|Elliott Sound Products||What is Hi-Fi (Part II)|
Copyright © 2004 - Rod Elliott (ESP)
Page Created 12 Jan 2004
Published 27 March 2004
In Part 1 of this topic some of the general principles were covered. It explained that many of the concepts applied to 'High-End' systems are at odds with one's listening pleasure, and included some reasons for this.
We tend to think that a hi-fi system should be accurate, but what exactly is accuracy, how do we assess it, and what happens if the system is not 'accurate' for one reason or another?
The purpose of this article is to attempt to cover the reasons that accuracy is not always desirable or achievable, and to allow the reader to make an informed judgement of a system and its abilities. This is a somewhat daunting task, and one that may raise the ire of many, since it appears that I am suggesting that an accurate system is not useful.
It's not that an accurate system is not desirable - it is desirable in the extreme, IMHO. The problem is not so much with the system we listen to, but the source material itself. The room also plays a very important part in the equation, as the average listening room has a great many anomalies at varying frequency ranges, and can make an excellent system sound very ordinary indeed if it is unsuited to the purpose of listening to music.
I do not intend to cover room treatment in any detail - there are many articles already that do just that, and it is a non-trivial exercise at best. As an example though, I am sure we have all seen photographs on the Net and in magazines of superlative looking systems in a stark room, having no carpet and a nice tiled floor, minimal furnishings, and lots and lots of glass windows. While this may well suit the purposes of the photographer, a system in such a room will almost always be a complete sonic disaster.
Likewise, having heard a piece of music on the radio (for example), you go and buy the CD, only to get it home and discover that it is over equalised, compressed to within an inch of its life, and has excessive and boomy bass. I have several such recordings, and my system is too accurate, revealing every flaw with punishing accuracy. Under these conditions, an accurate system is not ideal - in fact it is a disaster! Music you might otherwise be able to enjoy is ruined, and a typical "high-end" audio system is completely inappropriate for a great many recordings. For example, it should not be necessary to switch off the subwoofer to play some CDs.
Accuracy in a general context could mean that the system will play the source material with no changes to the frequency response, harmonic structure, relative phase of the signals, or anything else. While this may seem to be the ideal situation, it is never achieved in a practical application. Non-linear distortion will be added, and this changes the harmonic structure of the signal. Very small amounts of non-linear distortion are added by most amplifiers and preamps, much greater amounts by all speakers. While the levels will rarely be objectionable with well designed equipment, that the signal undergoes some distortion is an absolute - there is no such thing as a zero distortion amplifier or transducer.
Certainly, the distortion levels may be able to be reduced to below the noise floor of the listening room, but it is still there - whether you hear it or not is the only thing that is important. The concept of 'masking' - where one signal is deemed inaudible because of a higher level signal nearby does not always work as well as many designers may hope. A perfect example is MP3, where all 'masked' signals are removed as part of the compression algorithm, and 'near CD quality' MP3 recordings lack any sense of imaging or 'space' because we use the 'inaudible' signals as a cue for image placement. Since they have been removed, we lose that ability.
Frequency response will always be modified to some extent. Again, whether these changes are audible is the only question that is important - you (as the listener) may actually want to modify the frequency response, either to compensate for recording quality, room effects or even hearing loss. Should you suffer from a loss of high frequencies, then the ability to compensate (at least to some degree) will enhance your enjoyment, although it is probable that others will find the sound overly "bright".
Phase shifts occur in all aspects of electronics to some degree, and are particularly easily measured with crossover networks. Fortunately, our sensitivity to static phase shifts is rather low, and we can tolerate huge phase shifts with no noticeable change in sound quality. Relative phase is another matter, and any piece of electronics or transducer array that changes the phase of signals dynamically (almost always deliberately) will cause very audible effects. Rotating speakers (e.g. electronic organ Leslie cabinets), or phase shifting networks (e.g. phaser pedals for guitar effects) have a profound effect, but these are rarely found in hi-fi systems. Loudspeaker drivers inadvertently wired out of phase cause static cancellations of some frequencies, and this is almost always audible.
Phase shift is one of the least understood of the distortion mechanisms, but it is important to understand that the phase of any signal is changed radically by distance, even if it started out as perfect. Sound travels at 345m/s in dry air at sea level at around 22°C, but this changes with altitude, temperature and humidity. This works out to be .354mm/us, or 35.4mm/100us. Given that a typical listening distance may be 2 metres, that represents a time delay of 5650us, or 5.65ms. The chart below shows the relative phase shift of a perfectly phase coherent signal, at distances of 1 and 2 metres.
Figure 1 - Phase Shift at 1 & 2 Metres Listening Distance
At a maximum of 2,000°, this is hardly 'phase coherent' (note that the chart is in k degrees), but it happens with every speaker (including the human kind) in every room, under all circumstances. This is not distortion, but simple reality, to which our hearing is fully accustomed. The phase shift is the result of the time delay suffered by the signal as it travels through the air, and is completely normal. Hearing is not affected by this (radical) change for the simple reason that it occurs all the time, with every sound source. It also renders the claims that some speakers can reproduce a perfect squarewave rather suspect, as shown below ...
Figure 2 - 1kHz Squarewave Response at 1 Metre
Not what I would call a perfect squarewave, and naturally the shape varies with distance and frequency. The room will also have a profound effect on the waveform, and from tests I have done, I know that perfectly ordinary (non time-aligned) loudspeakers can also reproduce a squarewave - provided one is sufficiently patient to find the 'correct' microphone placement. As a matter of interest, it is almost impossible for any loudspeaker to introduce a phase shift of this magnitude, regardless of crossover topology or baffle layout. As I have pointed out on several occasions, phase shift in and of itself is usually considered to be inaudible, but the effects of combining signal sources with different phase shift characteristics can be very audible indeed.
The greatest distortion is one of time itself - the re-creation of a piece of music without the musicians being present is obviously a "distortion". 100 years ago, if you wanted to listen to music, then the musicians would be right there with you, and there was no distortion - you heard exactly what you heard at your seating position. Others would hear things differently, because they would hear the same music from a different location, thus changing the sound. It is no longer necessary to have the musicians present, but for the purposes of any discussion, this distortion is usually ignored. It shouldn't be, because this is the one thing that people strive for - to make their system sound like the original performance. The fact is that this is impossible, because the original performance was recorded under conditions that cannot be duplicated in a home system, in a normal room, and with all the changes that have been made during the recording process itself. Indeed, even when present at a concert, you will hear the sound differently from different locations - which is right, which is wrong? Obviously, no position where all instruments are heard clearly is wrong, but they are different regardless.
If your or my system is accurate yet cannot reproduce the music in a manner that is enjoyable, what is the point? One could scour the record shops in a quest for CDs that are well recorded and properly mixed (i.e. without excessive compression or use of effects), but if the CDs you can find have music that you don't particularly like, again, what is the point?
If you happen to enjoy "pop", contemporary or alternative music, then you absolutely need some method for correcting the often appalling mix quality, since a great many of such recordings are deliberately mixed to sound "good" on exceptionally average systems (such as factory installed car sound and department store "hi-fi"). If you like classical or string quartets, choral or organ music, you are not necessarily safe there either, since it is very common for "post processing" to be performed by the mastering engineer.
The fact of the matter is that the only way you will hear exactly what was intended by the recording engineers is to hear it in the studio where the recording was mixed, using the same speakers at the same level - that this is impractical in the extreme should be fairly obvious. No system and home environment will match the acoustics of the studio or the monitor speakers used. Even if you got lucky with a "fairly good" match, it would only be applicable for a single studio at the same level that was used for mixing.
Equalisation is considered a rude word by most in the high-end (or audiophile) community, yet an equaliser, or even tone controls, can rescue an otherwise unlistenable recording. Exactly when and why tone controls got such a bad name is unclear, but my next preamp will most certainly use them. It is probable that they will be used only on a very few CDs, but at least my investment in the music is retained if I can actually listen to it.
What is really at issue here is that "accuracy" has become the catch-cry of some of the high end of the audio fraternity, and while it is absolutely essential that a system is capable of a flat response and full dynamics (accuracy), this is a very open-ended situation because the quality of recordings is so variable. It would be nice if orchestral recordings (for example) were labelled "No EQ" or "No Post-processing" so that we would know that the response had not been tampered with beforehand, but this is not the case (at least for any I have seen).
There are Audiophile recordings available, but if you can't find any with the artists of your choice or the material of your desire, then they are of no use to you. In some cases, while the recording quality may be excellent, the musicianship may be left wanting. This is a no-win situation, and will remain so until tone controls take their rightful place in systems once again.
It goes without saying that tone controls are not the real answer (even though I just seem to have said they were), but if they can be used to rescue a recording (or several), then they are worth inclusion. Ultimately, hi-fi is about music and enjoyment, and if someone really enjoys listening to everything with the bass turned way up, then the facility should be available. Musical enjoyment is a very personal thing, and for a great many people the audiophile world has shut them out because such systems do not have the features they desire.
No-one has the right to impose their standards on everyone else, yet by insisting that systems have a flat frequency response that cannot be adjusted (not even balance controls are used in many systems) you are allowing recording engineers to impose their preferred sound onto you - if you don't like it, there is nothing you can do about it, apart from return the CD and demand your money back (not a bad thing to do, by the way, but you may deprive yourself of a lot of music you'd otherwise enjoy).
I have commented elsewhere on these pages that RIAA phono equalisation does not have to be super accurate, since no-one has any way of knowing what additional EQ was applied to the cutting lathe at the time of mastering. That the playback EQ should be accurate to within a dB or so is a sensible approach, and deficiencies can easily be made good by the use of tone controls - again assuming that they are available.
What of systems where the user can adjust each frequency band in a multi-way active system? By the normal high-end standards, we should be appalled by such an idea, but it makes sense. Given that music is such a personal thing, why should anyone have to listen to the system as it was designed and built?
There is no reason that people should not be able to tailor the sound to their preference if it increases their enjoyment of the music of their choice. It will rarely be accurate, but what difference should that make? If you don't like it the way the user has set everything up, that is your problem, not theirs. As long as the listeners (owners) are happy with what they can achieve, to get the sound they like, then that is all that matters. You can go home and listen to your system set up the way you like - naturally, I consider the ability to get a flat frequency response (as well as good transient response, low distortion, etc.) as essential in all of this, but whether it will be used like that is another matter altogether.
Accuracy in the sense that I use it here is primarily about frequency response, with phase and transient response following closely behind. Linearity is usually taken to mean freedom from non-linear effects, and specifically harmonic and intermodulation distortions. As noted above, these are always present to some degree, and usually vary with amplitude (loudness) and frequency. Most at risk are extremely low and high frequencies - low frequencies because they demand large amounts of movement in the transducer, and high frequencies because of limitations within many tweeters.
Quoting THD (Total Harmonic Distortion + noise) is a pointless exercise if the waveform of the distortion component is not disclosed (which it generally is not). Two amplifiers having apparently identical amounts of distortion can sound very different indeed, and this has led to THD measurements being declared 'useless'. Provided you know what kind of distortion is present, it is generally easy to make a prediction as to whether it will sound "good" or "bad". Solid state amplifiers with 1% crossover (or notch) distortion sound unquestionably bad, while a valve amp with 1% 2nd harmonic distortion will sound good (especially by comparison).
Linearity is important, since even relatively small amounts of THD can create much larger amounts of IMD (Intermodulation Distortion), where each frequency being reproduced beats or mixes with other frequencies to create new frequencies that were not present in the recording. This happens with electronics, but is usually much more pronounced with transducers, and for this reason (amongst others), it is common to use separate transducers for different bands of frequencies. The simplest is a woofer and a tweeter, with a crossover network that divides the signal and sends the low frequencies to the woofer, and high frequencies to the tweeter.
Transducer non-linearity usually exceeds that of electronics by a great deal, but tends to be 'softer' and less audible. Not so however when the linear excursion limits of a speaker are exceeded - the distortion produced can then be very audible and highly objectionable. For systems that are used at low levels this is not so much of a problem, but to obtain good linearity at even reasonably high levels requires that drivers be limited to a maximum of about 1 decade (a little over 3 octaves). To manage the whole audible frequency range, the signal may be split as follows ...
|Low Bass||Mid Bass||Upper Mid||Treble|
|20 - 160Hz||160-1280Hz||1280-10kHz||10kHz-20kHz|
Unfortunately, for a number of reasons, this is not a practical separation of the signal. Because of the difficulties with bass (in particular) a more common split may be as follows ...
|Low Bass||Mid Bass||Upper Mid||Treble|
|20 - 80Hz||80-500Hz||500-3kHz||3kHz-20kHz|
This too has problems - there is a crossover frequency at 500Hz that may cause problems for the natural reproduction of voices (a very critical requirement for any system). This is not insurmountable though, and the second variation is fairly common (with some minor changes depending on philosophy, driver specifics, etc.).
Distortion actually covers every aspect of the system. Technically, if the volume is changed, you introduce distortion (amplitude distortion), since you are not listening at the same level that was used when the music was recorded or mastered. In turn, this changes the response of your hearing, with the bass and treble seeming to be less pronounced as volume (SPL) is reduced (and vice versa). The change in amplitude is not considered as distortion, but it should be, since the relative loudness of the high and low frequencies is changed relative to the original.
The term "distortion" is usually taken to mean non-linear (harmonic or intermodulation) distortion, but the true meaning is any change that makes the reproduction different from the original. Since THD and IMD have already been covered, I shall only look at other changes to the system that change the way the sound is heard, versus the way it was recorded.
From this perspective, it is quite obvious that all hi-fi systems introduce distortion (of the original performance). The ability to manipulate tone controls, graphic or parametric equalisers, or crossover network frequencies and/or levels may be used to reduce the distortion introduced by the simple act of changing the listening level.
Years ago, it was common for systems to have a "loudness" control, often incorporated with the volume pot. That these almost never worked correctly is a given, since no attempt was ever made to provide any method of calibrating the "effect" to the SPL in the listening room. Needless to say loudness controls went away, but with them, in most high-end systems, so too did tone controls and even the balance control. While the loss of the loudness switch went almost un-noticed, the other losses should be missed, since they are useful.
It is far more convenient to be able to change a balance control or tone controls, than to have to move furniture about or listen at a specific volume all the time. Consequently, it could be said that most systems are "distorted" most of the time, since what you hear is not the same as what was recorded.
The dream of a system that is 'dead flat' (in frequency response) is not necessarily a good idea, since it is likely that it will be far less accurate than expected for a great many recordings. While technical excellence is always a good thing, it is unwise to place that above all else, and even less wise to assume that ever greater levels of excellence will lead to ever greater enjoyment.
What of the average person (non-audiophile) who wants a really good system? There is precious little available catering to that market at present, and the choices are limited to the department store offerings (often with speaker boxes that look like something from outer space - and sound like it too!), or they can go and argue with the sales-thing at a high-end audio outlet, who will try to convince them that they don't want (or need) tone controls, balance controls, X-Bass or any of that "rubbish". Well, guess what? They do want them - all of them (well, most anyway). Will they use them with a good system? Probably not - initially they will all be used (usually radically), but after a while they will adjust and learn to listen to all of the music, and not just the bass. If it turns out that they do listen with the bass right up, then again, that is their right - it is their system, after all.
The purpose of a sound system in your home is for one thing only - to listen to music (and movie soundtracks, of course). Whether your system is optimal from a purist's perspective is immaterial. If you enjoy the sound, then you have a system that is perfect from your perspective. If others enjoy it too, then so much the better. The idea is for you to put together a system for yourself and (perhaps) your family - not other people. Should others in your household prefer a different sound, then there are two choices ... either you have two or more systems for different sounds, or use a system that can be tailored to suit each listener, or modified at will to suit classical music one session, rock the next, or tweaked further to suit the DVD currently playing.
The bottom line is that any sound system's ultimate purpose is for the enjoyment of music - not to follow the latest trend, or as a test vehicle for the latest magic stone, newest speaker leads or pure silver interconnects, but to enjoy music. Period.
Once you are in the position of listening for flaws in the system, you are no longer listening to the music - certainly you hear the sounds, but if you are so interested in the reproduction you can no longer concentrate on the music itself. You may use test instruments to analyse the response, phase alignment, spectral decay and distortion of your system, but if, after all of that, it still sounds good - who cares?
What of SET (Single Ended Triode) amplifiers and other systems that introduce often quite high levels of distortion? If this is what the listener likes to hear, then that is his prerogative, but it means that all music will be subjected to the same distortion. This may sound very nice with some recordings but revolting with others - again, limiting one's choice rather drastically. However, if this is what the owner wants, then fine. Such owners should resist the temptation to insist that others follow their path to 'nirvana' - most people don't want and don't need distortion. I consider SET amplifiers and their ilk to be "effects units". For those who like the effect, fine, but don't try to tell others that you have a hi-fi, because that's not the case.
|Copyright Notice. This article, including but not limited to all text and diagrams, is the intellectual property of Rod Elliott, and is Copyright © 2004. Reproduction or re-publication by any means whatsoever, whether electronic, mechanical or electro- mechanical, is strictly prohibited under International Copyright laws. The author (Rod Elliott) grants the reader the right to use this information for personal use only, and further allows that one (1) copy may be made for reference. Commercial use is prohibited without express written authorisation from Rod Elliott.|