HMV Model 328A

Concept

This project is a compact self-contained entertainment unit.  Its primary role is a high quality music player.

Speakers positioning is not negotiable for serious listening.  The only thing that can be done is to make them as small as possible and supplement them with a common subwoofer.  The positioning of the subwoofer is not critical.  If you want video, you need a screen at the front of the entertainment area, at a convenient viewing height.

That leaves the remainder of the system.  Ideally, you would combine all of these things into a single unit.  In the past, this was probably a giant cabinet with your turntable, cassette deck, video player, amplifiers and so on.  Now we can replace all of this with something much smaller.  We don't need lots of separate players, except for the purpose of transferring stuff from old media.  We could also include the subwoofer in the same unit.
7.1 Speaker Layout
7.1 Layout.  For stereo, angle should be within 30° - 45°, but most audiophiles lean towards 30°.

Design Goals

The system must be designed to be simple and convenient to use, with minimal controls.  The system should include:


Design Philosophy

I want the core system to last as long as possible.  With things like the cabinet and the audio electronics this should not be a problem.  I built an amp in 1968 which I still use!

One thing we can count on in the consumer media market is the fact that "standards" change.  If you buy a dedicated entertainment system such as a home theatre system or even an iPod, it will almost certainly become obsolete within a few years.  Basing an entertainment system around a standard PC seems to be by far the best option at this time, allowing for the upgrading of individual components as long as possible into the future.

Although this project uses PIC microcontrollers, in general I do not believe in using specialised components, especially exotic audio chips, which have a nasty habit of becoming "obsolete" and unobtainable within a few years.  I would rather use commonly available components where possible, even if this means an increase in circuit complexity.  This concept worked well for MCI, whilst many newer so-called professional audio products became useless within a short space of time, due to the unavailability of specialised parts.


Loudness Control

The human ear tends to be less sensitive to low and high frequencies at low listening levels.  Traditional hi-fi amplifiers include a loudness switch to compensate for this effect.  In low-end systems, this is just a fixed low frequency boost.  In more expensive systems, the bass boost is higher at low volume settings, reducing to flat as the volume control is turned up.

However, very few systems have true loudness compensation.  For a start, the overall listening levels need to be calibrated for the loudness compensation to work correctly.  In the late 50's / early 60's, some manufacturers experimented with elaborate switched controls with hand-wired networks to perform this compensation.  This was expensive and with the advent of stereo, most people conveniently tried to forget about the idea.  Today, if anything, suitable rotary switches are even harder to obtain and the number of steps are limited.

The idea here is to construct a digitally controlled FET switching network using cheap and readily available CMOS chips (4051, 4052 and 4053).  By carefully choosing the operating conditions of the surrounding audio electronics, it is possible to construct a high quality fully contoured loudness control with low distortion.  A PIC microcontroller is ideal for controlling the CMOS switches.


Subwoofer

Most medium priced commercial units I hear have a peak around 60Hz to accentuate the kick drum to make them appear to have a lot of bass, but fail to get to the lowest note of a bass guitar, let alone low frequency effects.  For that matter, most recording studio monitors don't get there either!

A well designed fully sealed enclosure is ideal for a subwoofer and will produce a smooth response, but it needs to be very large in most cases.  Build a concrete enclosure the size of a refrigerator and I promise you will have bass!  Ported designs allow much smaller boxes to be built to do the job, but they have limitations.  Below the tuned frequency of the system, there is no back-loading for the speaker cone and the speaker can be easily overloaded or modulate the higher frequencies - the last thing you would want for a speaker that could be exposed to very low frequencies.

For studio monitor or domestic listening situations, one solution is to put the subwoofer into an undersized sealed box, which will have a predictable low frequency roll-off of 12dB/octave at a frequency which you can calculate, then use an electronic equaliser to compensate for this.  The resonance of a large speaker in an undersized box will be relatively high, producing a peak at some higher frequency.  However, this can also be compensated for with the electronic equaliser.

The down side of this is that you require an enormous amount of power for this to work.  For example, if you require 20dB of boost at 20Hz to make the system flat, you require 100 times the amount of power needed to produce the same level at (say) 1KHz.  The exact relationship between the power requirements for the main speaker system and the subwoofer depends on the relative sensitivity of the two systems.  In practice, the energy levels at low frequencies in most recordings are relatively low, reducing the demands on the subwoofer amplifier.  As a rough guide, if you wanted to run your main speakers at around 30W, the subwoofer amplifier might need to be capable of handling 300W!


Some Thoughts About Amplifiers

 

Music Power

Most conventional amplifiers have unregulated power supplies consisting of a transformer, rectifier and some filter capacitors.  When operating close to their maximum power, the power rail voltage drops.  The continuous power is measured with a pure sine wave into the rated load.  However, music is not like this and has short term high level transients.  When a conventional amplifier is not delivering its full power, the power rails rise and due to the energy stored in the filter capacitors, is capable of delivering short-term bursts of power which are greater than its continuous maximum power capabilities.  With music, this gives the amplifier an effective headroom of maybe a few dB above its nominal continuous power rating.  It is for this reason that some amplifiers also have a "music power" rating.  This does not happen with amplifiers which have a regulated power supply such as a switched mode power supply.  In this case the music power rating equals the continuous power rating.  Similarly, a class D or PWM amplifier has an absolute 100% maximum level.  This means that such amplifiers need to have a higher power rating to match the effective power of an old school amp.  A few dB may not seem like much, but bear in mind that 3dB represents double the power.


How Much Power?

This is a difficult question to answer, due to the number of variables involved.  Sound levels are inversely proportional to the square of the distance between the speaker and listener.  Speaker efficiency (which is very low) varies enormously.  For example, I have a pair of B&W DM5 speakers which are so efficient they sound amazingly loud on a 2W computer sound card amplifier!  These were designed to work with valve amplifiers, where efficiency was a key priority.  However, most modern speakers trade off efficiency for other parameters such as size, since it is no longer a problem to build higher power amplifiers.  Speaker efficiency is usually rated as the sound level in dBA at a distance of 1M with an input power of 1W.  In practice the efficiency of different speakers varies by up to 10dB and this represents an enormous difference in power requirements to produce the same loudness level.  This makes talking about how big your system is in watts almost meaningless!

For PA applications, the amplifier and speaker power ratings are usually matched to minimise the risk of damage to the speakers.  For hi-fi and private studio applications, the sound quality is often improved by using a higher power amplifier.  This is for two reasons.  Firstly, the amplifier is running well within its limits and usually performs better.  Secondly, transients are reproduced better when the speaker is running close to its limits.  For example, if you combined an amplifier with a continuous rating of 30W RMS and a music power rating of 40W with a speaker rated at 30W, the maximum average level you could play the system at would be 40W less the peak to average ratio of the recording.  If the peak to average ratio was 10dB, that would mean the average power level could only be 4W (1/10th) if all the peaks in the recording are to be reproduced cleanly.  If you replace it with a 300W amp, you can now run the speaker up to an average level of 30W (8.75dB louder than before), with peaks going up to 300W.  Provided you do not exceed the continuous power rating of the speaker or its maximum cone excursion, a given speaker system will usually sound cleaner and more powerful with a bigger amp.  Naturally, you have to be very careful not to fry your drivers!


Why do Amplifiers Sound Different?

Real world speaker loads are not pure resistors but instead a complex network of resistors, inductors and capacitors as a result of the speaker cables, the crossover network, the speaker voice coil itself and so on.  But the story doesn't end there.  Speakers are also generators and you can think of them as very large diaphragm, very low impedance dynamic microphone.  In fact, you can connect one to a microphone preamp input and use it as a microphone, or connect a dynamic microphone to an amplifier and use it as a loudspeaker (although this is not recommended!)  If you rapidly push a speaker cone inwards, you will generate a voltage at the speaker terminals.  Therefore, mechanical properties such as the speaker cone mass, suspension stiffness and enclosure volume all form part of the complex speaker load.  Imagine a point in time when the speaker cone is rapidly hurtling forward whilst playing music.  As we just said, there will be a voltage across the speaker terminals due to this movement.  So what happens if the input waveform is suddenly reversed?  Due to the inertia of the cone, for an instant in time it continues to move forward, generating a voltage.  An output transistor suddenly has to deal with not only its own rail voltage (as would be the case with a purely resistive load), but also this additional voltage!  In a similar way, the reactance due to the capacitance and inductance of the speaker load can produce "out of phase" voltages which increase the voltage across the output transistors.  In a worst case scenario, we can assume up to double the amplifier's rail voltage.  Just in case, the diodes between the amplifier output and its power rails are there to prevent the voltage across the speaker load from exceeding this.  The output transistors need to be rated at a voltage of at least double the maximum rail voltage.  Similarly, the maximum current that a power amplifier needs to be able to deliver is at least double what you would expect by calculating the peak current into a resistive load at its rated maximum power.

These unexpectedly high current and voltage demands on the amplifier output devices can sometimes exceed their ratings.  A bipolar transistor output stage without V-I limiting will usually be damaged in the time it takes for a quick-blow fuse to blow if the output is overloaded or shorted.  V-I limiting usually involves the addition of two transistors to limit the operating region of the output transistors to their rated Safe Operating Area (SOA).  But this limiting often reduces the ability of the output stage to deliver the necessary short-term heavy currents to the load and many people (including me) agree that most amplifiers without V-I limiting sound better when operating close to their maximum power.  In domestic and private studio (but not PA!) applications, you might be prepared to take the risk and just be careful.

Many designers fail to take real world loads into account and perhaps this is one of the reasons why some amplifiers can sound so different.  Another factor is how the amplifier recovers from overload.  The output waveform when some amplifiers clip can be ugly!  Furthermore, the output protection itself can produce bizarre results when active.  This can produce artifacts into the sound when the amplifier operates close to its limit.  This reduces the useful effective power of an amplifier.  Using an over-sized amplifier as discussed above helps to reduce these problems.


Amplifier Design

A FET output stage without V-I limiting tends to withstand short-term overloads better and a well designed stage will usually safely blow the fuses if shorted without destruction of the FETs.  FETs have a softer "knee" characteristic and tend to have lower crossover distortion than bipolar transistor versions.  However, FET amplifiers require a relatively heavy bias current to achieve this and they usually run hot when idling.  This, along with Class A amplifiers which are even worse in this respect, represents an unnecessary waste of energy.  The other thing that turns me off using FETs for this project is concern about the long-term availability of FET devices and the reduced likelihood of finding a suitable replacement.  Although I have heard some very good sounding FET amplifiers, I am yet to be convinced that there is any inherent sonic advantage to using FETs.

It seems you cannot go past the simple conventional bipolar long-tailed pair differential input stage followed by a Class A voltage gain stage feeding a Class AB output stage.  However, for higher performance and higher power designs, the simplest form of this design starts to put demands on the transistor specifications.  It would be tempting to throw in a current mirror to increase the gain of the differential pair, or a buffer to increase the drive power of the Class A stage, or an extra driver transistor for the output stage to increase its current gain.  But sometimes this can result in too much overall gain, leading to potential instability.

Interestingly, some of the better high power bipolar amplifiers I have heard have either series output transistors or operate in bridged mode.  In both cases, this doubles their slew rate capabilities.  This suggests that the speed of the output transistors is an important consideration in larger amplifiers.  With lower power amplifiers, it is not as difficult to achieve the necessary slew rates to produce a clean sounding full range amplifier.