Wednesday, 28 May 2014

Setting the scene for success: Designing a demo system - Chapter 3

In the last chapter, we made an overview of the SPL needs. Now we can have a look at the Home Cinema sound system configurations.

A professional approach to sound is often disturbing the typical Hi-Fi salesman’s or consumer audio professionals.

It only takes to read a so-called « technical » brochure about a loudspeaker to understand how deep is the problem: You will find supposedly meaningful words like “rhythm”, “pace”, “precision”, “fast”, but nothing that will really help the installer to design a system, like:

    • Directivity response vs. Frequency either polar or Matlab
    • Directional gain
    • Maximum SPL @1m without cheating (like when forgetting the thermal compression)
    • Frequency response at maximum level (generally different from what it looks like at 1W)
    • Distortion curves at various levels (at least 10V)
    • Waterfall response diagrams
    • A CAD drawing

Without this information, the system designer is left to either do an approximate job, or to perform all these measurements himself. This is particularly critical when matching the loudspeaker to the room acoustical environment and the audience geometry.
Such information is compulsory and most commonly provided with professional  (I mean here: non-residential) loudspeakers, as sound engineers simply do not accept working without it.
I have to say, strangely enough, I never found a directivity curve provided with any loudspeaker aiming at the residential market.
Now, if you start believing that such a concept as directivity control has never been considered in the design of residential loudspeakers, I cannot really say that it is 100% true.
Maybe only 99%... (I'm kidding ;-) )

I am not suggesting here that most residential loudspeakers are ill designed; there are good choices around. But in many cases, you will have to pay for a nice veneer or a glossy lacquer when your speakers will end up hidden behind a screen!

And the sound system designer is left alone without the proper information…

  • Unbalances

A Home-Cinema sound system comprises much more components than a Hi-Fi system, needless to say.
So it is easy to understand why the most demanding customer, willing to get only the best available components for each function of his home cinema, will end up postponing the purchase of his new Bugatti Veyron.
For the vast majority of even affluent customers, there is a limit to the budget, and hence compromises.
 To illustrate this, let me first describe a typical Hi-Fi system. It comprises:

  • 1 CD player
  • 1 vinyl turntable (optionally)
  • 2 analog line-level cables
  • 1 integrated amplifier
  • 2 loudspeaker cables
  • 2 loudspeakers

Diagram 1

Now if we describe the audio inventory of a basic Home Cinema sound system, there must be at least:

  • 1 Blu-ray player
  • 1 HDMI cable
  • 1 digital audio cable or a 2nd HDMI cable 
  • 1 AV receiver
  • 7 loudspeaker cables, at least 100m total
  • 1 line-level audio cable
  • 3 front loudspeakers
  • 4 surround loudspeakers
  • 1 amplified Subwoofer

This, of course, is not including the video part of the installation.

Diagram 2

And for a more sophisticated installation, we will find:

  • 1 Blu-ray player
  • 1 AV server
  • 1 switcher
  • 2 HDMI cables
  • 1 power conditioner
  • 1 preamplifier-AV processor
  • 8 audio line-level cables
  • 2 DSP crossovers
  • 12 XLR balanced audio cables
  • 12 channels of power amplifiers
  • 2 Subwoofers (passive)
  • 3 bi-amped front loudspeakers
  • 4 passive surround loudspeakers
  • 12 loudspeaker cables, total at least 150m

Diagram 3

It is now easy to understand how important the cables budget could be, and that esoteric audiophile cables are to be discarded (in addition, their stiffness and diameter can become a nightmare for the installer).
It is also easy to understand that mono-block valve amplifiers weighing 50 kg each and rated 12 Watts are not appropriate.
So you will have to forget about spending days doing A/B comparisons between single components or cables.
An HC system needs to be optimized in the perspective of its total budget, even if large.
So, in this case, optimization means balance. Let me get to the point:
It is known that what you hear in a system is its weakest link. Balancing a system means changing the weakest link to a better one, until there is no weakest link at all.
Now, where do you start?
Well, you must have an opinion about components quality.

This is where the image becomes plain white and the sound an unbearable silence…

How to evaluate the respective quality of the various elements in a complex system as pictured above in diagram 3?
I’m sure my answer will disappoint all fanatic readers of AV magazines: You need professional experience.
If you read tests in the medias, you will never find anything that justifies the 1 to 20 price ratio between components which are all rated as “excellent”. So, how will you select your components?
A tiny minority only selects world-renowned brands, top class ranges, and very, very expansive stuff. Their customers are the like who will not even have to postpone the Bugatti Veyron purchase…
Others (they are more) spend their time on forums. Actually, they don’t need to get involved in HC installation, they do not have time for this as they can spend all days on the internet.
Some others rely on 3 letters products certifications which prove that their suppliers have paid the flat fee plus the royalties (it’s funny, all certifications have 3 letters labels! none is for free)
What I consider as the best approach is to rely on your own experience with the equipment, as you don’t have the time to spend your life testing and comparing components. As a professional, you have installation jobs on your schedule.
However, you need to get a new component under test from time to time, otherwise the risk is to become outdated.

When designing video systems, things are crystal clear: You choose a screen (and Excellent one, preferably) and a projector after checking their compatibility (lumens vs. dimension) and then you calibrate the projector, play an image and evaluate the result.

When designing a sound system, it is not so simple

An HC sound system is involving quite a few components, and the complexity of the system is an exponential function that has the number of components as exponent. Then, the results are audible instead of visible.  This makes a real difference, as the sound is by definition vanishing immediately, whereas an image can be made steady.
There are some measurement methods which are rigorous and reliable, fortunately. Again, they are better known in the professional audio industry than in the residential one.

  • The mains supply

Considering the number and diversity of devices connected to the mains, it is quite difficult if not impossible to have a thorough evaluation of the power supplies of each element.
A well designed power supply will not incur any problem even with a relatively unsteady or polluted mains supply.
However, many electronic devices, even in the “high-end” ranges, suffer from dependence with respect to the mains supply quality.
This quality varies with the location of the installation (it is generally better in cities than in the countryside) and with the time of the day.
Therefore, it is wise to be careful about the electrical part on the installation, not only for safety reasons, but also for sound quality.
A few things to do:

1)    Draw a direct line from the mains connecting board. Only audio devices must be connected to this line. Any other electrical appliance, especially light control devices and/or machines comprising an electrical engine have to be discarded.
2)    Check the grounding quality (< 10 Ohms preferably)
3)    The cross sectional area of the conductors must be sufficient to handle at least twice the total maximum power consumption of all connected devices.
4)    Use a power conditioner providing sufficient available power for at least all devices that are operating at line level. When estimating the necessary power. Provide a large headroom.
5)    Install a surge protection device.

As far as 4) is concerned, typical « audiophile »  sockets strips are not providing sufficient available power and are sold at a price that can be justified by the amount of silver, gold or complex engineering, but not by the insulation they provide.

Ideally, the power amplifiers should also be connected to a power conditioner. However, their power consumption requires a very powerful, bulky and expansive power conditioner (see image below). For this reason, such devices are not common in Home Cinema installations.

 22 KVA mains conditioner

  • The sources

Many audio/video sources provide a decent quality at very affordable prices. This does not mean that you should use a Blu-ray player sold at £99.99 to feed a £100,000 installation !
There are typically 3 types of A/V sources :

Ø  Blu-ray players

These devices are the simplest to use and to install. They are also the most affordable ones. Still you need to checK
  • HDMI standard (which generation ?)
  • The nerve-wrecking duration of the starting process
  • The mechanical sturdiness
  • The image and sound quality, by comparing it to the competition 
  • At the moment, one brands seems to have a leading edge (check our stand at the past 2014 ISE)



One advantage of  HTPC  devices is the  « all in one » aspect. You can record blu-rays, download video, shortcut lousy legal announcements, get a fast access to recorded material, in short, happiness!
Either you are a true Geek, and configuration problems, compatibility issues etc. are peanuts for you,
You know how to feel like Mr.Everybody and design interfaces that even your client’s grandma will find funny to use
You do not have both of these skills, and you are heading to a real nightmare.

Ø  Audio/Video servers

Audio / Video servers are a bit like HTPC, but they are pre-configurated and (nearly) ready to use. They are seemingly the ideal solution, combining a user-friendly interface and a huge storage capacity (count in Terabits).
In fact, most still suffer from two limitations:
-       Very high prices
-       Or outdated performances (Full HD was not available on some servers for quite a while, for instance, or no lossless sound format)
-       Or sometimes both problems

So, the remarquable comfort of use of these AV servers has a price   
The choice is yours…

  • Decoder-preamplifiers

For once, this is easy: Since lossless audio formats are available (DTS Master and Dolby True HD), you don’t really feel like listening to anything else (in Home Cinema, I mean).
So you first need to check that decoders support both these formats.
Other things you will need to check are:

  •   HDMI quality and stability. It sometimes “bugs”
  • Menus clarity and ease of use
  • Clarity of the various functions and configuration, especially the ambiguous “Bass Management” and “LFE Mix”. You definitely need to understand what the device is doing, otherwise you will not be able to achieve what you want. 
  •  In the case of integrated “receivers”, the availability of pre-out connectors is essential. It will allow you to insert control devices, like a digital crossover or an equalizer.
  • Availability of a manual “lip-sync” adjustment.

Now, if your preamp is provided with an automatic equalization function using a measurement, do not worry. What you only need to do is to bypass this automatic function.
Anyway, never use it !
We will see why in a further chapter about EQ.

Now that the first elementary requirements are met, what makes a difference between AV pre/pro is the sound quality.
I do not mention here the video quality, as I strongly believe that the video signal of the main source should be connected directly to the video projector. You will need 2 HDMI outputs from your source, as one will be sending the audio signal to the decoder.
Checking the sound quality is something, I believe, you do not need to be advised on. You’ll make your own evaluation.

  • Amplifiers

The amplification of a Home Cinema system can be complex, as there are quite a few amplifier channels involved.
The above diagrams #2 and #3 illustrate this.
According to the setup on diagram #2, 8 channels are needed, and on diagram #3, 12 channels.
People used to 2 channel audio may think it is too much, but this is only because they are so used to traditional Hi-Fi that they are reluctant to get into the logic of real AV sound systems.
So, rule number one is like when scuba diving: never panic!
Looking closely at diagram #3, you will see that the loudspeakers are bi-amped. This means that the HF amplifiers do not necessarily need to be as powerful as the LF ones.
It is your choice: If you are budget-oriented, you will use less powerful amps for feeding the treble channels and the surround channels, whilst using the most powerful channels for L,C,R bass channels and LFE.
If you are simplicity –oriented, then you will use large power amplifiers throughout.

In the setup pictured at #2, we can choose between 3 types of solutions. The simplest (and most common) one is to use an amplified subwoofer and a 7.1 integrated receiver. However, this does not deliver the best…
In the vast manjority of the AV receiers, the 7 channels (usually identical) are supplied by one single power supply, to minimize manufacturing costs.  If some of these receivers can offer 7 x 200w, most are limited to 7 x 175 W and oddly enough cost twice the price of entry level devices rated 7 x 155 W.
Ain’t that strange?
Looking more closely, you will find that this power rating is for a non-standard 6 Ohms load. Even more interesting…
Then, you will discover that the rated 155 W per channel is only possible when only 2 channels are in use, but that all channels driven simultaneously, you will only have 7 x 112.63 Watts!
This of course is written in extremely small fonts.

This image is not displaying the virtual AV receiver I am describing above.
 It only shown that there is a single power supply for all channels

There is an easy explanation to this: The power supply is designed for providing a cost-optimized juice for the 2 so-called “main” channels (this is irritating: there is only one main channel in Home Theater, the centre one). When more channels need power simultaneously, the power supply just gives up.
Pursuing our investigations, we will discover that this magic do-it-all device affords a distortion level which is less than 0.5% and an s/n ratio > 90 dBA.
Not bad for a 60 years old single-ended valve amplifier!
Well, it is not, actually.
I am not trying to discourage the thorough and smart attempts of the marketing departments to dissimulate the fact that the receiver is an entry-level, low cost and low performance product by presenting it as a high-end one. The marketing manager job is difficult these days, and I really appreciate their creditable efforts.

In the pro-audio industry the use is to nickname an amplifier “black box with gain”. This really gives the right idea of what it is. Or should be.
(I call a single channel an amplifier, for the sake of simplicity)
Its function is to use the electrical energy from the mains to transform an input signal of X Volts in an output signal of X * G Volts, G being the gain.
Professional amplifiers have a normalized value for G, either 26 dB or 32 dB (G=20 and G=40, respectively, as 20 Log 20 = 26 and 20 Log 40 = 32)
This is an ideal world situation.
Please note that X*G is a voltage, not a power.
In the real world, the output signal is: (X Volts * G = X.G + e1 + e2 + e3 +….. ex)
where e1, e2, e3…are harmonic distortion, non-harmonic distortion, noise level, etc.

Here comes the maximum output power…
The X*G product is limited to a definite value which is inherent to the amplifier. When X increases so that the output is becoming near to X.G max, G is reduced and e1 is seriously increased, especially in odd harmonics (the most unpleasant to the ear).
So it seems straightforward to select an amplifier: Maximize X. G max, minimize e1 + e2 + e3.
The larger X. G max, the less you are likely to get near to this limit, which is called clipping level.

In some literature it is stated that some valve amplifiers can sound “louder” than what can be expected from their maximum output. This is supposed to be because their clipping is more “musical” than the one of other amplifiers, so you can drive the amplifier harder. Personally, I am very suspicious about the “musicality” of clipping.
This does not mean that amplifiers all sound about the same when they are below clipping level. My point is: I strongly believe that an amplifier that does not clip sounds better than one clipping.
Of course it suggests selecting large power (voltage) amplifiers, which is not the cheapest solution.
Now, let’s look at the power figures, which depend on the load impedance. The impedance of an actual loudspeaker is not a constant, like a resistance, but a function that varies with frequency.

The power figure is given by the Ohm formula: P = U2/R

Therefore, the power P is inversely proportional to the impedance (here expressed as the resistance R). So, if you divide R by 2, you double the power.
The problem here is that X*G max is, in theory, constant for a given amplifier. In reality, it is only constant within a certain impedance range.
Having a look at the figures in the literature, we will see for instance and amplifier capable of delivering 200 W in an 8 Ohms load can only provide, say,  320 W in a 4 Ohms load. Applying the Ohm law, we will find that the X*G max value is 40V into 8 Ohms, and only 36 V into 4 Ohms. This shows that 4 Ohms is outside of the linear range of X*G max.
Bad news!
Looking at various amplifiers figures, we will see that for, say, 200 W into 8 Ohms, some amplifiers will deliver 400 W into 4 Ohms, whilst others will only deliver 280 W, for instance.
There are even amplifiers delivering less power into 4 Ohms than into 8 Ohms. This sometimes explains why the power is expressed into a 6 Ohms load: It is the load into which the power is maximum, this being used as a (poor) selling argument.
In some instances, this can be due to the circuit design, but in most cases it is a limitation from the power supply.
The technology of a good power supply is straightforward and well known, so it is not a design issue.
The real issue is cost: In most power amplifiers, the power supply is the main part of the cost (and weight in traditional analogue designs).
So, the trade-off in limiting the cost is limiting the ability of the amplifier to drive low impedance loads.
And the more linear is the power vs.impedance relation, the better the amplifier.

Well, this does not explain it all. There is more...

Nearly everyone knows, amplifiers have different sound qualities. We are in the real world, actually very far from the ideal “black box with gain”.
We can look in detail at e1 + e2 + e3, etc; beyond a certain level of “decent quality” it will not give us any significant information. If distortion is below 0.01% (-80 dB below signal level) it will not be perceived as it is masked by the sound of the signal itself. Now the noise floor (which does not depend on the signal) will not be audible if the s/n ratio is higher than 110 dB (even with high-efficiency horns). By the way, the s/n ratio is a very interesting indicator of the built quality of the amplifier.
Stranger even is the significant variation of tonal balance between one amplifier and another, which is never suggested by any frequency response measurement (it is always ruler-flat).

About this, I would like to mention a strange experience I have made.

I was working at the R&D of professional sound reinforcement loudspeakers. Some products in our range still had passive crossovers (cost, again!), and we were setting up a QC measurement of the crossovers.
We used to do it a rather high levels (circa 50 Volts), as they were heavy-duty loudspeakers.
We measured the frequency response of the voltage across the crossover load. Also this is not meaningful, what we were looking for is the identical reproduction from one crossover to the other. It was only QC.
Once we found results that significantly differed from the standard. Strangely enough, all crossovers in the batch had an identical response, but significantly differing from the previous batch. The implementation was correct, and the components were not different. The only different element was the power amplifier that had been replaced by another one.
So we checked the frequency response of the power amplifier itself, driving the complex load presented by the crossovers under test.
We had the surprise to discover that it was very far from ruler-flat, so we thought it was faulty.
Then we re-installed the former amplifier. It had a quite different response, but still far from flat!
Then, we reduced the voltage by steps, and discovered that the curves became nearer and nearer to the ideal straight line when the voltage was reduced. About 10 dB below clipping, the response of both amplifiers was the same straight line.
The most concerning part was the serious unbalance of the response at high levels, with deviations of up to 6 dB for the “weakest” of the 2 amps.
I found reassuring that the “weakest” amp was the one which was not so good sounding…
It is unfortunate that I did not have the time or later the opportunity to further investigate this experiment, but I did think about it and found a possible explanation, although it is only a not-verified hypothesis.

Its is therefore to be taken with great caution

Looking back at the available power from, for instance, amplifier A:
  • 200 W into 8 Ohms
  • 300 W into 4Ohms
Whereas amplifier B delivers:
  • 200 W into 8 Ohms
  • 400 W into 4 Ohms

It is easy to understand that at frequencies where the loudspeaker impedance is at its minimum, 4 Ohms for instance, amplifier A will have its output voltage limited at a lower value than at frequencies where the impedance is higher. This of course only occurs at high levels when the voltage is near to its maximum.
Still, the result is affecting the frequency response of the amplifier, as the voltage will be higher at some frequencies than at some others.

In the same conditions, amplifier B will not meet this issue, as it is capable to deliver the same voltage into 4 Ohms and into 8 Ohms. Its frequency response will remain unchanged.
The serious variation in amplifier frequency response I described above suggests that the limitation phenomenon starts occurring at levels that are far below the clipping level: There would be a progressive modification from the straight ideal response to a load-limited clipping response as the level increases.
It looks like a “soft-knee” clipping, which would occur at different levels according to the load impedance and the power supply ability to handle it.

I insist that this is only a hypothesis, I did not investigate this phenomenon deeply enough to derive a definitive conclusion. However, this could explain why the tonal balance seems to significantly differ from one power amplifier to another one.

And there is no risk to deduct the following preferences when choosing an amplifier:

1)    When the budget allows it, it is better to use one channel per loudspeaker driver (multi-amplification) rather than passive wide-band loudspeaker systems: The impedance of individual drive units is typically easier to drive for an amplifier than the wide band load of a passive loudspeaker with a reactive crossover.

2)    When the budget allows it, it is better to use very powerful amplifiers. They are more likely to work in their linear range anyway. Further, their (unpleasant) clipping threshold is rejected at higher levels.

3)    Choose amplifiers provided with a properly dimensioned power supply and a significant headroom, so that the maximum voltage does not vary too much with respect to the load impedance.
4)    Select preferably amplifiers with a high s/n ratio (>105 dB)

Well, this has been a bit long, and I just hope you did not fell asleep.
To be followed soon…

No comments:

Post a Comment