your Ncore by some accident fails but your warranty is already expired Who will repair with original parts that broken amps for you then ? , Bruno ?
My MOSFET amp, I showed before, has the loop gain(simulated) 63 dB at 60kHz and 54 dB at 100 kHz, ULGF is 4.2 MHz. Is that good enough?
Design goal, best sound possible.
Sorry Dadod. That's not a design goal, that's 'let me horse around a bit and see where I get'. 😉
Jan
Sorry Jan, what is design goal when design an amp, not the best undistorted sound? Must be my bad English.
Damir