Post by mgbpuff on Dec 28, 2017 16:47:33 GMT -5
The slew rate which is necessary to avoid creating distortion depends on the frequency involved, the signal voltage, and the gain of the amplifier (the amount of feedback it uses).
So, for example, an amplifier that has a higher output power would require a higher slew rate in order to avoid distortion at high frequencies.
However, as long as you exceed that minimum by a reasonable margin, exceeding the minimum by 10x is no better than exceeding it by 5x.
In theory, there would be no reason to not want infinite bandwidth.
However, in practice, there are two reasons why you would want bandwidth to be limited.
First, as you suggest, noise is a problem.
You wouldn't want your tweeters to catch fire when you walk near your amplifier with your cell phone.
(This really could happen if the length of your interconnect just happened to be an even multiple of the length of the antenna in your phone.)
And you certainly wouldn't want your head to explode when your microwave oven turns on.
Now, obviously, this is only going to be a problem if the amplifier is sensitive to high frequency noise.
However... guess what... one of the ways designers reduce sensitivity to high frequency noise is to limit the bandwidth to what is useful.
(It might be cool if your amplifier could accurately reproduce 10 mHz square waves; but it won't sound any better; and it's a lot more likely to blow up.)
Second, each individual design has inherent relationships between frequency and phase.
In most cases, this means that, as the frequency increases, the phase shift also increases.
With a normal amplifier design, if you allow the gain to exceed unity, while the phase shift exceeds about 270 degrees, the amplifier will oscillate.
(Which will destroy the amplifier and/or your speakers.)
The way we avoid this is to design the amplifier so that the gain decreases at high frequencies - where phase shift is increasing dangerously.
We make sure that the gain drops to below unity before the phase shift is high enough to cause oscillation.
In other words... we limit the bandwidth.
A well designed amplifier has plenty of bandwidth to accurately reproduce any legitimate audio signal...
But is designed so that it won't try to reproduce high-frequency noise or signals with enough phase shift to cause instability or oscillations.
(Many older designs were "just naturally limited" to a safe bandwidth; and some were "just barely good enough"; but many modern components have such improved performance that they must be explicitly limited.)
The "take away" from this is that modern amplifiers are not designed to have "the widest possible bandwidth".
(Modern components would allow us to build amplifiers with bandwidths far wider than would be useful or prudent...)
Modern amplifiers are virtually always designed to have a specific bandwidth - chosen to result in the best overall audio performance.
As a result of this, the idea that "an amplifier with a wider bandwidth is a better design" is simply no longer specifically true.
However, I would challenge your blanket assumption about "there must be something there for the money".
The reality is that, while there are plenty of expensive audio products that do deliver exceptional sound quality, or really nice fit and finish.....
There are also lots of expensive audio products out there whose only excuse for their inflated price seems to be that "someone was willing to pay it"......
I'm also a little curious why you would assume that the Krell has wider bandwidth... or a higher slew rate... or that this would be desirable.
(While insufficient slew rate or bandwidth can lead to various sorts of distortion, excessively wide bandwidth can also lead to problems, so a "happy medium" is usually the design goal.)
But reproduction of a transient signal accurately depends on a summation of frequencies, many beyond the audible, and those frequencies must not only be magnitude amplified accurately but they must occur in phase precisely as per the input transient signal in order to reproduce an accurate but amplified transient copy. I mentioned slew, but I think that refers to an internal amplifier integration speed limitation (or is there really any difference between slew limitation and overall transient performance?). I fail to see how a wide bandwidth could be detrimental unless you're thinking noise amplification, but the high frequencies would not be audible and many speaker systems would not even respond to such higher frequencies. However, some tweeters, such as wire type or extremely light metal cones in this age can respond; So, the reproduction of a transient depends on the sound transducer also.A disclaimer is in order here! Although I am often blamed for talking to myself out loud, I have never done so on line.
The first six paragraphs above are by KeithL. 1)Although the delay the signal takes on in passing through the amplifier is transport delay and has nothing to do with slew, I agree on the effect of slew (a rate of change limitation). It is an internal limitation of the ability of a given amplifier to change level quickly. If it's fast enough to not interfere with the input signals speed, then it need not be faster. 2)Humans can hear certain audio transients such as cymbals and their crispness of sound. It takes frequencies beyond just the ordinary audible band to create such transient sound both in reality and in amplifier reproduction. Heads won't explode! 3) No one is proposing a bandwidth wider than is necessary to reproduce sound accurately, but that bandwidth may be greater than the typical 20 - 20khz +- 3db. Most amps exceed this anyway. And finally 4) What you are describing as decreasing gain with frequency so as to not allow oscillation is another way of saying that a given amplifier with a given bandwidth must be designed to be stable. That is what engineering system design is all about and any bandwidth desirable can most likely be designed to be stable. If it is not stable, then it is poorly engineered.
