If you can audition it, that should settle the case for you.
47 responses Add your response
Audiophiles are often fond of attributing all sonic traits of a component to a single issue. It's rather analogous to a car buff assigning everything about an engine's performance to cubic inches.
It is almost never that simple for either audio components or cars. Negative feed back is just like anything else. It can help or it can hurt. It depends on implementation and how it is integrated with all other technical aspects of a design.
That said, I don't know anything about Unison amps, so take a listen if you're truly interested.
Some of the best measuring and worst sounding amps ever made had massive amounts of negative feedback. The Japanese amps of the 70s are the best examples of this, .0001 distortion. It is not that distortion makes an amp sound good, it is what you have to do to reduce it to tiny levels that makes it sound bad. While negative feedback may reduce measured distortion its correlation with sound quality is another matter. A good amp can have no global feedback or a moderate amount. Mine has none and has a considerably higher level of distortion than the vanishing level of distortion amps of the 70s but is a far better sound amp than they were. But intelligent design is everything; we are still searching [after all these years] for either design principals or measurements which will guarantee good sound. Peter Walker , of Quad fame, said years ago that he could build an amp that would look good on every test but that would be unable to play a recognizable tune. Neither design principals nor simple measurements is really going to tell you what an amp sounds like in a given situation. If distortion were the overriding factor then no one would ever buy a tube amp, obviously there are other factors at work.
I have two sets of amps that use zero feed back (Atmasphere, Electra Print), one amp that uses minimal feedback (VAC), and another amp that uses moderate feedback (Music Reference). I use and love the sound of all of them in my system.
I suppose we might also clarify whether the discussion is about global feedback, local feedback, or both. While we're on the topic, when a company markets their product as a zero feedback design what are they really referring to? It's been my understanding that this generally refers to global feedback only. I had an interesting conversation with Roger Modjeski on this once. It's a topic you don't want to get him started on unless you have the time and knowledge to keep up with him.
"...Negative feed back is just like anything else. It can help or it can hurt. It depends on implementation and how it is integrated with all other technical aspects of a design. "
Absolutely true !!!!!!!!!!!!
Still, I wish to add that NF is dangerous weapon and many nmanufacturers simply don;t know how to use it and it took decades to learn its positive and negative contributions to the sound.
My favorite amplifier and I own it, Spectron bases its design on "control theory" where they treat amplifier as a "control system". They do not use large amount of negative feedback but theirs - about 10 times faster then typical amplifier...and they can control speaker with the load of 0.1 Ohm...and reproduce music, particualrly dynamic peaks more realistically then any other amp I owned or auditioned. For hard rock - ain;t better.
Please refrain from making bold statement like that. Please don't take it the wrong way but Spectron can not drive the 1 ohm Scintillas though tough load 0.8ohm from 20hz to about 2khz and then hovering around 1.5 ohm up 20khz. That is a tough load but still far cry from 0.1ohm. The Krell master reference is probably the only amplifier which can handle such load cause now we're so close to a dead short. Below is the real testimony that Spectron cannot drive 1ohm load, namely the Apogee Scintillas. Again, I meant no offense. Peace
If the amplifier has other design considerations, it will not need to have feedback to be low distortion. For example, a fully-differential amplifier will have even-order cancellation at each stage in the amp, so it will, without feedback, generate only the 3rd harmonic.
Just an example.
Now, the flip side of the coin is what happens to the amp if a feedback is used. The problem is that with any amplifier, there is a time delay for the signal to move from input to output. This propagation delay causes the feedback signal to arrive slightly behind the actual signal at the input. Now with sine waves, the amplifier can reduce its distortion after succeeding iterations of the waveform. It does not do so with non-repetitive waveforms, like real music. It has also been shown that feedback, due to the time delay, actually **increases** certain distortions, namely the 5th 7th and 9th harmonics.
Now the increase is slight, but there is a problem: the human ear uses these harmonics to determine how loud a sound is. So if you mess with these harmonics, the electronics will have an artificial loudness about them, a tonality, which audiophiles describe as bright, harsh, brittle, etc.
In a nutshell adding feedback decreases certain distortions that the ear does not care about a lot (we hear them as 'warmth', 'bloom', etc.) while **increasing** the distortions that the ear cares about a lot.
Some designers pay attention to this and others don't. The subject has been controversial for over 50 years. For more information see
Finally, Chaos Theory has something to say about this, confirmed by Norman Crowhurst several decades before Chaos Theory became generally accepted: The addition of negative feedback to an amplifier results in a chaotic behaviour with both stable and unstable states. Additional harmonics and also inharmonic information is added (Chaos Theory calls distortion 'bifurcation'). This results in a noise floor quite unlike normal hiss- a noise floor that the human ear cannot hear into (whereas we can hear 20 db into hiss) that effectively masks information below the noise floor.
For those interested, Wikipedia has a nice primer on Chaos Theory: http://en.wikipedia.org/wiki/Chaos_theory
Perhaps I've brought this up before, if I have please forgive me. As I understand it great deals of negative feedback are often used in Class D amplifiers. One poster has already commented that since it is done at much greater speeds, the lag is inconsequential. I've wondered with digital designs such as the Tact, if might be possible to use some sort of, for lack of a better description, "digital fly wheel" to synchronize the feed back with the original signal. If such were the case then feedback might used to it's advantage with out the disadvantages often attributed to it, with the final signal funneled out. Any thoughts?
The faster the amplifier, the less this is a problem! However, faster is not the same as 'real time'. I would say that this is in the realm of 'progress' where we have an improvement over something that was done 15 or 30 years ago.
I think you can see though that if the input and feedback were synchronized, that the delay of the circuit is going to have to go somewhere. You might need a substantial memory for that. The faster you do the scan, the more memory you are going to need...
I think that H2Oaudio has the full right to participate as a "private user". The reason is that he is not a designer but a person who buys mass production B&W Icepower module and installs them in his chassis (I bet makes some modifications too) and place the label "H2O" on it. Incidently ihe same is with Bel Canto, Wyred4Sound, D-Sonic and bazzilion of others buyers of ICEpower modules.
The single exception, IMO, is Jeff Rowland who while uses B&W modules yet developed his own highly sophisticated power supply and with PFC at it. My hat off to him!
Athmosphere designed his equipment form A to Z and his post is 100% accurate. He is particualrly right in: " It has also been shown that feedback, due to the time delay, actually **increases** certain distortions, namely the 5th 7th and 9th harmonics."
In his 2nd post he acknowldge that "The faster the amplifier, the less this is a problem!" He correctly noted that "... faster is not the same as 'real time' " and he is right here as well - otherwise Spectron for example would list its distortion level as zero which they are not (but very small and even order of magnitude smaller in their monoblocks but its different mechanism in force).
Also, you are mistaken, not all class D amplifiers use the same fast negative feedback as you believe. Spectron explains its method on its web-site but, for example, Bruno Putzeys (formerly of Philips) uses exactly the opposite approach and his powerful class D amplifiers (under "Kharma" name) while expensive at $25k - $30k, are said to be excellent.
Regarding what H2Oaudio wrote in relation to the capabilities of Spectron amplifiers - I copied his post and e-mailed it to them. If they want they will answer.
I believe that B&O recommends to the users of their Icepower modules not to apply them to the loads below 2 Ohms - at least until very recently so I can understand this concern... Surely, Athmosphere amplifiers, particularly since he explained the "badness" of negative feedback, cam drive such loads with ease....otherwise I will start to believe that even Athpmosphere designs are not "perfect" and he is only human ... :--) :--):--)
Cheers, life is gorgeous!
Unsound - As I understand it class D doesn't need a lot of negative feedback since duty cycle is much more linear than transistor voltage in class AB output stage*. In addition it has inherent low output impedance since speaker is always connected to zero source impedance (GND, VCC) with very low resistance Mosfets. I would just guess that it is less than class A amps (gain before feedback is few hundred) and much less than most class AB amps (gain in order of few thousands). It is also worth mentioning that conversion of voltage to duty cycle is done in one stage oscillator to drive output Mosfets while traditional amplifier requires multiple output stages (hence delays) to get signal from input to output. Negative feedback causes, as Atmasphere mentioned, late summing with input signal and overshoots (time domain) or enhanced odd harmonics (frequency domain). It is called Transient Intermodulation TIM and was unknown until 70's. Some class AB amps even have saturation of output transistors that causes momentary gaps (charge trapped at the junction)with fast changing input signals (our brain compensates but makes us feel tired) - not even possible with class D since output transistors are in saturation anyway (time is analog quantity and not the voltage). For the same reason class d amp cannot become unstable.
Icepower has multiple feedbacks. One feedback, I believe, is for output voltage while another might be controlling time.
* Class AB has additional problem of different (changing) trans-conductance (gain) crossing from one transistor conducting to both transistors conducting (around zero). It is voltage dependent gain that needs negative feedback.
Dob, as long someone is receiving financial gain beyond the customary hobbyist buying and selling, I think it appropriate that the potential for a conflict of interest exists. I welcome and applaud those in the business that provide a disclaimer for contributing here, and perhaps even take their opinion more seriously despite any possible perceived prejudice that might exist. On the other hand, I am very skeptical of those in the business that do not provide a disclaimer. It's very apparent that Henry is fairly new here, and deserves a pass.
I never said all Class D amplifiers use the same negative feedback, I don't believe that, and I'm not mistaken.
We used to have a regular contributor here (I'm sorry I don't remember his user name) that has been designing amplifiers (and provided a disclaimer) for many years. His latest company ((?)located in Texas) now specializes in Class D amps. It was he that suggested that when designing typical ss amps that he avoided negative feedback when and where ever possible, but that many if not most Class D amps have plenty of negative feedback, and it didn't seem to matter as much, perhaps due to the faster speed. Please accept my apologies for not being able to provide any verification.
Kijanki, has regularly provided excellent insight into amplification technologies and I don't doubt what he's suggesting for a moment.
Unsound, My bad and sorry I forgot that I'm not well known as Ralph of Atmasphere or Charles Hansen of Ayre or Nelson of Pass Labs, and I would never dare to even compare to these exceptional designers. I'm just a rookie. So I thought by my moniker it is known that I'm a manufacturer. I believe I registered as manufacturer and not private user. First Let me say that I don't usually post much in any forum. If you check, I might have less than 5 posts through the years in Agon, and when I actually post, never have I bragging and/or pushing my own product. At any rate, You're absolutely right. Rules are rules and we need to follow. As for negative feedback or not, all I can say is negative feedback doesn't make it automatically BAD, and conversely, NNFB does not automatically makes it GOOD. It all depends what load and more importantly system as a whole. and even more important, listener's preference. Of course I can go on.
Unsound, the way you described THAT GUY sounded like it is me though I'm not sure. I have designed conventional class A for many years. In fact I still have a 100watts pairs of monoblocks and an 85 watts stereo class A. The link show my 85 watts stereo class A which I used to drive the Scintillas for years.
Dob, Thanks very much for the understanding and again I did not mean to offend you in anyway.
Brml, if you have an inefficiency and/or tough load speakers, NNFB may not works too well. Then again, It is all up to your ears.
"As I understand it great deals of negative feedback are often used in Class D amplifiers. One poster has already commented that since it is done at much greater speeds, the lag is inconsequential "
- which I understood that you generalize all negative feedbacks in class D amlifier. However, since you said that you do not believe that then my sincere apologies for misunderstanding.
I haven't read every post in this thread, but has anyone mentioned GLOBAL vs single stage feedback?
global, from output to input while other designs apply feedback at a single stage or a stage-at-a-time.
I can understand the time dependent nature of feedback, but can't that be helped by more....limited applications?
I have received reply from John Ulrick of Spectron - their chief designer:
Issue #1, stable driving a 0.1 ohm load. We can prove a Spectron amp will drive a 0.1 ohm load by providing lab data in the form of a oscilloscope graph of the amplifier driving a 0.1 ohm load. I will send a scope screen shot tomorrow [ I don;t need it - Dob].
Issue #2, driving the Apogee Scintilla speaker load. This is the issue of driving a low value complex load. The Spectron understand customers have chosen the Spectron amp because, for them we do drive their Scintilla speakers. There is a guy in, I think, Australia, who refurbishes Apogees that claims we do drive the Apogee speakers.
From me - Stereo Times reviewer Don Schaulis own Apogee speakers (don;t remember the model) and he gave Spectron spectacular review -http://www.stereotimes.com/amp013008.shtml
I have received reply from John Ulrick of Spectron - their chief designer. I sent him your first post as well:
"The above comments are only relative to analog circuits. Its true that negative feedback in slow circuits with a Z term in their transfer function have inherent ringing and do not enjoy the benefits of negative feedback. Spectrons forward loop does not have any analog circuits, its digital with a propagation delay of .2uS. Engineering requires quantitative analysis. The first order look reveals some insight. Applying super-position theorem: The period of a 20KHz sine wave is 50uS. The ratio of the wave period to the forward loop propagation delays is 50uS/.2uS = 500 In simple terms, this means that the control loop can imitate 500 control vectors at shortest wave period of the input function."
Now, if you look midrange then the ratio would be about 5000:1 and for the bass 50000:1. So, while you are correct (in your 2nd post) that fast feedback is still not in real time - I would submit to you that in practical terms you can call it "nearly real time speed" it so fast!. While John mentioned here that the control loops used in Spectron are not analog but DIGITAL - you can read it in much greater details on Spectron web site.
I will not work as a mailman here anymore - if you want to discuss NF with John or Simon - take it to them, directly.
Dob, have you ever heard or checked out the H2O Audio gear? His amplifiers are not slap-an-ICE-module-in-a-box affairs. They have very stiff linear power supplies and they sound different from other ICE module amps. His preamplifier has, as far as I know, a very unique design resulting in extremely high quality sound. While subjectively one can like or dislike his products, objectively, he is a skilled and knowledgeable designer.
Regarding negative feedback, I agree with most other end-users who have posted in this thread. In my experience, some amps that advertise no negative feedback sound very good when driving speakers that offer a benign load, but tend to sound soft and rolled off when driving more difficult speakers. While having auditioned and owned gear from a number of good designers, I've had better results with electronics that have some amount of feedback when driving my particular speakers (Avalon Opus).
By the way, I've heard the Spectron amps as well and I thought they sounded good in the two systems I heard them in.
In my post I expressed NO JUDGMENT of the sound of H2O and other amplifiers which use B&O ICE Module whatsoever. I knew in moment I will do it people like you will attack me and I like to be attacked exactly the same as next guy. If you still feel offended then my sincere apologies - we really talked about differences in designs and their effect on even/odd order of ditortions as well as capability to drive difficult loads.
"...they sound different from other ICE module amps"
Here, I must take issue this this statement. Firstly, if linear power supplies used are BEEFY then most certianly this amplifier will have better dynamic range then all others based on tiny switching power supplies (again - huge exception is Jeff Rowland). So here I agree with you but in regard of the dynamic range only.
Secondly, if your amplifier is reasonably neutral and B&O modules certainly are then any change (say in PS wire or input capacitor) will produce somewhat different texture, so obviosuly each of them will sound slightly different but their main sonic signature is determined by their input/output stage...which are identical or nearly indentical (with some mods) in all of them.
"better dynamic range then all others based on tiny switching power supplies (again - huge exception is Jeff Rowland)"
Not only that most (if not all) of Rowland Icepower amps use SMPS supplies but he uses them as well in Capri preamp and new generation of class AB amps. Rowland amps like 102, 201, 501 use standard B&O modules with built in SMPS.
To talk, as you did above, about Jeff Rowland (switching) power supplies and not to mention that he is first in the high fidelity audio world (to the best of my knowledge) who produced probably the most technologically advanced power supplies with fully or partially regulated PFC (power factor correction)and DC:DC converter (allowing safe high voltage headroom - thus his dynamic range is "exception" from all other SMPS in audio,I mentioned) is the same as to visit a zoo and miss the elephant there.
I hate to go personal but I feel that many of your posts are not for the discussion or observation or sharing experience but for "Gotha" purposes. Here, you missed and missed by miles.
Magfan, to answer your question, in a power amplifier of conventional design (not class D IOW) the problem you face if you use local feedback as opposed to global is the issue of gain. IOW if the stage that the feedback is around does not have a lot of gain, it may not help you all that much. Where this really plays a role is in the output section, wherein if the amplifier is to conform to the ideals of the Voltage Paradigm (IOW, be a voltage source), there is rarely enough gain in the output section for said local feedback to do the job.
I did say rarely: the Ayre amplifier is able to be a voltage source without any feedback.
Anyway, you encounter many of the same problems with local feedback as you do with global, but overall I would say they are less pronounced, as (in theory) a single stage is going have a shorter propagation delay than an entire amplifier. This will tend to push the time domain issues to manifest at a higher frequency. That does have its good and bad side, so again a lot depends on the actual design!
How I see all this is very simple: as long as the amplifier fails to add **any** distortion to the 5th, 7th and 9th harmonics, then it all should be good. It will be even better if the lower ordered harmonic distortions fail to appear as well, although this is far less important.
Mentioned Rowland 102, 201, 501 have standard SMPS without anything coming from Rowland. You can buy amplifiers with built in PFC like model 312 or buy additional power supply but most of Rowland Icepower amps have only standard B&O modules with integrated SMPS.
Please check power amp listed in my system before you make another silly personal remark involving elephants.
Magfan - local feedbacks are always better than global one but their application is limited.
Class AB amp design has to start with excellent linearity and wide bandwidth since one of conditions to eliminate TIM is to limit bandwidth at the input to one amp had before feedback was applied. Feedback can fix inherent class AB problems like "gm doubling" (different gains for small and large signals) but cannot fix poor design. AFAIK there is always some form of feedback (local one) even in zero feedback amp.
Atmo / Kij,
Processing your replys now........and they agree generally with what I had heard, which is why I brought it up.
Talking about feedback without qualification doesn't help a lot.
I also have felt that the B&O 'd' offerings have more in common than differences, no matter WHO packages the modules and what mods are performed.
One question though, about 'd'. For at least the ASP modules from B&O, they come WITH integrated SMPS. How do you get around that? Go to another module...the ASC(?) series which has no PS included?
Magfan - ASC series has also SMPS. I think that it was AP series but now Icepower makes different modules.
B&O SMPS are pretty good - very quiet (zero voltage / zero current switching) and very strong (1000ASP can deliver 40A output for 0.5s). What people often don't realize is that SMPS supplies are line and load regulated while linear supplies aren't. Linear supplies require a lot of capacitors to keep voltage steady and filter out 120Hz. They are also noisy since switching is done in audible frequency at max voltage. Linear supplies are also huge with big transformers. Small 1"-2" dia. toroidal transformer at 100kHz can deliver same power as huge toroidal transformer at 60Hz.
Best Icepower amps use 1000ASP with addition of extra supply that feeds over 400VDC. From what I read it has positive effect on the sound but module still uses its own supply. It is perhaps bunch of extra capacitors at high voltage (low losses).
Smaller Icepowers like 200ASC used in my Rowland 102 have higher carrier frequency and therefore wider bandwidth.
If you plan to use one, you might want to investigate input circuit that increases input impedance from 10k to 40k based on THAT1200 instrumentation amp (tiny board). That's the only Rowland contribution (other than beautiful case) to my amp.
I have to admit that he deserves credit for recognizing genius of Karsten Nielsen and B&O company. B&O was so fascinated with the project that they sponsored Nielsen's doctorate and gave him shares of company - first time in 70 years of B&O (being private company).
I've lost track of what B&O makes. Here is link to ASP module....I suspect as used in my PSAudio GCC series integrated::
The data sheet lists 50v and 80v outputs...with which to feed the extra, non power supply equipped modules. The 80v is listed as 7 amps, If I remember correctly. These numbers are with the 'home' amp having zero output....
The GCC series is a so-called variable gain design. The input circuitry is handled thru something called a 'gain cell' which I HOPE takes care of any impedance mis-match with any source connected.
The balanced inputs from my CD player are wonderful....
Thanks for the responses. My speakers are Coincident peII's w/ a flat 8ohms load and 92 db. Aljordan's comment about znfb used with easy to drive speakers is true, I believe at least in my experience. The Unison would be used on the same speakers. I try to stay away from "hard to drive" speakers as it limits your amp choices too much. I prefer SET amps, or tubes in general. Atmasphere hopefully someday I can listen to one of your amps, no dealers in my area.
"...SMPS supplies are line and load regulated while linear supplies aren't. Linear supplies require a lot of capacitors to keep voltage steady and filter out 120Hz. They are also noisy since switching is done in audible frequency at max voltage. Linear supplies are also huge with big transformers. Small 1"-2" dia. toroidal transformer at 100kHz can deliver same power as huge toroidal transformer at 60Hz. "
Kijanki, I am not EE so I cannot argue professionally. However, everybody knows that with exception of Jeff Rowland only class D amplifiers use SMPS supplies and the rest of the world use "huge" linear PS. So, there is something "there" in SMPS not so positive... Again, I just am making an observation.
Dob, I disagree with you. Jeff Rowland uses his SMPS with PFC (for Kijanki - on his premiere amplifiers) but he is not the first. See below from Spectron web site:
" Spectron designs team won a contact with pro audio giant Inter-M (among others in competition were Bang & Olafson and Philips) to build a highly reliable continuous output 8000 watt power compact digital amplifier. We successfully demonstrated a prototype of the first stage of this unique amplifier at the NSCA (National Sound Contructor Association) meeting in Orlando Fl , in March of 2007. The highly innovative design includes very high level power factor corrected switching power supplies using the latest PFC conversion technology and resonant mode DC:DC converters"
I can't even imagine amplifier with continuous output (rms) of 8000 watts and who needs it (besides its use in Madonna concerts :-) but they also use SMPS supplies with PFC and DC:DC convertor at least since 2007
I spoke with Simon Thacher from Spectron about the fact that they use in their Musician III amplifier linear PS and not their SMPS, a while ago but I can't remember a word of his reply, today. So, I can make two points:
1 - If to use SMPS then only with PFC + DC:DC Convertor
2 - Even SMPS w/PFC are not automatically better then linear PS - as Spectron has both types and uses both on two different amplifiers.
Finally, I agree with Magfan and Dob that ".. also have felt that the B&O 'd' offerings have more in common than differences, no matter WHO packages the modules and what mods are performed ". I think, one need to hear Kharma, Weiss, Spectron etc to see true potential of class D technology.
Rowland uses SMPS in all class D models (including integrated). In higher models like 312 or integrated Continuum he uses additional PFC unit. For the rest like 102,201,501 you can buy external PFC unit. Such unit supplies about 450VDC and you don't need any DC/DC converter (term tossed often to impress people) because B&O module can be supplied directly with 450VDC.
There is more than 100 companies manufacturing class D amps any many of the use SMPS. Example of company that uses only SMPS is Bel Canto, NuForce, Red Dragon.
New Rowland high end class AB amps costing $48k also use SMPS. Read this thread:
For class D amplifier SMPS is obvious. If you think that class D amp is good enough then SMPS, that is the same thing (class D), should be as well.
I know, you know all that and am just restating but as for the rest of the world using only linear power supply - I don't know. I wouldn't risk such statements without checking. Others with more experience can perhaps help but I remember that Linn amplifiers have SMPS and are not a class D amps. Even some preamps like Rowland Capri use SMPS. I can tell you that linear power supply is easy to design in comparison to SMPS. In addition there is natural resistance as it was at the beginning to class D amps. It is very difficult to change opinion of people (like you) so designers design for current demand (I don't blame them).
One more thing - PFC is not conversion but correction. It is beneficial for power company but not so much for the end user (lower overall efficiency since PFC unit has its own efficiency <1). What counts is just whole bunch of electrolytic caps at high voltage fortifying SMPS but the same thing (extra capacitors) could be included in SMPS to start with. Again "PFC correction" is more impressive than just "extra caps". Bel Canto has similar thing inside of cabinet but they call it "increased energy storage". Their amp Ref1000M made Stereophile class A rating. There is not so much sound difference from the original Ref1000 but it allows some critics who made previously stupid reviews to save face and reevaluate (Martin Colloms gave this amp score of less than one of 10).
Absolutely true, but linear power supplies also create HF noise in the AC mains by drawing current in short narrow spikes of huge amplitude. They are practically switchers at 120Hz. Modern SMPS with zero current / zero voltage switching can be extremely quiet. I don't see other reason for Jeff Rowland using them in preamp (Capri) where regulated linear power supply could be used.