Add distortion from 2-80,000hz, dual mono, well designed power supply that doesn't need cleaning up with ISO transformer, power cord, etc.
20 responses Add your response
Given all else being equal. All five of your parameters are all desirable goals for amp design.
Unfortunately all else are never equal. Designing andmanufactoring an amp at a given price range are always about compromises. That's why these goals are becoming more-or-less gimmicks instead. Every amp is always compromising something.
Unless the budget is astronomical, it's pretty unlikely to find optimize for all these goals.
Take a look at the statement amps from Levinson, Dynaudio, Krell, Classe, Musical Fidelity, Pass and etc should give you an idea. Even these guys don't have class A output if the rated output surpass 50-100 watts range.
The biggest class A rated output is probably Jadis 200 or the Pass Aleph series. Somewhere around 120W/ch, but they both have other issues as well.
To me the design parameters are important to me in the foll. order:
* Zero global feedback - reduces distortion & improves 3D imaging
* Class A output - for best linearity. We can have a class-A amp w/ lots of feedback but its worth very little!
* Instantaneous Current Delivery - this implies a very well designed power supply. Clayton Audio amps seem to rank the highest here. They are rated at 200A continuous! One could weld with them.
* High Damping Factor - this speaks about the amp output impedance. Since power amps are voltage sources, the best voltage source has zero output impedance. Of course, this is not physically possible so the next best is as low as possible over the entire audio band. Such amps will have killer bass response such as Symphonic Line & Wolcott tube amps. There are others, of course, but these come to mind right now.
* Balance In/Out - according to me this feature is the most hyped & most marketed! It has some advantages but not as much as what is hyped. I have no personal experience w/ BAT but this co. seems to make fully differential designs from input to output rather than create a single-ended design & then create an out-of-phase signal & call it balanced! Fully balanced designs seem to make sense if common-mode noise is an issue. I have heard single-ended designs that sound awesome so I'm not fully convinced that a balanced design brings a significant advantage to the table. Of course, JMO.
If only it were this simple... I have heard amps with negative feedback sound better than ones without negative feedback and vice versa. Same goes for class A versus AB and balanced versus single ended. As far as capacitance and dampening goes it really depends on the type of speaker you are using. As with most things design, execution and application are actually more important.
What I have found is that lower powered amps (including tube amps) generally sound better than higher powered amps in the same price range and if power is not an issue.
Ask the manufacturer of your speaker for amp recommendations.
How about the Belles 350A which is almost a purely class "B" design. It has a ultra high damping factor. Give one a listen if you get the chance or check out the review on Sounstage.com
Also, in the above discussion, a goodly number of amps use a cap to block DC on the input. This, to me, imparts a slight change in the sound for the worse. It seems input blocking caps have been substituted for DC servos (I know it's cheaper to do.) Also, as stated above, many amps go just a few watts into class A then crossover. With this, waveform distortions can occur. The positive and negative half cycles don't match up.
What i'm saying here is alot can happen to influence the sound of an amp besides the above listed things.
Why do people love tube amps---most of them stink on the test bench in any of these parameters. Most tube amps have a damping factor of around 20 or less! Could it be we like the "Distortions" a tube amp produces? (they sure test like crap!)
All the discussion points to one thing. Trade-offs for the typical audiophile amps. They were all optimized for a few particular things but not everything and they can't be due to cost consideration.
It's possible to design and build a 1000W/ch amp class A, balanced, zero-global feedback, balanced and high current delivery, but it will probably cost a quarter of a million dollars per channel and the size of a sofa.
So all these compromises designer made will basically force the marketing to emphasize specific strength which usually make all these terms unimportant and impossible to compare one on one.
If you are amp buyer, don't take these terms too seriously. If you are an amp designer, that's a complete different story.
BTW, if you can't tell yet, I am by training an eletrical engineer.
1, 4 and 5 are bare minimums for a nice amp. 2 is cool if you can balance the entire signal chain. 3 is nice if you can live with the heat and the electric bill. If you dig spec'ing out solid-state amps, compare the number of output devices/channel and the amounts of distortion and noise. After a while you'll get a feel for where the corners were cut.
I usually take a peek at the slew rate. To me it says something about how careful the amp design is; if the amp sounds good to boot, then somebody did his homework. Higher is better but if the topology is sloppy, a high SR can mean instablity, distortion, and noise - especially with more output devices. IMO, this parameter gives the amp its speed (pace), adds to the illusion of realism by presenting transients closer to real time, and presents quieter silences. Of course, it doesn't pull everything together - there are many other factors (I wouldn't call them "marketing gimmicks", they are just specs, which show performance, which in turn only hints at the sound).
I was reading on the VAC website a while back about damping factor, and it caused me to rethink it a bit. Their point is that amps with high DF are essentially using back-emf as feedback to the amp, thus modifying the current delivery from the amp in real time. They then go on to make the points that (1) the voice coil is a lousy sensor, as it wasn't designed for this, and (2) the back-emf almost always passes back through a crossover network, which further changes its amplitude and phase characteristics. Interesting points for those who think high DF is the way to go. It may not necessarily be so. It may actually be better to have a zero-feedback output stage and simply reduce the output impedance as much as possible. Then it's up to the speaker to accurately reproduce the waveform being sent to it. Just some food for thought; I personally don't know which is better.
Interesting feedback comments!
Likewise I was reading the GamuT Audio website (http://www.gamutaudio.com
On the right side click "FAQ" & then click "what is damping factor") & they claim that damping factor of just 25 is sufficient.
When I think of damping factor I mostly think "amplifier output impedance". The lower it is, the better off you are (& the more naturally you get bass control). I'm not aware of the various techniques of obtaining high DF but your comments have me pondering....
The more the speaker impedance changes, the more the variation in DF, which is not so good i.e. load dependent DF. I have inferred that those amps that have a constant high DF over the audio band have low output impedance over that same freq. range. Might be a simplistic view but it is one criteria I use to short-list an amp. Listening to it provides the final verdict.
A small point: DF is the ratio of the load impedance to the internal impedance of the amp. If we average the speaker load to 8ohms, just how do we get a DF of 25 or 50 (unless we're marketers)?? In other words, how low must the amp's impedance be to reach these beautiful ratios?
Where's Sean and other EE's?
The way you get really high DF (ie, really low output impedance) is with negative feedback in the output stage. This isn't the only factor, just the easiest way to reduce the output impedance to really low numbers. And with negative feedback comes the issues I mentioned above. Visit vac-amps.com; there are numerous thought-provoking essays on this and other topics. I'm not saying they're right, or that there aren't other issues, but they do offer some interesting points to think about.
I would like to make a comment here:
The post 'Karls' made on 7-30-03 seems to suggest that ALL NEGATIVE FEEDBACK is bad!! This statement is not correct!
As an EE, I know that we rarely design an amplifier (with large gain) WITHOUT feedback! It will clip to the + or - supply rail the moment you feed it an input signal esp. with music transient signals that are rampant in classical, Jazz & blues music.
The key to negative feedback is that the designer must be intelligent to use it properly. I will agree with Karls that it if global feedback (from output of amp to input of amp as was done a lot in the 1970s) is used then the sonic signature of the power amp is botched up pretty damn well nicely! However, if local feedback is used most/all of the negative feedback issues can be overcome. Local feedback is defined as feedback that exists local to 1 device or a few devices or an amplifier stage.
I cite 2 prominent examples that use negative feedback. Let me point out here that I simply use these 2 amps as examples - I'm not bashing nor am I praising. Pass Labs X series amps claim "no negative feedback" & many would believe this. However, read his white paper on his website that clearly states that he (Pass) considers "no feedback" to synonymous with "local feedback". He even challenges the reader to submit to him a "true no feedback" circuit so that he (Pass) can point out the hidden feedback! Per his literature, it is true to say that the X & XA series amps operate heavily into class-A hence use no feedback in the output stage (but there is local feedback in the 1st gain stage). That is because, for a MOS device, output impedance is inversely proportional to drain current. If one uses a "healthy dose" of bias drain current in the output stage, output impedance will drop to practically nothing. This will result in a NATURALLY high damping factor.
The second example is Spectron Musican II. The designer clearly states the advantages of negative feedback on his website. The Musician II seems to use local negative feedback & those of you who own it, can testify to its great sound. I will agree with the Musician II designer that the amp has to be high bandwidth (i.e. fast) for the feedback to work. Most of the time feedback makes the amp sound sloppy 'cuz the amp stage(s) to which feedback are applied are not high enough in bandwidth & by the time the amp reacts, that particular music passage is done & long gone. In this case, the issues with feedback that Karls pointed out are very valid. This does not mean that feedback is bad! It just means that the amp designer is incompetent!
However, not every amp on the market is class-A operation. In fact, most are class-AB & some of those are class-AB with emphasis on class-A upto a certain wattage. The better sounding amps in these 2 classes use LOCAL negative feedback. They HAVE to! No way the above cited amps can have the much sought after attributes of unconditionally stable into any speaker load, low distortion, low output impedance, low noise, etc, etc. w/o some sort of feedback.
Anyway, for power amp design, negative feedback is a fact of life. Less in being today as compared to ss designs from the 1970s but it is STILL being used! With intelligent use of negative feedback, the issues associated with this technique can be overcome with good/great results. My experience is with solid-state designs only. Thus, I will not comment on negative feedback or lack of as used in tube designs. Some of the concepts of tube design can be ported over directly to the ss domain but there are a number of pecularities of tube design that cannot. These are exploited positively by creative designers to produce some great sounding amps that not always measure A1 in the lab. It is very possible that what Karls read on the VAC site is particular to tubes only & that technique cannot be applied to ss designs to reduce output impedance. BTW, I have yet to read that VAC site info.
Just how high the D.F. has to be to get good/great bass - I don't know for a fact. The Gamut site says 25 is enough. The Pass site says a large number is not needed but doesn't state how large is large (1? 10? 100? 1000?). You can be quite sure that, with the X & XA series amps producing killer bass, their D.F. is on the high side rather than the low side. Providing a low impedance to the woofer VC thru the amp output impedance is important to start/stop it quickly. That's what gives us "fast" & "tight" bass. Also, do not forget that D.F. IS frequency dependent 'cuz the loudspeaker impedance is frequency dependent. Each time the loudspeaker impedance dips, the D.F. reduces. If the amp. has to provide "fast" & "tight" bass regardless of this impedance fluctuation then the amp output impedance BETTER be real low! Local negative feedback can help us here unless you have a high power dissipation pure class-A power amp. Many of us cannot afford such monsters but that doesn't mean we cannot have a good/great sounding amp!
Bombaywalla, great post. Please do not assume I have bought into the "all negative feedback is bad" camp, as that is not the case. I merely mentioned VAC's site as interesting food for thought, and much of what they have to say is probably applicable primarily to tube amps due to the presence of the output transformer. I will say, however, that two of my favorite amp designs presently are Pass X and Ayre V_x, both of which are touted as "no feedback" designs, and both of which sound purer than most anything else out there. And I would venture a guess that both of them have quite high DF to boot.
Very interesting and well put, Bombay! (The EE in you manifested himself, thankfully:))
Engineering a constant current device is a bit tricky, (albeit nice for speakers) especially when we go over the sacrosant 2,83V level. Astonishingly, it seems that some SS amps have been designed with the full throttle in mind, linearity falling as we lower the volume.
For that matter, how many of us nowadays listen at more than ~4V at the amp output -- with a 89-90db efficiency speaker that's enough to tire the ear pretty fast!
The reason I ask is because some have very stong feelings on this subject. For example, Richard Hardesty (Audio Perfectionist Journal) consistantly claims that the best sounding amps that he has heard are fully-balanced in design w/ no global feedback. However, he doesn't really talk about any other factors of amp design (at least, not yet).
Other reviewers have ranked amps of *very* dissimiar design very closely together. A Clayton M-100 is very different from a McCormack DNA-500, but I have seen them referred to as comparable. Throw tube amps into the equasion, and on paper, things get confusing really quickly. However, the confusion is usually quickly sorted out *if* you are able to do side-by-side listening comparisons (which is generally logistically difficult to do).
Oh...and I apologize for my spelling of 'architecture' in the original post.
Understand why you posed the question in the 1st place. I agree that it's very hard to discern an amp's sound quality by just looking at the specs & knowing what sort of technology was used to design it. I also agree that the best is to listen to it.
You know, we might be surprised when we open up the M-100 & DNA-500 & place them side-side! It is quite possible that the 2 independent designers used techniques more similar than dis-similar to design their resp. amps! Neither designer provides much info about the techniques used & we have to rely on the IAR journal to ferret these out for us. I suppose that you are referring to that IAR80 article, which I also read with a lot of attention, that put both these amps in the same class.
We just don't know what these designers are putting into their amps esp. if they as tight-lipped as Wilson Shen & Steve McCormack. I read a review on the M-100 amps, which was a reprint of a Bound For Sound review, where I was informed that I should not use high impedance 10-16 Ohm speakers nor should I use a tube pre with the M-100 if I really wanted the amp to sing. J. Peter Moncrieff never mentioned this in his article! So, should I/we append this tit-bit of knowledge on the M-100 to the knowledge gained from the IAR80 article? Don't know......