Thursday, 7 April 2011

How much power does an amplifier need

Its obvious from posts one sees on various forums that most people have no idea how much amplifier power is needed to reproduce modern recordings properly and neither do they understand the relationship between Decibels and Watts.
This is understandable given the nature of magazine reviews and because so many advertisers amplifiers simply don’t have enough headroom. TVs for example have about 12 WPC and many expensive so called specialist hi fi products are still well below 100WPC.
Most speakers are around 87-89 dB/W/M and this represents a loudish listening level with pink or white noise, Heavy and Death Metal or simply a recording of a Black and Decker drill. Continuous sound appears much louder than music.
Music is very different and a typical continuous listening level is more likely to be 80 dB, which means 1/4 of a Watt! The problem is that this 1/4 Watt is interspersed with enormous peaks, some requiring hundreds of Watts and if the amp can’t produce them, it’ll clip, and ruin sound quality. 
Speakers are not a fixed load either, at low frequencies impedance rises to 30 odd Ohms and in the mid, where all the musical energy is, it falls to about 5 (assuming 8 Ohms nominal impedance) and as frequency rises, so does impedance and much more so as the crossover is approached.
This means that an amplifier must produce more Volts when impedance is high and more current when it’s low as well as continuous flow of peaks that may run into the hundreds of Watts for the music. 
The solution, where high quality sound is required, is an amplifier that is capable of several hundred Watts per channel as are our ADM9Ts.
In simple terms, if 80 dB is an average listening level, the system needs a good 25 dB extra in reserve for the peaks and to cope with the varying impedance of the load. If higher volume levels than normal are required, say 90 dB, then the amount for headroom must be added to that. It all sounds pretty simple until you realise that doubling amplifier only gives you an extra 3 dB, which is why so many Watts are required to properly reproduce modern recordings.
To Sum Up
A typical average requirement is about 80dB, at this level various music instruments including drums, pianos, even human voice produce instantaneous peaks that may require hundreds of Watts and if an amplifier can’t produce them, it will distort or clip trying. Therefore an amplifier should be able to produce at least 25 dB more than an average listening level to avoid problems. This means approximately 400 times the power or 100 Watts. However many enthusiasts and professionals require far more because they may like to listen at live or at least higher levels from time to time. 
A concert grand piano produces 110 dB at a metre, which equates to 400 Watts and a Snare or Tom Tom 139 dB! Think about it.
In our opinion 100 WPC should be considered a minimum power requirement for average domestic listening levels and 250 WPC for serious enthusiasts and professionals and it is important to understand that these power requirements are recommended to improve sound quality, not necessarily to increase sound pressure level.
Interestingly AVI measured music (as have Crown the US amplifier manufacturer) many years ago to establish these parameters, we even appeared at a major hi fi show with a special device that showed precisely the requirement and we demonstrated it to most hi fi journalists of the day. Although they were obviously surprised, none ever mentioned this in a magazine and to our knowledge, no reviewer has ever used a Scope to see if an amplifier is clipping when he’s listening to it. Lamentable but true.
We found huge variations in the dynamic range of music, one track in particular that didn’t sound very loud required peaks that equated to 600 instantaneous Watts when the average listening level was only slightly above normal and the continuous average power requirement 600 milliWatts. In our opinion Watts are misleading where decibels are not. The enormous peaks that we’re describing in decibels are actually volt/amp peaks that average out to a fraction of a Watt.  

Saturday, 11 December 2010

A Third Way

Over the last thirty years speakers have evolved into two basic types, two way and three way. The former are by far the most popular because they are smaller, they only have one crossover and it is high enough up the frequency range not to be overly intrusive. 
Three way speakers have more problems and tend to be less well liked, unless they are in a large enough room not to boom and for listeners to be a good distance away from them. This is because they have a second crossover in the region of middle C, which is very intrusive. They harshen sound  and the phase distortion they produce causes discomfort.  Also passive crossovers are very lossy and make for boomy bass. 

Big passive three ways struggle to avoid boom and tizz and even small passive two ways have intrusive crossovers as the tremendous success of ADM9s has shown. They don't suffer from the unpleasant artefacts produced by a passive crossover.

We've called our forthcoming generation of speakers the Third Way because they are different and better.

1. ADM40s are floor standers approximately the same size as ADM9Ts on stands, each housing a purpose built 10" Sub driver in a sealed volume large enough to allow useful output down to 30Hz. 

2. Because the bass driver in the ADM40s crosses over to the mid at 100 Hz rather than 250-400 Hz in older designs, artefacts that are so intrusive at higher frequencies are inaudible and merging is seamless. 

3. This low crossover point means the ADM40s can have an unfiltered Low Frequency Effects input for an AV processor and a gain control (from handset) to adjust normal stereo bass levels to suit different room acoustics and or recordings. 

4. At the same time, applying a crossover to the lower end of the 6.5" mid driver reduces its cone excursion and intermodulation distortion, which in turn allows a worthwhile increase amplifier power for more dynamic range. 

5. ADM 40s will have a total RMS power per speaker of around 750 Watts and be able to produce instantaneous peaks of an incredible 3000 Watts or over 120 dB! To anyone considering one of a recent proliferation of "one box, audio only solutions"  with measly 25 Watt amps, this may seem ridiculous, but it isn't at all because modern movie and TV sound tracks (and the best music recordings) all have a dynamic range and bass extension that needs it! 

Needless to say this doesn't mean ridiculous SPLs are the norm, although ADM40s will play as loud as a Wagner Opera or AC DC without distorting if you want them to. Much more important is that that they avoid the inevitable (and irritating) congestion/distortion that lower powered amps produce on loud passages. With adequate power everything appears cleaner and more clear at all levels including normal family ones. 

Our next blog will provide a much overdue explanation of the need for more power where high quality sound reproduction is the objective.

None of the above is practical with big old, three way speakers, so if you buy a pair of them and they boom when you get them home, your money is wasted!


By combining the best of modern subwoofer and two way speaker technology in a compact active, three way, AVI have produced floor standing speakers that have the most dynamic range of any available, the cleanest, purest sound you will hear and without any ugly metal boxes or even a separate sub. 

ADM40s  will easily outperform more expensive, old fashioned three way speakers several times the size and price.

Tuesday, 6 April 2010

Passive Crossovers and the Impact of Amplifier Output Impedance

In previous blogs I've referred to the enormous amount of distortion introduced by passive crossovers and in this one I show how a power amplifier can react with a loudspeaker to change the way it performs. A good quality, competently engineered loudspeaker can have its performance sabotaged by a power amplifier.

If you model second order passive crossover as there may be in a typical two way loudspeaker assuming a typical amplifier output impedance of 0.01 Ohms and then use an amplifier with a higher output impedance, say 0.3 Ohms, you'll find that the bass driver crossover frequency moves up and the tweeter crossover frequency moves down, producing an overlap in the crossover region. The higher the amplifier output impedance the more pronounced the effect.

As an example, using speaker with a nominal impedance of 4 Ohms and a classic 2nd order crossover at 2.3 kHz, increasing the amplifier's output impedance to 0.3 Ohms gave a bass crossover point of 2.4 kHz and a tweeter one of 2.2 kHz (being defined as the frequency at which 90 degrees phase shift occurs). Therefore it is likely that higher output impedance amplifiers will sound noticeably harsher, because of the extra energy produced in this region compared to a conventional amplifier.

In a recent review of amplifier with an output impedance of 0.27 Ohms, the reviewer also noted significant frequency response variations due to the interaction of the amplifier output impedance complex impedance of the speaker load.

We've measured a good quality 15 Watt valve power amplifier with slightly better than 0.1% distortion and it exhibits an output impedance, via the 4 Ohm tap, at 100 Hz  of 0.28 Ohms, at 3 kHz 0.4 ohms and at 10 kHz 0.9 Ohms. Not only will there be even worse overlap at  the crossover frequency, but also a loss of HF output (nearly 2 dB down at 10 kHz).

Single Ended Triodes would be much worse.

Saturday, 3 April 2010

How To Assess Hi Fi or Studio Monitors?

For the last few years hi fi fanatics, especially those that might be described as flat earthers, have been in denial. They really don't like Apple and iPods, despite the fact that they are probably the only reason that legacy separates systems haven't disappeared altogether. Many commentators are of the opinion that the Apple thing has re-awakened interest in hi fi.

However, so much vitriol has been directed towards poor old iPods and the AAC downloads of which 10 Billion have now sold, that many don't realise that with good headphones, or even the standard ones forced into your ears so that you get some bass but are uncomfortable, are probably the best sound quality most of us will ever hear. iPods have extremely low levels of distortion and can provide a level of clarity and neutrality not possible, even with the best hi fi system. Therefore they make a very good portable reference that you can use to help you judge the quality of any system you may be considering. They will not have the stereo image or the three dimensionality of speakers, but in other respects they are very useful indeed.

Which iPod
I don't think it matters, the later ones certainly sound better than previous generations and other, better quality PMPs are just as good. I use a Touch, which is an extremely useful computer as well as having an excellent remote control application in it for my Apple TV, but I recently heard a Nano which was at least as good. I'm sure a Shuffle will be too. My Macbook Pro is excellent as well, but whenever I've tried PCs they've sounded poor by comparison.

Which Headphones
Differences between good headphones are far smaller than between speakers which vary enormously and even more so if sub 100 Watt amplifiers are used, especially with 4 Ohm speakers, so choice is far easier and it's not necessary to spend a fortune. I use Sennheiser HD25-2 Professional Monitoring ones, they are absolutely superb and have a 70 Ohm impedance that suits PMPs. However others may prefer more discrete in-ears ones from Shure or Etymotic. The main thing is that they couple well to your head so that you don't lose bass as you do with standard ear buds.

Choosing Appropriate Recordings
The purpose of Hi Fi or high fidelity equipment is to more nearly or, as accurately as technically possible, replay what has been recorded. A Piano or voice should sound like the real thing, a group of instruments should spread out in front of you, remain in position and you should be aware of the acoustic characteristics of the venue where the recording was made. To someone who loves Classical Music, Folk, Jazz, World Music, even live Rock Bands, the comparisons are easy to make and people do successfully all the time.

However modern music is a free for all, sound quality is not an issue, but selling it fast is and certain characteristics are often exaggerated to make it sound as loud as possible so that it stands out from other tracks being played on the radio. Adverts between TV programs get similar treatment and it's called compression; the quiet bits are made as loud as the loud bits. Because it then appears louder but flat sounding, more treble is applied and more bass too, not to mention various other electronic tricks to make it "exciting". The net result is that a lot of it doesn't sound much better on a good system than an ordinary one and often it appears bass light because the treble is so brutal. None of this means you shouldn't enjoy it and collect it, but you do need to bear this in mind when assessing hi fi for possible purchase. You need also to use music with a real sound and proper musical instruments as well. If something live has been recorded to sound as it did on the day and the system does a very good job with it, it is a good one, but if you are impressed by the sound of purely electronic music without real references in and then find the real stuff sounds wrong, be wary because it probably indicates fatiguing distortion that will ultimately spoil all enjoyment.  An old Ella Fitzgerald recording is very much more useful as a reference than something modern and mostly electronically generated.

In previous blogs I've explained the benefit of better sound quality to minimising fatigue, but another is that is with an accurate and neutral system of the highest quality, you can enjoy a wider more diverse collection of music and movies without discovering irritating distortions that flatter some material while ruining everything else. I remember once a sound engineer friend explaining that his particular monitors did wonders for Saxophones. I asked if they made Pianos sound like Saxophones as well and he laughed and agreed they did!

I do hope this helps.

Ashley James

Thursday, 1 April 2010

How We Hear Stuff

The above Fletcher-Munson Curves  show that, unlike speakers, our ears do not have a flat amplitude response, in fact they are not very sensitive at low frequencies and don't become particularly so until 200 Hz or just below Middle C and from then on it increases until about 5 kHz and then decreases again. The best explanation for this is that we're designed to understand speech and do so by filtering off the high frequencies to discern clarity and intelligibility. Voice has a range of approximately 80 Hz - 3 kHz, but we listen mostly from 250 Hz-3 k Hz. We hear what people are saying by using our ears to locate the sound source, focussing them on it and then singling out the sizzly bits to understand what's being said.

In my opinion this shows clearly that we expect clarity and intelligibility from whatever we are listening to and amplitude is just one of several factors that enable us to achieve this. However as we age, so our sensitivity to higher frequencies deteriorates and by thirty five an average male's hearing has lost approximately 10 db (reduced by a factor of ten!) at 3 kHz. Although he may not realise it, he will be finding it more difficult to hear what people are saying or to discern clarity from his hi fi.

Our ear canal produces distortion, tests indicate this to slowly rise to about 8% at 105 dB and we use this to tell us how loud something is, but as we age and our hearing deteriorates and it too is distorting, so we tend to perceive sound as louder than it is and find it more difficult to tolerate. We need clarity at low levels if sound/music is to be relaxing to listen to.

We also have a sort of automatic volume control that can turn down the ear's sensitivity if sound is too loud for us, which is why young people are warned not to listen to iPods at high levels. The process is called temporary threshold shift and it can become permanent if we are exposed to high SPLs for prolonged periods. It probably accounts in part for thin, bass light recordings of some older rock music of which Led Zeppelin might be an example. The producer may have been playing much too loud and so added treble or removed bass to  compensate for threshold shift.

Anyone can test their hearing on the net these days and it might be a useful exercise to compare yours with a notional average and to get an idea of how well or poorly you hear. Women tend to be about 6 dB more sensitive than men and a 10dB drop is colossal, so it's not surprising that young people are more likely to tolerate distorted sound than their elders or want more bass to lessen much stronger (to them) treble.

The Brain
The ears are merely a means of gathering information, which they pass to the brain for processing, so the better the information they are able to gather, the less the work the brain has to do for it to be intelligible. It's hard work listening to indistinct sound.

Sine Wave Speech tests show how the brain is able to process quite distorted sound to get what it needs from it. If you do the test, play the sine wave speech first, then the proper recording and finally sine wave speech again. The first time you hear it, it will just be an incomprehensible noise, but once you know what has been said, it becomes intelligible. I'd guess this is the process of "running in" hi fi that the subjective brigade sells.

Sine Wave Speech probably explains more than most realise. For instance it could explain why most people are convinced they have the best hi fi and that nothing else is as good, or that reviewers who describe a system in very negative terms on first acquaintance, but having improved after a period of "running in", or why it is that often when people are offered significant improvements, they don't always recognise them. They have subconsciously "programmed" themselves to hear music through their system and need reprogramming before they see benefits of another.  Good headphones and an ipod are a tremendous help in providing an accurate reference, which is why professionals use them to listen out for detail speakers miss.

The ears, first and foremost need clarity, but the brain can process a poor information from them to achieve the same result. Clarity becomes more of an issue as we age because of lost HF sensitivity, which means we have to concentrate harder to hear whatever we are listening to.

Therefore hi fi, which is probably part of the process of relaxation, needs to be clear and probably at lower levels than in real life for maximum enjoyment.

It follows that adding treble by tone controls or through distortion in the replay system may also add clarity, but often it just makes things harsher and less bearable, which may explain why some hi fi enthusiasts think they don't like clarity.


Monday, 22 March 2010

Active versus Passive Speakers THD.

In my previous blog I explained the various different distortions that occur in passive speakers and which made active ones so much better. In this one I will present measurements made of both an active ADM9.1 and a passive version driven by a low distortion Class A test amp.

Both speakers were measured at a nominal 86 dBa at 0.5 Metres. The Mike was a Bruel & Kjaer type 4165 measuring microphone and the source was the low distortion output of an HP339.

The above is a distortion plot of the active ADM9.1 at 100 Hz and it shows the second Harmonic at -48 dB and third Harmonic at -50dB just visible above the noise level in the factory.

The above shows a passive ADM9 at 100 Hz where the second harmonic has remained at the same level as the active version but surprisingly the third Harmonic has increased by 12 dB and the fifth Harmonic, absent from the active version, is at -45 dB.

 The above shows that at 3 kHz (in the region of the crossover) second and third Hamonics are almost lost in the noise at -55 dB.

The above shows the passive version at 3 kHz, the second Harmonic is now -46 dB and the third Harmonic at - 48 dB.

Saturday, 20 March 2010

What Constitutes a Good Drive Unit?

The Picture above is of the 5" drive unit we use in our Neutron Five loudspeakers. It is excellent and typical of the best available today.

Below are measurements pinched from a DIY forum to illustrate how they compare with an extremely expensive "high end" driver from another manufacture.

Plot one shows a smooth amplitude response to way beyond where a crossover would go and plot two shows  harmonic distortion.

Plot three shows a much more expensive "hi end" driver from another manufacture, which is remarkably similar in amplitude response and plot four shows that it is in harmonic distortion as well. The reality is that either drive unit will give excellent and similar results. However there are other characteristics to consider that will also affect performance.  For example; power handling and maximum linear travel, where ours still compares well. 

What these measurements show is that the performance variations between the between modern paper diaphragm drive units from reputable manufacturers are comparatively small,  though nothing like as small as those between DACs and other electronic components.

There is however a divergence of opinion of what constitutes the best diaphragm material and many prefer  hard or metal ones. 

The above illustration is of a good quality 5" metal diaphragmed driver, which has a smoother amplitude response to 4 kHz, will probably have slightly less harmonic distortion, but which breaks up catastrophically at higher frequencies. AVI doesn't like them and wouldn't use one, but others prefer them and use an extra filter in the crossover to reduce/remove the spike. We don't like this either because it is further distorting phase. 

In our opinion the paper or soft diaphragm is the best compromise because it acts as a shock absorber to suppress and iron out the tendency that cones have to break up above a certain frequency dictated by their size. I think 1.4 kHz for a 5".

These differences aside, all high modern drivers have relatively shallow cones to aid dispersion, large diameter spiders to minimise distortion, they are spaced off the front plate by about 10-12 mm with ventilation underneath to avoid resonances and they have longer coils than the front plate to give lots of linear travel. Ours has + or - 5 mm which is very good and double that before there is risk of mechanical damage. Some may remember the "clack" "clack" of the spider hitting the front plate of an old Kef B110 in LS3/5As!

This last illustration shows a large diameter and compliant spider and ventilation through to the voice coil underneath. This is typical of almost all modern, good quality drive units and has been for some years, although small incremental improvements take place all the time.