10 Steps To Excellent Sound

I've spent over 20 years passionately involved in audio both professionally and as a keen DIYer / hobbyist. I've learnt much about the art of audio during this time and I think my personal system can attest to that. Here I've presented my list of the ten most important things needed to achieve a truly transparent audio system. HiFi or High Fidelity has become a much used but rarely achieved goal in audio. To me, High Fidelity means true to the original, in every sense. This sounds easy in practice, but in reality, everything must be done right and this is actually very rare. Have a read of my 10 Steps below and have a think about the ramifications of these statements. I believe my current system is pretty much there, except for #1 and #5. I miss my target at the very lowest frequencies, despite having 1600W RMS powering eight high excursion 12" drivers.

  • THD and IMD< 1% at all normal levels and frequencies
    This should keep distortion down to inaudible levels. I've seen 3% as the level where THD starts to become audible, but this does change significantly with the order of the distortion, higher orders being audible before the lower orders. This spec most likely refers to the latter.

  • Thermal compression less than 1dB at all normal levels
    Lots of voice coil area and well designed voice coil cooling is required here at low frequencies. Higher up high efficiency can be a great benefit. When attempting to reach high output levels, a little dome tweeter or small voicecoil on a 10" woofer is just not going to be able to provide low enough thermal compression to be inaudible. Thermal compression will rob the system of macro dynamics.

  • SNR greater than 90dB
    This specification appears easy to achieve, but in practice is quite a challenge. Attention must be paid to the gain structure in the audio chain to ensure the SNR available from the electronics is achieved in practice. All stages in the signal path should approach clipping at about the same time. Any deviation from this ideal is simply robbing the system of precious dB's of SNR. A 90dB SNR gives a noise floor of 20dB SPL in a system that is capable of 110dB output. This should be at or below the ambient noise level in most domestic environments. As the output capability of a sound system increases, high SNR becomes even more important.

  • Response flatness +/-2dB from target above 300Hz, +/-5dB below 300Hz at all listening positions
    Pretty self explanatory, correctly designed equalisation is your friend. Correctly executed eq also has the advantage of undoing any phase errors within a drivers passband. A large well designed room and controlled directivity can also help.

  • Output above 110dB at listening position
    This should assure sufficient output for movies or dynamic music. Remember that a 4 meter listening distance will reduce the output from most speakers by up to 12dB compared to the 1 meter output. This requires more than 122dB output at 1 meter which is a tough challenge for most speaker systems. A true line source will only reduce in output by 3dB for each doubling of distance, making this spec easier to achieve in large rooms. Very few loudspeaker systems can claim true line source behaviour at all frequencies, they are simply too short.

  • Controlled Directivity and/or a very well treated room
    It's important to keep reflections in front of the listening position to a minimum. If reflection points exist, they can be removed with intelligently placed absorbent panels.

  • Smooth power response or anechoic room
    The off axis response must be very close to being the same as the on axis response, or any reflections off the walls making their way back to the listening position will have a different spectral balance and will ruin the sound. Very absorbent side walls can reduce the effects of a loudspeaker with poor off axis response. The ceiling and floor reflection points must also be dealt with. It can be helpful to use a mirror to help determine exactly where the reflections are occurring.

  • No reflections closer than 10mS from direct signal
    The ear has trouble differentiating reflections that are very close in time to the direct signal. 10mS to the first reflection will ensure these reflections don't adversely impact the sound. As the time to the first reflection becomes longer the brain

  • Change in frequency response of less than 1dB with change in output level
    This is closely tied to the thermal compression issue. It's a sad fact of life that many systems change their response significantly with changes in output level. It's always very telling to do a response measurement at 110dB(1M) and compare it to the response plot at 90dB. Differences of 5dB or more are common. As voice coils heat up, their DC resistance can almost double. This change in Re causes a huge shift in driver parameters and de-tunes your carefully aligned boxes. The higher order boxes (Bass Reflex, Band Pass etc) suffer more than the low order alignments such as the sealed box.

  • Bandwidth from 20Hz to 15KHz
    This should cover all audible input signal for most of us.

    Now the controversial part.

    Having a look at this list you might notice that most of the problems highlighted are room and loudspeaker related. It's a sad fact of life that this is where most of the non linearities in a system exist. Apart from SNR there is really nothing that's electronics related, and even that has more to do with the application of the electronics than it's actual design. Even the cheapest CD/DVD source will be "blameless" in even a high end sound system. VERY FEW components can't achieve 20Hz-20KHz, 0.1% distortion and 100dB SNR. If they can't, they don't have a place in ANY high end sound system - PERIOD. If any component will colour the sound AT ALL it is not worthy for use in a high end sound system.
    Some of you might think "where do all the fancy electronics, valves and high end cables fit in?" Well, IMHO, the fancy electronics and cables are there to make you feel good about your system, and they do. You are simply kidding yourself if you think a $500 cable pair will sound different to an interconnect using Neutrik connectors and good quality Belden microphone cable, or similar. As long as the connectors are clean and sound, the cable capacitance is low enough not to impact the top end and the shielding is sufficient not to impact the SNR, there is no benefit spending more money on the interconnects. This money will have a far greater effect if spent where it counts, on the speakers and room. I maintain that most of the budget should be spent on the speakers and the room they are in. The end result will always be better than dumping a pile of money on the electronics and signal path. Valves are another can of worms. Sure they can give you a "warm glow" but to base a high fidelity system around a valve amp is a real challenge. I have not yet heard a valve based system that was not coloured. Lacking in dynamics usually describes them too.

    My Perspective on audio is from a purely engineering based perspective, however I can see where tweaky stuff has it's place. It helps the local audio shops that we buy CD players and electronics from keep their doors open. There is not a lot of profit in main stream electronics these days and the retailers need their little golden egg.

    The funny thing is it's actually pretty easy these days putting together a reference class audio system for not too much money. To tick all the boxes in my 10 steps (above) all you would need to do is stick a pair of, say, Danley SH50's in the corner of your room, power them with a simple 100W RMS per channel amp, team the lot with a sub that can match the output of the SH50's (Horn, IB, multiple sealed or vented boxes) and you're done. The Danley boxes come to mind because they are one of the few well engineered full range loudspeakers on the market that can meet my output, distortion and directivity specifications laid out above. Another loudspeaker that would provide what I am after is the Gedlee Summa. If the amplification and crossover are from one of the many pro sound manufacturers (many honest well priced products), the entire system could be put together for probably $10K and rival the monitors in most recording studios. The biggest hurdle would be to make it all look pretty in a domestic environment, but as I've proved with my Unity build in, even that's possible. Please note that I have no affiliation with Thomas Danley or Earl Geddes, I do however have great respect for their electro acoustic engineering ability.

    If price is not an issue and you can't live with the poor aesthetics of the above choices (And don't want to hide them behind a screen) the Wisdom Audio l150i is probably the best option available. This system solves several acoustic and aesthetic challenges.
    Back to my home page