Choice, application installation and set- up

Choice, application installation and set-up

Fortunately, the dedicated users and the more intelligent reviewers of audio power

amplifiers have become good at assessing quality, in all areas. So the frequent

comment by those who are astute about quality in everyday life, “They don’t make

stuff like this any longer” when surveying surviving 40 year old artifacts against

many modern goods, may be untrue when it comes to power amps. The reverse may

actually be true. The development of domestic high performance audio has

undoubtedly benefited from the continuing, public scrutiny of product quality by

experienced reviewers, in magazines.

One of the most important discoveries of the second part of the 20th century is that

while quality is hard to define, humans have intrinsic quality sensors that precede

intellect.

“Reality ... is the moment of vision before the intellectualisation takes place. There is

no other reality.”

And “..what is good... and what is not good, need we ask anyone to tell us these

things ?”

Robert Pirsig, in Zen and the Art of Motorcycle Maintenance [2]

Listening to equipment is an example of this skill being honed. But in any

manufactured item, quality is more than this, and the skill of ‘picking a good one’

and ‘by instinct’, is there for the using.



9.1.1 Choosing the right power amp, domestic

Many domestic power amplifiers are of the ‘integrated’ (pre+power) type, but a minority

of these may be described as high performance. A ‘separate’, ie. a discrete power amp,

as used by dedicated listeners and professionals, reduces the number of opposing

performance and feature compromises, and increases the scope for future changes.



How much power ?

Be aware that even if the cited watts are ‘honest’, the relation between wattage figures

and loudness is counter-intuitive. Most notably, it is under-powered amplifiers that

are more likely to burn-out HF drive units. Specifying the ‘right wattage’, a power

rating that will satisfy the listener and also suit the speakers without risking ‘blowing

them up’, is complicated by the uncertainty that ‘apples are reallybeing compared with

apples’.



Programme ratings

For example, most low-budget hi-fi speakers are already rated (in watts) for

‘programme’ (music), so the power amplifier’s, average watts per channel rating

shouldn’t range much above this. A power capability up to twice the speaker’s ratings

is likely to be harmless if programme material always has reasonably wide dynamics

and no sustained high levels above 5kHz, i.e. avoid playing highly compressed sounds,

and raw percussion and synths, at high levels. Then a 200 to 300w/ch amplifier

would be the most for speakers rated at 100 watts to 150 watts programme.



332



9.1 Manufactured goods

With more honestly rated, possibly more expensive loudspeakers, the rating given

may be a continuous type, based on the AES/EIA test. If so, programme ratings will

be x3 to x10 greater. So a 60w AES rated enclosure will handle programme up to

from 180 to 600 watts, provided it is not compressed, and that the amplifier is not

driven into clip other than occasionally and momentarily.



Power rating truth

The amplifier wattage needed for a given loudness depends a little on how near you

are to the speakers, and a lot more (often) on how sensitive the speakers are. The

loudness difference between the least and most sensitive Hi-Fi Speakers is up to at

least 8 fold for a given drive level. This represents a 1000 fold difference in power

conversion efficiency, alias a 30dB difference in SPL. Just changing to a 3dB more

sensitive speaker (e.g. from 90dBSPL @1w @1m to 93dBSPL @1w @1m) is equal to

doubling the amplifier power rating.



Scaling

Increased watts give only inverse or anti-logarithmic (read: slack) increases in sound

level, yet cost linear money. A 500 watt amp and speaker will both cost getting on

for ten times more than 50 watts. But a 10 times more efficient speaker need not cost

anything like 10 times more.



Power satisfaction

The easiest and most certain way to know what power rating is required is to forget

about precise power ratings and listen with the prospective ‘bride’ speakers that the

amplifiers under selection may be mated with.

The maximum useful loudness can be quite different to expectations, and may depend

on the music being played, according to whether the amplifier has output protection,

and whether this is being triggered by the speaker’s dynamic load.

Active systems (bi or tri-amped) are well known for producing clean sound levels

higher than the simple aggregation of the four or six (or more) amplifiers channels’

power ratings would suggest.



High level defects

Another cause of inadequate or unsatisfactory loudness (e.g. where 100w of power

amplifier Y sounds quieter than 30w of amplifier Z) occurs when the sonic quality

degrades half a dB or more below clipping. A well-designed power amplifier using

NFB maintains its sonic character up to just below clip, say within 0.2dB. But above

clip, the distortion sets in extremely abruptly and if prolonged, soon reaches

unbearably high levels. Some listeners prefer to accept ‘gracefully decaying’ sonic

performance in the 10 to 3dB leading up to overload. Called ‘soft clip’, this is second

best to adequate headroom. Yet even the high-end industry and the informed public

have a historic and continuing avoidance of high headroom amplification, in part

through fear of the high power numbers (One Thousand watts and above) that adequate



333



Choice, application installation and set-up

headroom confers, and partly through a belief that such high powers (which actually

have only up to an order higher voltage and current swing than conventional designs)

must necessarily degrade sonic quality.



Blind specification

If buying without listening (or being able to listen) to how loud the amp will go with

the speakers it will be partnered with, you will be reliant on a power figure. However

notional this is, at least check what impedance a prospective amplifier’s power is

developed into. The issue is not that higher powers are always developed into lower

impedances than your speaker has, but that the advertised or brochure’s ‘headline’

power rating, is the one that is developed into the nominal impedance of your speakers.

In other words, a specified ‘100 watts’ won’t be available into your 16 ohm speakers

if the 100 watts is only developed into 2Ω. Only about 13 watts will be available.

Misreading − or not reading − the maker’s (or a good technical reviewers’)

specification, then not listening, could leave you paying dearly for hypothetical power

and headroom.



Quality

Somewhere along the line, decor and aesthetics will ultimately matter to someone,

and likely more so with physically bigger units, particularly if it (or they) will have

to be positioned dominantly as a ‘room sculpture’. Nothing short of a physical

inspection and a relaxed period of hands-on use can allow quality to be perceived in

an apposite perspective − and all the more so for astute and skilled purchasers.

Reliability must somehow be gauged. Parts inside most power amplifiers will be

subjected to more stress than any other item of Hi-Fi equipment. Most domestic

power amplifiers are still heavy, so the cost in carriage can be substantial if they

have to be sent back for repair. But then an amplifier in a sturdy case will more likely

be built electrically sturdily, so the likelihood of the need is less!



Rationalised controls

Domestic power amplifiers are usually even more bereft of unnecessary or tangential

features than utilitarian pro-PA amplifiers, features that regularly appear on massmarket, Mid-Fi domestic and low-budget DJ-grade power amplifiers, most notably

gain controls, ‘swing needle’ (or LED) meters indicating ‘power’ output , and any

other kind of superfluous LED indication. The money saved can potentially be used

to increase real quality. However this should not be taken too far.



Recommended features

A clip indicator LED is invaluable (with conventional solid-state amplifiers having

high global NFB) for knowing for certain when you are exceeding limits that may

quickly burn out part(s) of your speaker system. Even if you have enough headroom

so you never see it light in the loudest passages, how could you be sure without it?

The clipping of amplifiers employing valves and/or low NFB is less abrupt, far softer

and far less likely to damage drive-units.



334



9.1 Manufactured goods

A ‘we have a problem’ ‘Error’ or ‘Protect’ LED is most useful in the eventually

probable event that something is (or goes) wrong, at least to save you wasting time

and energy trying to work out why there is no sound.

A second set of parallel output terminals or sockets, for bi-wiring.

Headphone sockets are rarely seen on any modern power amplifier, excepting

integrated types. For most users, they are superfluous. Fortunately, if required, usually

for when you want to use medium-to-high impedance (rated at 100 ohms or greater)

headphones, it is straightforward to add (or have added) a headphone socket. Switching

of speakers is best avoided. An advantage of using the professional Speakon speaker

connectors in that speakers can be disabled (and the plug contacts cleaned) without

removal, by simply twisting. Low impedance (50 ohms or lower) headphones should

never be plugged into an amplifier capable of more than a a few tens of watts, as

both the headphones and the wearer’s ears stand a chance of being traumatised if

anything like full level is passed more than momentarily.



9.1.2 Choosing the right power amp, for pro users

In touring PA, reliability, low weight, a certain fashionable ‘u’ size for a certain

power, and ability to ‘kick ass’, have often taken precedence over general sonic

quality in any selection listing.

In recording studios, there is more chance of the sonic results being the priority, yet

the amp may be ‘chosen’ by playing safe and buying on a name or untested

recommendation alone; or the monitor speakers’ maker or installer may specify what’s

to be used. The latter may be the best route, but the suitability of such a specification

should be demonstrable by listening and comparison.



Power

The average power rating is commonly selected to be higher than the AES/EIA or

‘rms’ or ‘average continuous’ power ratings cited by most professional drive-unit

makers. Provided clipping does not occur, or only happens occasionally and very

briefly, both the speaker and in practice, the amplifier power ratings may safely be

three to ten times this. For the speaker, this rating over the continuous is well known

as ‘Programme Power’. This procedure naturally allows a given amount of hardware

to be used to its fullest. This is important not just for operating costs, but also for

sonic quality, since it allows the number of sound sources to be minimised. Unless

there is severe compression (e.g. as might be caused by over-processing, or by clipping

back at the mixer), the amplifier’s unclipped maximum power delivery will be largely

in reserve.



Bridging

With a cursory inspection, bridging sounds like a thoroughly good thing. It is after

all balanced drive. But it’s commonly overlooked that bridging halves any given

output damping, doubles the amount of circuitry, internal connectors and solder



335



Choice, application installation and set-up

joints in the signal path, and perhaps most important, almost wholly cancels all

even-order harmonics created in the amplifier. As no amplifier has solely even harmonics, odd harmonics are left, and most are highly unpleasant and fatiguing [3].



Bridge frustrations

Usually, rated power is barely doubled. Amplifiers capable of about 1kW into 8

ohms when bridged are relatively commonplace and ‘old hat’. Bridging makes a

nonsense of precious power-to-size ratios: 2 channels of 1kW into 4 ohms become

1 channel of 1.2kW into 8 ohms, without any halving in size to show for it ! In its

defence, bridging is fine if it does the job, but the potential cost in size, weight and

sonics must be evaluated carefully.

On the other hand, bridging ideally reduces supply rail noise intrusion, and can be

used to build more reliable amplifiers, where one half of the bridge continues with

only a modest volume reduction, if the other side is sick or being repaired. Overall,

bridging two amplifier channels remains a mixture of good and bad in engineering

and sonics, a frustration to the designer and the user alike.



High power

Bridged or not, amplifiers that produce above 1kW/ch into 8 ohms/ch or above

2kW/bridged are increasingly available. In PA, appropriate use of these multi-kW

per-channel amplifiers with the new class of drivers and suitable enclosures will

reduce the number of sources, help simplify acoustic calculations and the related

predictive accuracy, improve sonic quality, reduce rigging time, reduce costs and

potentially increases system reliability, as there is simply less kit to go wrong.



Sonic chemistry

Some amplifiers sound good whatever they’re used with. With others, reliability and

sonics may be considerably influenced by the boxes they are driving. An amplifier

that sounds poor or shuts-down driving one type of cabinet, may shine with a different

kind − even one working over the same frequency band.

Amplifiers that measure badly can sound good to some users, but inaccurate to others.

Pleasant amplifier distortion can mask unpleasant distortion elsewhere in the system,

but if the whole system omits unpleasant distortion, the masking is unmasked. This

means if you choose a given amplifier on the reasonable basis that it sounds best in

your PA or studio, you may find that a completely different amplifier is preferred

when or should you change the mixer or other major components.



Listening test caveats

How well a given amplifier sounds on the day you check it out may depend on

whether it was warmed up, the quality of the mains supply, and the general health of

the system. For example, if unbeknown to you, one channel of the mixer source

happened to be marginally stable and putting RF at 1MHz onto the signal, then Amp



336



9.1 Manufactured goods

‘B’ with better RF filtering might win over Amp ‘A’, which has lesser filtering, and

would otherwise sound better.



Reliability tips

While reliability can be estimated by seeking out the experiences of other owners,

inspection of the construction, and asking the maker specific questions, there is no

way to be absolutely certain. Production of any previously reliable amplifier model

can prove unreliable in use at any point in the future if key factory staff are lost, or

change, or if the design is ‘re-engineered’ by accountants. And even a well-engineered,

generally rock-solid design can fail if a particular user and their equipment accidentally

uncovers out some weak spot.

Equally some users are happy with, and get reliable use from, designs that can be

readily blown up when bench tested. An amplifier that blows up when driven

continuously into a 2Ω load is a non-problem if the amplifier is only driven with

music and driven into 8Ω speakers that do not exhibit significant impedance dipping.



Transistor reliability

With good protection, and otherwise solid design, it is the junction temperature and

temperature cycling ‘world line’ of the power transistors, that have the foremost

effect on audio power amplifiers’ reliability. Like aeroplanes, power devices have a

finite number of cycles before metal fatigues, while every 10°c rise in temperature

approximately halves a semiconductor’s lifespan, through chemical diffusion.

Considering amplifier cooling, the large (and especially full depth) cooling tunnels

seen in many high power amps are less potent than they seem. Frequently, when

working hard, the temperature of the hottest device at the outlet end of the tunnel,

can be as much as 20°C hotter than the one by the air inlet (Figure 9.1, overleaf).

And this is just the surface temperature. Internal, junction temperature differences

can be greater, especially if the transistor fixings differ in tightness. Such temperature

differentials confuse power sharing and will lead to unreliability as the hotter devices

blow prematurely.

An amplifier suffering significant temperature differentials between the output devices

(whether this is caused by careless cooling design; or by poor mounting of devices)

should be derated for reliability, to well below the potential power capability. Because

the maker can only derate so far while meeting a salable price tag, some trade-off in

reliability and sonics is inevitable.

A number of modern designs (e.g. specific models by ARX, BGW, C-Audio, Chevin,

Crown and MC2 Audio) employ a short but wide lateral heat-exchangers cooled by

a plane airfront. With all the power devices exposed equally, they operate at near

identical junction temperatures, sharing out the thermal stress equally (Figure 9.2).

This approach allows every device to be safely used to its full rating, and fewer

devices (and less stringent protection) may be needed as a result, for a given level of



337



Choice, application installation and set-up

Figure 9.1

The power devices on blown, tunnel heatsinks rarely

operate at the same temperature. This degrades

reliability and sonic quality and wastes precious

silicon resource.

© Studio Sound



Figure 9.2

On lateral heatsinks, power device temperatures can be almost perfectly equal, so

thermal stress is experienced equally.

© Studio Sound



power handling and reliability. Suitably designed short cooling tunnels can also offer

isothermal operation.

While one cannot really estimate the reliability of an amplifier by counting the number

of devices alone, a broadside-cooled array of power transistors is a good point you

can tell about an amplifier, just by taking the lid off.

Users can contribute greatly to reliability by shielding amplifiers from stressful

occurrences, e.g. see section 9.7.



Mains universality

In territories where the AC mains differs from that in the maker’s territory, users (or

importers) should carefully evaluate and monitor the maker’s success at the

conversion. For example, the transformers in USA made equipment are designed for

a nominal 60Hz and 110 to 115v AC. In the UK, the supply is not double as is often



338



9.1 Manufactured goods

presumed, but 240 volts and often up to 256 volts (irrespective of theoretical EU

normalization). Also, the frequency is 50Hz. Alone or together, the lower frequency

and higher voltage will cause a transformer designed down to the last cent for 60Hz,

and down to an inadequate primary voltage, to run hot, buzz loudly, and possibly fail

at an early stage. Sonic quality may also be affected by line frequency, at bass

frequencies in particular, as power supply energy replenishment is naturally less

frequent at 50Hz (50 cycles per second), vs. 60Kz.

Users operating internationally should also check that the amplifier’s power supply

is suited to use on other mains supply voltages and frequencies, and conservatively

rated, not forgetting that mains voltages outside of the UK and Western Europe are

not so standardised and/or not nearly so tightly adhered to both in the ‘3rd world’

and in the rural outback of spacious countries like the USA, Canada and Australia.



Listening

Relaxed listening is a good start (forget A-B tests unless you have an strong

‘photographic’ short-term memory of music). Make the repertoire extensive; listen

over a few days, with lots of different source equipment, different speaker boxes and

different people. If you care about checking in on finer differences, make a point of

listening in stereo. This applies even for PA, and even if stereo content in concert PA

has mostly so far been minimal. If you’re looking for a bass amplifier (say), and one

of the shortlisted amps you are trying is obviously not giving the performance you

were after, it might be worth briefly trying it in other roles. This is the way great

discoveries are made ! Equally, if you experience disappointing results, it’s valuable

to contact the dealer or maker. Getting them to supply another sample unit and making

doubly sure that you’re using it optimally may make no difference, but equally, the

tables have been turned before on such occasions.



Mechanics

A mechanical layout that spreads the load can make a big difference to the comfort,

safety and longevity of a touring power amplifier.

Examples of bad mechanical design include those amplifiers with:

(i) Much or most of the weight focused in any one area away from the centre

other than towards the front and symmetrically about the centre-line.

(ii) More than about 6"/150mm deep without rear rack supports.

(iii) Sharp (un-radiused) metal edges, liable to cut or injure installers.

Good designs may include such features as:

(i) Recessing of vulnerable or critical controls or switches.

(ii) A single secure but easily removable cover giving access to all the breakable

or wear-prone parts most requiring attention.

(iii) All air-spaced PCBs, allowing rapid drying out in the event of flooding.



339



Choice, application installation and set-up

Individual reality

Overall, power amplifiers are like other machinery that humans experience intimately:

their ‘world-line’ interaction with your system in your part of the universe is not

predictable from any spec sheet. Even with in-depth technical understanding, no

amount of technical data and bulleted, internal technical features can tell you anything

definite or specific about the unmeasurable qualities. Therefore you may find it useful

to do the listening tests before reading any spec sheets and technical papers, or

dissecting or abuse-proving the unit under evaluation. If the amplifier sounds bad,

you never have to wade that far.



9.2 Howlers

Here are some common and ‘it really happened’ defects to look for in manufactured

units. Based on hundreds of reviews of both domestic and professional power

amplifiers costing from £150 up to £5000, or even more.



Illuminations

*

*

*



‘Clip’, ‘Peak’, ‘Error’ or 0dB bargraph LEDs that warn of overload more than

½dB early or far worse, more than ¼dB late.

‘Clip’, ‘Peak’, ‘Error’ or 0dB bargraph LEDs, where the accuracy of clip

indication depends on the speaker loading and mains voltage.

Important panel LEDs that are invisible unless viewer’s eyes are almost on-axis.



Noises

*

*

*



Transformers that buzz audibly with normal mains harmonic variations.

Overly noisy fans for domestic and some PA applications.

Hum at output with no input connection.



Anti-social behaviour

*

*

*

*

*

*

*

*



340



High inrush current at switch-on, which trips house power breakers, making

use almost impossible.

Farting noises and sweeping whistles from speakers, when amplifier is switched off.

HF signal continues to pass when output is muted.

Binding posts deeply recessed between heat-exchanger fins.

Binding posts that are too slippery to be tightened, and/or undo themselves

despite attempted hard tightening.

Sharp, un-radiused metal case corners.

Protection by a specially fast fuse with no spare(s) supplied. Fuse fails at

worst moment through old age.

Manual intervention required to reset an amp – after just a momentary power

down, or the briefest short at one output.



9.2 Howlers

*

*

*

*

*

*



Under-PCB insulating material that traps moisture despite hours of drying out

in warm air flow. Important for outdoor use and even for parties.

Thermal grease (especially anything silicone based) oozing around mounted

parts, or smeared about.

Unfiltered fans blowing into electronic areas; and un-louvred or shielded air

vents, allowing dust and filth to accumulate on PCBs.

Fans switch off when the amplifier gets too hot.

A fault on one channel stops the other channel or the whole amplifier working,

without specific indication.

Confusing, complicated or illogical LEDs (‘a green LED, marked ‘OIC’

flashes dimmer when headroom is exceeded’) or unindicated events (‘The

power LED will continue to glow even though the amplifier has shut down’).



Poor Q-A

*

*

*

*

*



Supplied ‘dead on arrival’.

Something metal rattles inside the case.

Uneven torquing-up of power device screws.

Un-alignable lid screw holes.

Random input polarity i.e. hot sometimes (XLR) pin 2, other times pin 3.



‘Unroadworthy’ construction

*

*

*

*

*

*

*

*

*

*

*



Unlocked screws.

Unsealed pots in an unsealed enclosure.

Unpolarised connectors.

Unretained, non-latching connectors.

Unmarked connectors adjacent to others having identical pinout.

Power devices, or audio or mains connectors or other wearable parts, hidden

under an hour of metalwork dis-assembly.

Sub-standard 4mm binding posts with short and/or slack receptacles

Sub-standard 4mm binding posts with a slippery, ungrippable body.

Flimsily mounted transformers that wrench free in transit, smashing other

innards.

Amplifier innards explode when outputs momentarily shorted.

Amplifier emits smoke or fails when long speaker cable connected.



A small, worldwide amplifier interface quality catastrophe

A type of 4mm output binding post made in Japan and regrettably fitted to thousands

of ‘competitive’ professional and perhaps millions of ‘mid-priced’ Hi-Fi amplifiers,

mainly made in Japan and the USA, is diametrically at variance with real needs. To

be of any use, the hexagon shaped posts, moulded from a slippery plastic that is



341



Choice, application installation and set-up

certainly ungraspable if the grasper is sweating from any exertion, require an end-on

hex driving tool, that is never seen or supplied. Second, the 4mm plug receptacle is

foreshortened and has slack tolerances, so even the best sprung plugs are not gripped

uniformly tightly, if at all. Third, the offending type is drilled to accept only thin

wires, usually not above 0.75mm2. Astute purchasers will insist on having these illconceived posts sent for non-audio metal recycling, and replaced by higher quality

parts, which are hardly expensive. Cliff and HCK are the foremost makers.

In the following sections, first use, installation and sonic optimisation are methodically

considered.

Figure 9.3

BGW’s PS2, a compact, solidly-packaged

yet budget professional amplifier,

showing (centre foreground) four of the

sub-standard output ‘binding posts’

blindly employed by pro and domestic

makers in the US and Japan. Note input

jack sockets on the right, which should

never be used for fixed installation.

The remainder of this design gains high

marks for ease of servicing and

inspection. Most of the air exits at the

sides – a type ‘C’ airflow. © BGW



9.3 AC mains voltage

New amplifiers should be factory set to the nominal mains (line) voltage for your

country, and this voltage should be stated on the rear panel.

Across the world, there are many combinations of voltages and frequencies. Five

AC mains voltage ranges cover most Western and many Pacific rim countries:

1. UK, Australia, New Zealand, Malaysia:

240v

Note 1: Since 1.1.95 UK is technically 230v +10%, −6% by EU law but in practice

240v +/−6% is maintained. In the year 2001, AC voltage throughout the EU will be in

theory standardized at 230v +/−10% but the UK power authorities will nonetheless

maintain 240v + 6, −10%. Likewise, continental European 220v will not increase to

230v in the short run, in many territories.

Note 2: Western Australia ranges 240 to 260v.



2. Western Europe (continental)

3. Singapore, Korea

4. USA, Canada, Mexico, Taiwan

Note 3: With regional and local variations from 110 to 120v.

5. Japan

Note 4: With regional and local variations from 95v up to 110v.



342



220v or 230v

220v

115v

100v