Eagle, Music Technology Ron Lebar
Feed Back
Feed Forward
Class A Vs B
Valve V Transistor

Preface, a brief definition of amplifiers & amplification.

A device for increasing the amount or amplitude of a parameter is frequently needed. Many work by increasing one value at the expense of another. A simple lever can increase mechanical force, but at the expense of travel, so there is no net gain. An electrical transformer can increase Voltage with a reduced current. Again no net increase in output power.

These are termed Passive devices.

An Amplifier is an Active device, a source of power is required apart from the input. Energy is drawn from this to provide the amplified output. A mechanical example is a brake servo, this uses power from a vehicle engine to increase force applied to brakes from a small pedal pressure.

There are many other mechanical examples, however this section is concerned only with the analogue electronic variety, for audio frequency use.

The general form consists of one or more Voltage amplifying stages followed by a power stage. Each stage uses a DC power source, with active devices to control current flow into an output load. The two main types of active device used are Thermionic Valves (vacuum tubes) and transistors.

Magnetic devices, such as saturable reactors, can be used in some applications. In general these have poor linearity for analogue audio & insufficient speed for digital. Their use will be covered in a separate section.

Magnetic technology should not, however, be dismissed. Advances in design may yet make the form viable. We are starting work on ideas for a stage amplifier. How this works out remains to be seen.

Digital amplification may use other devices, with switching capabilities, such as Thyristors or SCRs. This type can be highly efficient & handle high powers. From the early years of transistor amplification various designs have been tried, but use in audio has been limited. The genre will be discussed in a later section.

Some amps are falsely called digital. In reality they may have digital displays, effects, or be supposedly intended for digital sources. Our language is often abused in the name of commerce.

Non-linearity inherent in active devices distorts the signal being amplified. A long time ago it was realised that feeding back a proportion of the output signal to the input, in reversed phase, reduces that distortion. The Voltage gain is reduced by the same factor.

This is called Negative Feed Back (N.F.B.) & was pioneered commercially by Harold Leak of Leak amplifier fame. For the first time high quality audio reproduction became economically practicable.

Feed Back
Feed Forward


Types, causes & effects

When an electrical signal is passed through any device the relationship between varying input levels & resultant output can be plotted on a graph. The result is often termed a transfer or characteristic curve. Ideally it should be a straight line (linear). The transfer function of an active device frequently departs from this ideal. The same is true of some passive components.

Iron cored inductors, such as output transformers, exhibit 'non-linearity', especially when driven hard. Capacitors, other than air spaced types, show a dielectric 'memory' effect. This varies with materials used, polyester has a strong effect. The resultant linearity errors cause distortion at frequencies where reactance is significant in relation to associated components.

The shape of a waveform is changed (distorted) by a non-linear characteristic. The result is measured in terms of the proportion of unwanted by-products created, expressed as a percentage or in decibels (dB). The dB is one tenth of a Bel, the common logarithm of a power ratio.

The common British measure of distortion is the percentage that all added harmonics form of the final signal (Total Harmonic Distortion or T.H.D.). One method uses the purest possible sine wave at a set frequency as an input. Filtering the output, to remove just that frequency, leaves distortion products, plus circuit noise, hum etc.

Measuring the output with no input gives the bachground noise figure. Subtracting this from the previous figure leaves just T.H.D. This is not very accurate, especially if noise is high compared to distortion. A more precise variant uses a narrow pass filter to measure each harmonic in turn.

Since changing the shape of a sine wave results in the introduction of harmonics this is a reasonable measure of a circuit's linearity errors. An apparent anomaly is introduced, in musical instruments the presence of harmonics makes sound brighter & more interesting. So why are harmonics produced during amplification so objectionable?

The short answer is that they are not. The full picture however is, as usual, more complex. Adding harmonics to an existing instrument can change its nature, making it less recognisable. There is a difference here between music production & reproduction. Players often make use of this effect, as when a solid bodied guitar's signal is deliberately distorted, adding timbral complexity.

Another type of distortion artifact occurs when two or more differing frequencies are amplified together. The curved response can be seen as varying gain at different points on a waveform. Each of the frequencies present will modulate the amplitude of any others as they are swept through the curve. This results in the production of 'sum & difference' frequencies.

The section on FM synthesis touches briefly on this subject.

This is called 'Inter-modulation' (I.M.) distortion & is the real problem with non-linearity. Its products are often not harmonically related to original tones so the ear picks them out as unwanted sounds, or 'dirt'. It gets worse when more & complex signals are amplified. Harmonic distortion compounds I.M. since sum & difference signals occur between all the additional frequencies created.

For this reason the American standard measure of quality is inter-modulation products (I.M.), again as a percentage of output signal. Two different sine wave frequencies are introduced, the measuring filter is set to the sum &/or difference. The two methods often produce vaguely similiar figures, depending on the type of non-linearity.

An old arbitrary standard for high quality was less than a tenth of one percent harmonic content. A twentieth of one percent I.M. is probably a near equivalent. We expect at least a factor of ten better today. Loudspeakers often have around five percent harmonic distortion at high levels. I.M. can be minimised by separating high, middle & low frequency ranges into suitable drivers.

The complexity & individuality of the human voice does not respond well to any kind of distortion. Amplification for speech or singing needs to provide the most faithful reproduction possible. The only addition sometimes desirable is reverberation, for adding life to an acoustically dull environment (or masking a dodgey singer).

Feed Back
Feed Forward

Amplification, Negative Feedback
1: Preliminary
2: Design Factors

An amplifier is proposed where very low levels of distortion & noise are primary design factors.

At first sight this seems straightforward enough, negative feedback reduces both, together with gain. For a simple sine wave the resultant distortion, below overload, is the original value divided by, one plus the product of original gain & the fraction of output signal fed back. Gain, noise & output impedance are reduced by the same ratio.

The feedback path, from output to input, completes a loop. The original gain, with feedback disconnected, is termed 'open loop' gain (OLA). The gain with the feedback path connected, is termed 'closed loop' gain (CLA). If OLA is very high in relation to CLA, the latter approximates to the reciprocal of the feedback fraction (FF). Distortion & noise are approximately divided by the product of OLA & FF.

A high open loop gain compared to closed loop gain appears to meet the design requirements. An additional benefit is predictability, gain, for example, is controlled, almost completely, by the ratio of feedback resistors. Negative feedback seems to be a designer's dream, a dramatic improvement in quality with no cost. Apart, that is, from an extra gain stage or two.

In reality however, there is more to it, nothing comes without a cost. Before moving on to factors affecting design, a simple comparison can be made.

On one hand an expensive 'high end' amplifier, on the other a cheap, unit construction 'Hi Fi' system. Both have very similiar specifications regarding noise, distortion & frequency response within the audio band. A tolerance of 3dB from 20 C/S to 20 KC/S is common for the latter, probably .5 dB for the expensive model. 2dB is just noticeable.

When tested, both models meet their specifications. Listening tests, however, paint a completely different picture. The expensive amplifier sounds pretty much as one would expect. No wild acrobatics of sound, just clear reproduction. The cheap unit sounds dull & boring. Treble is lifeless, something seems to be lacking.

Yet specifications are similiar, what can the ear hear that test equipment does not show? One essential parameter is left out of most specifications, 'phase linearity'. Equipment exists to measure this accurately, or a simple two channel oscilloscope setup can display phase errors reasonably well.

Amplification, Negative Feedback

2: Design Factors - 1: Preliminary
Feed Back
Feed Forward


Feed Forward, an Alternative Approach
The Beginning

At 15, on starting work, my early interest in music & electronics developed in practical ways. Quality music reproduction became something of an obsession. Experience at work was of limited help as the company business was radio & television. A complete manufacturing facility, components designed & made on site.

I knew of Negative Feedback but thought 'It merely reduces distortion, why not cancel it, together with hum & noise?' I was young & inexperienced. A preliminary circuit was devised, refined over a period & finally built. Amazingly it actually worked, the sound was quite a surprise.

Using vinyl records it gave a degree of transparency not heard previously. No distortion analyser was available, an oscilloscope showed visibly pure waveforms throughout the audio spectrum. More importantly, reproduction was pleasingly clear & realistic when judged by ear. Using four television type output valves, eight Watts was obtained, the design target was twelve.

Several were built for family & friends. Minor evolutionary changes took place, the current injected distortion correction was later replaced by Voltage drive. This required a change in transformer ratio, gave a more elegant looking circuit & achieved the power efficiency originally expected. With better valves, outputs up to 50 Watts were produced.

After that no more were built, solid state seemed the way forward.

Later a transistor stereo version was built, current summing was again employed. It was compared with a carefully designed feedback type using similiar components & transistors. Distortion analysis showed it to be slightly better & frequency response was as good. However it sounded much better, with the old transparency.

The feedback model gave more power & a greater damping factor.

With the passage of time & more experience the reason for its sonic clarity has been established. After a gap of many years, work is starting on a modern dual channel version, using current professional valves. With an output to suit today's requirements & Voltage drive.

A higher power, current injected, transistor version may also be built.

This will be a separate line of research to our proposed monitoring & stage amplifiers. However there may be some cross-pollination between them. Each will be subjected to long term test & analysis before any decision is made on marketing or otherwise.

Feed Back
Feed Forward

Feed Forward, an Alternative Approach
Elements of Design

The principle of feed forward error reduction is straightforward. Initially an audio signal is supplied to the input of an inverting amplifier, this is the Main channel. Its output is compared to the input, using a suitable circuit (comparator). This gives a difference signal, consisting of distortion products plus any hum & noise.

This signal is applied to the input of a second inverting amplifier, the Difference channel. Its output is connected to the same load as the Main channel, this tends to cancel any distortion products etc. Obviously, unlike the original design ideal, total cancellation can not be achieved.

Any distortion & noise in the Difference channel will not be cancelled In the original 'pure' version. In practice this will however be low, because this stage only needs to supply sufficient power to correct the small errors in the primary stage.

For the later version of the Voltage summed type, the comparator drive signal was taken from the combined output. As distinct from the primary stage output as in the earlier versions. This is the version shown in the schematic below. This arrangement causes any errors in the difference amplifier to appear in the comparator's output as an error & be amplified as a correcting signal, by the Difference channel. This mechanism is, of course, negative feedback and thus can only reduce such errors..

Despite this, the overall system is not a feedback amplifier. The Difference channel does not see the original signal at all, only distortion & noise products from the output. The Main channel does not require feedback for distortion reduction, it can be simpler, with lower gain. This usually means less phase shift & noise.

Phase shift is a problem in amplification, it is inherently worse with more stages & with higher gains. The human ear is exceptionally sensitive to phase errors, this explains, at least partly, a lack of realism from many well specified amplifiers. The feed forward system sees a phase shifted signal as an error & tends to correct it.

Feedback systems have difficulty with this, at some frequencies the shift may be enough to result in positive feedback. Often gain is reduced at these frequencies to prevent oscillation. This results in further phase shift, plus increased harmonic & inter-modulation distortion.

The section on negative feedback shows that correction by this method ceases to work at the point of overload. There is simply no power available to reduce the resultant high distortion. With feed forward, overload of the Main channel can be corrected, at least patially, provided the power required does not also overload the Difference channel.

Feed Forward

With correct design the maximum output should be the sum of the two channel's power capabilities. In the early models, with current summing, the result fell short by a third. Voltage summed versions reached the design target. This difference, also found in a transistor version, will be investigated. It is presumably due to each channel providing a low impedance load to the other, when driven hard.

Normally the two amplifiers are similiar. A transistor variant, tried early on, configured the Main channel as a lower power Class A amplifier. With the Difference channel providing higher power in Class B or AB. This should have combined the best of both worlds, improved inherent linearity at lower powers & greater efficiency at high output. Current summing worked better here.

Results were not promising at first, using germanium transistors, they were simply too slow to correct overload in time. When suitable faster silicon power devices became available in Britain the design worked well. By then I had moved on to other things and the idea was not pursued further. Now I am developing the idea again, to produce an amplifier with class A sonic clarity combined with high efficiency.

An important point that is often overlooked. Whatever form of error reduction is employed, the basic design needs to be good without it. Feedback or feed forward should improve on excellence, not try to create it. It is all too common to find N.F.B. used to bring a poor design 'up to spec', with uninspiring results.

Another common misunderstanding regarding N.F.B. is its effect on noise. Put simply it can not improve the signal to noise ratio of an amplifier or pre-amp. In many cases it makes the situation appreciably worse.

From the front end a specific degree of amplification is required. To raise a specified input signal to the specified output. N.F.B. appears to lower the noise, but it lowers the signal by the same amount (more of less). So additional gain is required to bring the signal back up, bringing back the noise.

The reason it can make matters worse is that noise is broadband. There are high frequency components. Due to phase shift, feedback does not exactly cancel these high frequencies. So the extra gain increases them. A similiar problem often shows with capacitor coupled amps at low frequencies (hum etc.). Using larger capacitors reduces the offending phase shifts.

This can create another problem, particularly with hard driven performance amplifiers. Overdriving causes stage inputs to rectify the signal, this happens with valve or transistor circuits. This charges the coupling capacitors, when the overload ceases the charge 'turns off' the stage momentarilly.

The resultant drop in volume is disconcerting, with a larger capacitor the time constant is longer. That is one reason why capacitor values are often chosen to only just do the job.

Feed forward will of course see noise etc. as an error & reduce it. As the stage gain does not need to be higher a genuine reduction is possible. The noise of the difference channel does of course interfere with this cosy situation. Nevertheless, with care, it is a distinct improvement, if a little more complex.

Feed Back
Feed Forward

Distortion in Passive Components.

The definition of a Passive Component is sketched out at the bottom of this panel.

It is commonly assumed that passive components, other than ferrous cored inductors, do not introduce distortion. Strictly speaking this is not true.

Capacitors especially are prone to non-linearity for a variety of reasons.

A simple capacitor is formed by two plates spaced a short distance apart. Capacitance is proportional to surface area and inversely proportional to spacing. Applying a Voltage difference causes a current to 'flow'. This continues until the capacitor is 'charged', i.e. Voltage across the plates is the same as the source.

Opposite potentials attract, the plates attempt to move together against their restraints. If they are able to do so, capacitance increases, requiring a greater charge to maintain the potential difference. Current again 'flows', with a practical source, of finite impedance, Voltage will fall. Due to the plates' physical mass, this fall is out of phase, causing distortion.

This movement of capacitor plates is the principle behind electrostatic speakers & headphones. The effect is reversible, creating the electrostatic or condenser microphone.

Even those 'in the know' consider air dielectric capacitors to be free from distortion. The validity of this assumption depends on design & accuracy of construction.

Most practical capacitors for audio frequency use have a thin insulating or 'dielectric' layer between the plates. This serves two main purposes.

(1) Plates can be closer, increasing capacitance for a given area. They can be of thinner material, allowing them to be wound into a coil for greater surface area in a given space. This further increases capacitance by using both sides of each plate. Alternatively a number of small plates can be stacked, for lower inductance.

(2) The dielectric material increases capacitance, by a pecentage called its dielectric constant (K) or relative permitivity. As with all things this increase comes at a price. Molecular movement said to be responsible absorbs energy, generating heat. This loss is greater with higher K materials. There is thus an additional resistive component, reducing the capacitor's efficiency.

More important from the distortion point of view is dielectric 'memory' or hysteresis. When stress from a charge in one direction is removed, the material does not immediately return fully to its original state. A small potential difference develops shortly after application & removal of a large Voltage. Various materials exhibit this property to differing degrees.

Polyester has a high K value, allowing high capacitances in a compact & economical format. It also has a large memory effect & substantial dielectric losses. The former causes distortion in some applications & also makes it a poor choice for sample & hold circuits.

Polypropylene is better in both respects, it is stable & robust. K value is lower, a given value is larger & more expensive. Polycarbonate materials have similiar properties & price.

Polystyrene has very low loss & memory effect. It is also very stable, ideal for high frequencies, tuning capacitors or sample & hold applications. Its low K value means practical capacitors are only in the lower value range. It is also fragile, vulnerable to heat & solvents.

An early standard material, oil impregnated paper seems to have advantages for capacitors working at audio frequencies. Lower memory effect & moderate dielectric loss. Not suited to high frequency operation, its lower K value requires larger size & it is currently expensive.

Dielectric constant or relative permitivity is the electrostatic equivalent of relative permeability in inductor cores. Dielectric memory is an aproximate analogue of magnetic hysteresis, both cause non-linearity & consequent distortion.

Wound or stacked plate capacitors may suffer from microphony, if the layers are not tightly secured. If a polarising Voltage is present at a low level stage, acoustic feedback may occur. This can cause distortion, even if gain is insufficient for oscillation.

Electrolytic capacitors are often used in audio frequency circuits. These are usually a variant on wound foil capacitors & have a much higher capacitance/volume ratio. The insulating layer is porous & impregnated with a conductive liquid or gel. This is the electrolyte.

The plates are usually aluminium foil. During manufacture a current is supplied, a process called forming. Electro-chemical action causes a non-conductive oxide layer to build up on one plate, the anode. Eventually this layer cuts off the current flow & the capacitor is ready for use.

Aluminium oxide has a very high dielectric constant, giving a high capacitance/surface area ratio. The plates are often etched to increase surface area. In addition the oxide layer is extremely thin, further increasing capacitance/volume ratio. Values are available that can not currently be achieved by any other technology.

Any slight leakage current further builds up oxide, until current flow is minimised. Capacitors are thus self healing up to a point. If applied Voltage is reversed any current flow reduces the oxide layer, causing capacitor failure. This polarity sensitivity must be taken into account. Non-polarised types have oxide formed on both plates, with lower maximum values.

Due to the winding, large values, as used in power supplies may have appreciable inductance. The supply can have a high impedance at high frequencies. High peak currents required by output devices cause Voltage drop, resulting in distortion. Ways exist to minimise this effect.

A newer 4 pole capacitor design reduces this problem considerably. It is inherently more expensive, most equipment makers failed to take advantage when first introduced. So high volume makers dropped production, making remaining specialist sources even more costly.

Electrolytics are useful for interstage coupling in transistor amplifiers. When compared with other types an additional source of possible distortion exists. Charging current between the plates must flow through the electrolyte, effectively a semiconductor. Current propogates via ions (charged molecules), rather than free electron flow. This can cause noise & non-linearity.

The use of open weave insulators, liquid rather than gel electrolytes and smooth foil can improve matters. Some designers solve the problem by DC coupling, eliminating coupling capacitors. This requires care to minimise offsets & is our prefered technique.

Defining passive components.

The dividing line between active and passive components is blurred. Various attempts have been made to define the boundary, none quite works.

One definition describes a passive component as one that does not exhibit non-linearity. This has been shown to be a non-starter. A better description is any component that does not use 'electronic' phenomena, including conduction in gases, a vacuum or 'semiconducting' materials.

This mainly means Capacitors, Coils (Inductors, Transformers) & Resistors. This is the division used here, borderline parts, such as Voltage Dependent Resistors (VDRs) etc. are lumped with active components for simplicity.

Active components may be defined as those able to vary current flow in response to a Voltage or current. Other than by mechanical movement. Certain inductive devices are able to do this, such as saturable reactors or magnetic amplifiers & may arguably be classed as active.


Feed Back
Feed Forward
Class A Vs B

Effects Unit.

The term 'Effects Unit', often abbreviated to 'FX', covers a wide range of devices. From simple distortion pedals through to complex digital sound processors. The one common purpose is modification of a music signal in some way. Various technologies are employed, most are electronic, some are electrical & a small number are electro-mechanical.

Purely mechanical devices have been used to change the nature of a musical instrument. The obvious example is a trumpet mute. In the context of this article only those that alter electrical signals are relevant. They will be listed by type of effect, followed by the various means of implementation. Not in any particular order, associated types may be grouped together.

( Phase Shifter. ) A long & boring treatise.

A phaser works by mixing two signals of equal amplitude. One is an original signal, the other is a version of that signal shifted in phase. In a practical electronic device the shift is created by a battery of series capacitor/ parallel resistor stages, commonly four, each buffered by an amplifier. Negative feedback maintains unity gain for each stage.

Below a certain frequency each stage begins to advance the phase of a signal. This shift increases as frequency is lowered, eventually approaching 90 degrees. Thus four stages can produce a shift approaching 360 degrees, an angle never quite reached.

At high frequencies where little shift occurs the shifted & original signals add. At low frequencies, where shift is near to 90 degrees per stage, the signals also add. At one frequency only, shift is 45 degrees per stage, 180 degrees overall. At that one frequency the signals cancel exactly, giving no output.

Cancellation (minima) occurs at only one exact frequency, but addition (maxima) happens over a wide range of frequencies, to a greater or lesser extent. A graph of the response shows two broad maxima separated by one very sharp, narrow minima.

If a large number of stages are employed a number of narrow minima, one for every 4 stages, separated by broad maxima, will be produced. With phase shift the minima are equally spaced & the response resembles a comb. It is thus often termed a 'comb filter'.

Using a large number of stages is not a practical proposition, for several reasons, one of which is cost. However there is a way of obtaining similiar results simply & at low cost. As the overall gain remains at unity the shifted output can be fed back to the array input with no danger of feedback oscillation.

This signal is shifted further on its next pass & then fed round the loop again. In a practical circuit the proportion of output fed back can be adjusted, fron zero to 100%. When set slightly below maximum the signal loops a number of times before being attenuated to inaudibility.

This creates the classic comb filter response. If the resistor values of each stage are changed together, the minima frequencies change. Moving the teeth of the 'comb' through the audio spectrum. If this movement is slow & cyclic it results in the sound usually associated with phasing. White noise as a source can sound like waves on a beach.

If the cyclic change is faster, around 7 C/S for example, with feedback reduced, the result is more like vibrato, with music signals. In this case the original signal can be disconnected, for a clearer effect. A similiar technique has been used for rotary speaker simulators, realism is not that good but it has served a purpose for years.

There are number of alternative ways to adjust several resistors together. A common one is to use the channel of a field effect transistor (FET) as each resistor. With all gates connected to a common variable bias. This bias can be from an oscillator, a potentiometer, or both.

A problem with most phasers is a fairly high noise level. From the ICs used as amplifiers & often from the FETs. Another reason to keep the the number of stages down. Spending more on better components helps here, as does using light dependent resistors.


A flanger works uses a similiar principle to a phaser. The main difference is the use of a time delay insted of a phase shift. The delay is mixed in equal proportions with the original signal as before.

This creates maxima & minima in a similiar way. These are however, spaced by an equal frequency percentage, rather than equal frequency amount. The sonic result is different..

A 'musical' ringing effect is produced, especially with feedback. Similiar to the effect of a pipe or closely spaced walls. This type of sound sometimes interferes with music, especially with certain types of chord structire. It should be used with care.

The delay is normally obtained by a digital delay line & is shorter than that required for reverberation or echo. Previously such techology was too expensive for general use oitside studios. Charge coupled delay or 'bucket brigade' circuitry filled a need for stage use.

The term 'Flanging' comes from the original way the effect was achieved, before electronic versions became available. Two identical tapes were loaded on a pair of precision tape decks. Both were set to a start point and played back in exact step.

Gently touching the feed spool flange on one recorder caused it to drop behind by a small amount. Another touch would increase the delay. Touching the flange on the other machine caused it to drop back into step.

Naturally such a procedure was only practicable in a studio, using expensive equipment. The advent of electronics opened up both phasing & flanging to a wider customer base. However both are now out of fashion, probably having been worked to death by overuse.

Spin Doctor

The first of a series of electro-mechanical signal modifiers has successfully survived a year long reliability test. This technology belongs to the future, with considerable evolutionary potential, & can give traditional digital electronics a run for its money in many applications.

This first gives Vibrato, Chorus, Tremulant & Phasing. In true stereo & unlike those from electronic alternatives. The absence of electronics within the device allows low distortion & noise. Potential uses include pre vibrato Hammond organs, electric pianos & guitars.

Spatial Vibrato is especially interesting, also working with Tremulant, Chorus & Phasing. A rotating sound pattern spreads across the stereo field with constant animation. It is complex, cyclic & effective at all speeds.

The nearest analogy is the Doppler shift & reflections of a rotary speaker. It is however unique & with good speakers delivers an even frequency response plus greater brilliance.

Its primary structure is our non-electronic analogue of an acoustic space, an Orbiter projects music signals around this space. These are collected at suitable nodes for amplification.
A twin orbiter version will be able to accelerate or brake at different rates for bass & treble.

Building of the production model will start when current restorations are complete.
Selecting here will show more information.


Information given is generally brief & is based on our experience. If you spot any factual mistakes or 'typos' please feel free to let us know. We are not perfect & won't sulk over constructive criticism.

All Brand & Model names are Trademarks and/or Copyright of their respective owners. The Eagle motif is the Copyright (1980) of Ron Lebar.

Regards & thanks for reaching our Site, an ongoing project.
Watch this space.

Courtesy, Excellence & Value.
The standard others are judged by.

E-Mail: info@alphaentek.com

Amplifiers. Updated on the 14th of May 2005. Ron Lebar, Author.