[Chris J]

 
Quote:

 
 
Impedance matching applies to all frequencies including audio frequencies though the reasons for it are different when dealing with audio frequencies as opposed to radio frequencies. You will always get the greatest power possable when impedances match, even with transistors that is if they didn't burn up first WHICH I GUARANTEE THEY WILL . Tubes on the other hand WILL NOT BURN UP GUARANTEED into a matched load
 
 
You first statement is not true for either transistors or tubes. Transistors if you tried to match the output impedance you would burn up the transistor because of too much current otherwise it would also be fact for transistors that they would deliver thier best power into a matched impedance. The transistor would try to deliver too much power, much more than it could handle in a matched impedance scenerio. Practicalities prevent us from matching the output impedance in transistor amps due to the output impedance being so low.

In Solid State Power Amps you don't match the output impedance to the load impedance. if you did then output impedance would = 8 ohms when driving an 8 ohm loudspeaker.
In virtually all Solid State Power Amps output impedance is approx. 0.1 ohm or lower. Reference:  Ben Duncan, Douglas Self power amp design textbooks.
Solid State Power Amps are not designed according to Maximum Power Transfer Theorem.
They are designed for maximum electrical efficiency and optimum voltage bridging.  Ben Duncan calls it a "current capable enough voltage source".
The output transistors burn up when load impedance is too low, for example, a short circuit.
 
 
 
 

 

[germanium]

Quote:

 

 
The output transistors burn up when load impedance is too low, for example, a short circuit.
 
 
 
 

My point exactly You don't because you can't. The design paradigm has changed. with tubes doing so was a must to get best performance before adding feedback, with transisters you can't
 

 

[Chris J]

Quote:

My point exactly You don't because you can't. The design paradigm has changed. with tubes doing so was a must to get best performance before adding feedback, with transisters you can't
 

Honestly, I am confused by that statement.
No offense intended, I'm just confused.

 

 

[xnor]

Quote:

My point exactly You don't because you can't. The design paradigm has changed. with tubes doing so was a must to get best performance before adding feedback, with transisters you can't

 
This makes no sense. You certainly can increase the output impedance with solid state amps, but only few do because it makes a lot of things worse. If anything, it's tubes that are the limiting factor.
 
 

 

[firev1]

With tubes, using NFB plays only a minor role in reducing THD as compared to the solid state design, where NFB is needed and is vital in producing a low distortion amp. Btw reading the last few pages, I don't think NFB increases bandwidth? In fact, the reason why tubes play badly with NFB is because many designers had NFB factors too high without consideration for the limited bandwidth of the transformer coupled tube circuit. Lastly, even if tubes are indeed more efficient, the fact they run in Class A would mean they waste a lot more power than today's modern amp designs.
 
As for on Dr Otala's work, here is what Douglas Self has to say of it,"Another objective was the elimination of so-called transient intermodulation distortion,
which after many years of argument and futile debate has at last been accepted to mean nothing more than
old-fashioned slew-rate limiting." - Audio Engineering Explained.
 
For his low feedback design, Dr Otala made use of heavy local feedback to reduce the overall feedback, so it isn't very accurate to call it a zero feedback amp. On another note, Otala use of 20 db global feedback  across the frequency band for his prototype so the Otala amplifier that was eventually produced added 10 db more feedback for 30db across the frequency band.

 

[qusp]

not sure whos posts you are referring to with the statement of NFB increasing bandwidth because I certainly didnt say it, it lowers it normally, I was under the impression thats one of the reasons it increases stability in some amps
 
maybe someone abbreviated No Feedback as NFB and you took the wrong meaning?

 

[Chris J]

Quote:

With tubes, using NFB plays only a minor role in reducing THD as compared to the solid state design, where NFB is needed and is vital in producing a low distortion amp. Btw reading the last few pages, I don't think NFB increases bandwidth? In fact, the reason why tubes play badly with NFB is because many designers had NFB factors too high without consideration for the limited bandwidth of the transformer coupled tube circuit. Lastly, even if tubes are indeed more efficient, the fact they run in Class A would mean they waste a lot more power than today's modern amp designs.
 
As for on Dr Otala's work, here is what Douglas Self has to say of it,"Another objective was the elimination of so-called transient intermodulation distortion,
which after many years of argument and futile debate has at last been accepted to mean nothing more than
old-fashioned slew-rate limiting." - Audio Engineering Explained.
 
For his low feedback design, Dr Otala made use of heavy local feedback to reduce the overall feedback, so it isn't very accurate to call it a zero feedback amp. On another note, Otala use of 20 db global feedback  across the frequency band for his prototype so the Otala amplifier that was eventually produced added 10 db more feedback for 30db across the frequency band.

Tubes and Negative Feedback, the short answer is, it depends on a bunch of factors:
Yes I agree, when you take into account the effect of the output transformer, then the bandwidth is reduced and usually must be reduced to keep the amp stable, i.e stop it from oscillating.
In addition, because most tube are more linear than most transistors, hence, normally tube amps don't need as much feedback. In addition, tubes amps are generally simpler, i.e, designed with less open loop gain.    Complicated question! So I'll stick with "it depends"
 
You certainly didn't read me saying that tubes are more efficient than transistors.  They all need a heater element which blows any hopes of low efficiency out of the water and out the window.  Sounds like you read sumthin' out of context?  Or maybe not? Which is cool,  this thread is getting a bit looooong.   I'm sure you are aware that output stage of most tube power amps do not run in Class A. 
But I agree, whether they run in Class A or not, they are less efficient than "equivalent" transistor amps.  
Personally, I don't hate Class A so I have no axe to grind.
 
Yes,  I am aware that more than a few designers question Otala's work.
If you read Douglas Self's book on amp design he actually uses a few techniques that Otala suggested!  i.e. local feedback in the power amp input stage.
They may disagree with his results, but his work continues to be extremely influential in modern power amp design. For example, high output current amps.
He seems to be the prophet that a few designers love to hate.

You certainly didn't read me saying the Otata based designs were zero feeback.
You are probably thinking about the Eletrocompant amps, they were designed with his input, as I understand it, but were not actually designed by him.  Maybe someone knows something I don't know?
Otala wrote many papers on TIM and reducing TIM.
He never said:  "use this much feedback".  He only made suggestions.  You could argue he suggested using feedback in moderation.  I can send you some of his papers if you like??
 
Quote:

not sure whos posts you are referring to with the statement of NFB increasing bandwidth because I certainly didnt say it, it lowers it normally, I was under the impression thats one of the reasons it increases stability in some amps
 
maybe someone abbreviated No Feedback as NFB and you took the wrong meaning?

Feedback and bandwidth:  the answer is "it depends".
For example, if you take an existing power amp with global feedback and remove the Global feedback (assuming you can!), you will see the bandwdth decrease and the gain increase.  This is called Open Loop gain and Open Loop Bandwidth.
My statement is a bit misleading,  when you design an amplifier you design it with certain goals and constraints in mind.
It's a bit disingenious to say OK, now I will remove the feedback from an amp design to run with 37.5 dB of Negative feedback (for example).  May as well remove the turbo charger from an engine designed to work with a turbo charger and expect it to be a fantastic non-turbo charged engine.  It ain't gonna happen.  You are changing the design intent and will now have to re-design the engine or the amp.
Try running an Op Amp without feedback, the bandwidth will drop like a stone.  Take a look at any Op Amp spec sheet.
 
Negative Feedback also stabilizes gain.  etc. etc.
 
references:
http://en.wikipedia.org/wiki/Negative_feedback_amplifier
Feedback:  Fred Waldhauer.
 
I would argue that if you like the sound of an amp and it drives your speakers properly then I wouldn't spend a lot of time looking under the hood or pondering how much feedback it has.  If it's an excellent amp then the designer probably used the right amount of Global and Local Negative Feedback, etc, etc, etc.
 

 

[germanium]

Quote:

Tubes and Negative Feedback, the short answer is, it depends on a bunch of factors:
Yes I agree, when you take into account the effect of the output transformer, then the bandwidth is reduced and usually must be reduced to keep the amp stable, i.e stop it from oscillating.
In addition, because most tube are more linear than most transistors, hence, normally tube amps don't need as much feedback. In addition, tubes amps are generally simpler, i.e, designed with less open loop gain.    Complicated question! So I'll stick with "it depends"
 
You certainly didn't read me saying that tubes are more efficient than transistors.  They all need a heater element which blows any hopes of low efficiency out of the water and out the window.  Sounds like you read sumthin' out of context?  Or maybe not? Which is cool,  this thread is getting a bit looooong.   I'm sure you are aware that output stage of most tube power amps do not run in Class A. 
But I agree, whether they run in Class A or not, they are less efficient than "equivalent" transistor amps.  
Personally, I don't hate Class A so I have no axe to grind.
 
Yes,  I am aware that more than a few designers question Otala's work.
If you read Douglas Self's book on amp design he actually uses a few techniques that Otala suggested!  i.e. local feedback in the power amp input stage.
They may disagree with his results, but his work continues to be extremely influential in modern power amp design. For example, high output current amps.
He seems to be the prophet that a few designers love to hate.

You certainly didn't read me saying the Otata based designs were zero feeback.
You are probably thinking about the Eletrocompant amps, they were designed with his input, as I understand it, but were not actually designed by him.  Maybe someone knows something I don't know?
Otala wrote many papers on TIM and reducing TIM.
He never said:  "use this much feedback".  He only made suggestions.  You could argue he suggested using feedback in moderation.  I can send you some of his papers if you like??
 

Feedback and bandwidth:  the answer is "it depends".
For example, if you take an existing power amp with global feedback and remove the Global feedback (assuming you can!), you will see the bandwdth decrease and the gain increase.  This is called Open Loop gain and Open Loop Bandwidth.
My statement is a bit misleading,  when you design an amplifier you design it with certain goals and constraints in mind.
It's a bit disingenious to say OK, now I will remove the feedback from an amp design to run with 37.5 dB of Negative feedback (for example).  May as well remove the turbo charger from an engine designed to work with a turbo charger and expect it to be a fantastic non-turbo charged engine.  It ain't gonna happen.  You are changing the design intent and will now have to re-design the engine or the amp.
Try running an Op Amp without feedback, the bandwidth will drop like a stone.  Take a look at any Op Amp spec sheet.
 
Negative Feedback also stabilizes gain.  etc. etc.
 
references:
http://en.wikipedia.org/wiki/Negative_feedback_amplifier
Feedback:  Fred Waldhauer.
 
I would argue that if you like the sound of an amp and it drives your speakers properly then I wouldn't spend a lot of time looking under the hood or pondering how much feedback it has.  If it's an excellent amp then the designer probably used the right amount of Global and Local Negative Feedback, etc, etc, etc.
 

Note I never claimed tubes were more power efficient only that 0 feedback tubes with thier matched output impedance do the best job of delivering the power they send to the speakers & actually given the right circumstance sound the best (speakers designed with matched output impedance in mind). Such amps do not nessessarily sound loosy goosy in the bass or rolled off at the top with the proper speaker setup..
 
 A 0 feedback amp with flat frequency response is far supperior to an amp that has tons of feedback to flatten an otherwise very poor frequeny response & yes 0 feedback amps can have very flat frequency response if designed properly. My friend amp is flat from 10Hz to 30KHz at full power. At normal listening volume in a normal room with average efficency speakers that response extends to 500Khz. Far beyond human hearing.Heck 30KHz is beyond most human hearing This frequency response is not a result of the gain of the amp being twisted to get flat response, it is flat from the get go. The power response of said amp is also very near flat across a wide range of load impedance. Note I said Power & Not Voltage response. Speakers are power driven devices in general taking both voltage & current to move them.
 
Put a tube in the place of an opamp (with all the proper circuitry) & you will find that they have excellent distortion & frequency response without feedback even when run single ended in this low output setup. They will also sound more open in a lot of cases compared to an opamp used in the same location. The tube will be running right in the heart of it's most linear range in this application.
 

 

[xnor]

Quote:

Originally Posted by germanium /img/forum/go_quote.gif
 
My friend amp is flat from 10Hz to 30KHz at full power. At normal listening volume in a normal room with average efficency speakers that response extends to 500Khz. Far beyond human hearing.Heck 30KHz is beyond most human hearing This frequency response is not a result of the gain of the amp being twisted to get flat response, it is flat from the get go. The power response of said amp is also very near flat across a wide range of load impedance. Note I said Power & Not Voltage response.

So the amp is both flat in voltage and power response? What kind of magical device is that?
Did you use speakers as load to measure this? My guess is it is highly non-flat with speakers as load..
 
 
Quote:

Speakers are power driven devices in general taking both voltage & current to move them.

X_x

 

[Chris J]

Quote:

Note I never claimed tubes were more power efficient only that 0 feedback tubes with thier matched output impedance do the best job of delivering the power they send to the speakers & actually given the right circumstance sound the best (speakers designed with matched output impedance in mind). Such amps do not nessessarily sound loosy goosy in the bass or rolled off at the top with the proper speaker setup..
 

The efficiency thing......you must have me mixed up with someone else.
I agree, you never said tube amps were more power efficient than transistors.
However, I apologize for any confusion.
Although our opinions differ on one or two things, I'm not trying to start WW 3
 

 

[germanium]

Quote:

So the amp is both flat in voltage and power response? What kind of magical device is that?
Did you use speakers as load to measure this? My guess is it is highly non-flat with speakers as load..
 
 
X_x

Flat in voltage response into a resistive load & flat into an  varying impedance load in the power domain. It is not majical. I'm speaking of 2 different type loads & 2 different domains that we are measuring for.It is resonably flat in the power domain with loads from 4ohms to 32ohms on an 8 ohm tap & from 8ohms to 64 ohms on the 16ohm tap. These are the common impedances found in such speakers as well.
 

 

[Mauricio]

I have no horse in this debate, but I am puzzled by a few of the terms used here.  During my EE studies I never heard--or needed to refer to--terms like the "power domain" or "current domain".  Time domain, yes.  Frequency domain, yes.  The use of "power domain" and "current domain" is a reflection of either ignorance or attempts to obfuscate.

 

[firev1]

 
I think I mixed up on the efficiency thing my bad. X_X
 
To qusp: I think Chris covered it for me
 
Quote:

 A 0 feedback amp with flat frequency response is far supperior to an amp that has tons of feedback to flatten an otherwise very poor frequeny response & yes 0 feedback amps can have very flat frequency response if designed properly. My friend amp is flat from 10Hz to 30KHz at full power. At normal listening volume in a normal room with average efficency speakers that response extends to 500Khz. Far beyond human hearing.Heck 30KHz is beyond most human hearing This frequency response is not a result of the gain of the amp being twisted to get flat response, it is flat from the get go. The power response of said amp is also very near flat across a wide range of load impedance. Note I said Power & Not Voltage response. Speakers are power driven devices in general taking both voltage & current to move them.

I assume you have some graphs(with the appropriate watermarks) for these "magical" claims.
 
On other note, I'm very curious about Otala's work, since currently I'm studying Douglas works exclusively(school library) for now. 
 

 

[Chris J]

Quote:

I have no horse in this debate, but I am puzzled by a few of the terms used here.  During my EE studies I never heard--or needed to refer to--terms like the "power domain" or "current domain".  Time domain, yes.  Frequency domain, yes.  The use of "power domain" and "current domain" is a reflection of either ignorance or attempts to obfuscate.

Are you an EE or an EE student..........just curious..............that's all.

 
I agree, if you have a virtually zero output impedance then you will get virtually flat voltage output into a speaker load.
Since most speaker impedances vary WRT frequency then you will not get flat power response, but if you are getting flat voltage drive into a loudspeaker who cares about the power?
 
I assume Germanium was referring to FLAT ACOUSTIC POWER RESPONSE IN THE ROOM???   
 


Quote:

 
I think I mixed up on the efficiency thing my bad. X_X
 
To qusp: I think Chris covered it for me
 
I assume you have some graphs(with the appropriate watermarks) for these "magical" claims.
 
On other note, I'm very curious about Otala's work, since currently I'm studying Douglas works exclusively(school library) for now. 
 

And as for you..........LOL.............are you also an EE student, sir?

I'm glad I got your back, Jack.
Excuse my dry sense of humour please.

I've read Self's "Power Amp Design Handbook", the movie wasn't as good as the book.
If you would like one or two or a whole bunch of Otala's papers, send me a PM.
I'll be glad to send them to anyone who asks.
Hs work was very groundbreaking, if nothing else he got everyone talking and thinking about TIM and high output current.
For only that, all modern power amp designers at least owe him a "tip of the hat".
 

 

[germanium]

This is what I mean by tubes having a flat power response
 
http://www.atma-sphere.com/Resources/Common_Amplifier_Myths.php
 
This is true whether you use feedback or not on a tube amp. The feedback may lower the output impedance & flatten voltage response of a tube amp but it still will not make much of any difference in the ultimate power it can deliver to the load at any given impedance at full power I.E. clipping. That is a function of the plate resistance. It will just be more load sensitive just like a transistor amp delivering 1/2 power into the load when you double the load impedance for the same volume setting. Whithout feedback the power will remain near constant as in it will remain most likely within 1 db of the power that was delivered into the lower impedance load instead of loosing 3 db for the same volume setting. This same amp can deliver flat voltage response into a resistive load but not flat voltage into a reactive load. It will however deliver flatter power into the reactive load than a transistor amp will as the transistor amp will loose 1/2 power each time you double the impedance