In many power amplifiers the op-amp circuit is constructed with discrete components specifically designed for higher rail Voltages. Output transistors are added to provide extra current to drive a speaker. Large output transistors only have a small HFE current gain, therefore driver transistors are placed in front of the output transistors to increase to total current gain to approx 200. Amplifiers that use power MOS-FETs do not require driver transistors. The bias string can now be placed in the Class A driver circuit.
Output transistors can be arranged in three different ways. This description is a basic overview. Electronic design detail including PCBs for constructing power amplifiers is available on sound-au.com
The first transistors were germanium which worked well for low power transistor radios in the 1960s and 70s. But germanium transistors were unstable and not reliable. Reliable Silicon transistors were invented later. High power amplifiers could only be built with silicon transistors.
The below pic shows parallel output transistors. Some large power amplifiers use many parallel output transistors. Large wire wound resistors 1/2Ω (R47) are placed in series with the emitters. These emitter resistors force the output transistors to equally share current and therefore will be equal in heat dissipation. Because a small amount of power is lost across the emitter resistors some amp designs use 1/4Ω (R22).
Darlington complementary is the basic order in how the output stage of an amplifier is taught. All output transistors are arranged as emitter followers. The collectors are connected directly to the rails. The emitter follows the signal on base within 650mV. The output transistors do not increase the size of the audio signal. Output transistors can only add current.
Quasi complementary is used in the majority of amplifiers. The first large silicon transistors (2N3055) enabled power amplifiers to be capable of 50 Watts but were only available as NPN and not as PNP. When PNP power transistors (2N2955) became available they were twice the price. The output transistors on the -V rail appear not to be wired as emitter followers. However the PNP driver transistor manages the output transistors collectively as a single compound large Emitter follower with a high HFE current gain.
Compound complementary NPN and PNP complementary output transistors designed for audio amplifiers are now available from many manufactures. Both NPN and PNP driver transistors manage the NPN and PNP output transistors collectively as compound single large Emitter followers with a high HFE current gain. The compound complementary arrangement has two advantages over the Darlington and Quasi complementary arrangements. Compound complementary has superior quiescent bias stability and the peak of the audio signal can get closer to the + - V rails, therefore slightly more power.
The negative feedback must be taken from the output of the amplifier. Power amplifiers have a signal gain of approx 20 to 40 (adjusted by R1 R2). But the additional driver and output transistors are now contained within the negative feedback loop, and this causes all power amplifiers to become unstable.
Power MOS-FET
MOS-FET Metal Oxide Semiconductor Field-Effect Transistors are a variation of a Bi-polar transistor and are used in some amplifiers. A transistor functions by having a small amount of current between the Base_Emitter to enable a larger current between Collector_Emitter. FETs only require a static electrical charge as a Voltage (3V to 12V) on the Gate to enable a current to flow between Drain_Source.
MOS-FETs can be easily controlled to turn on and off at high speed (Mega Hz) and are mostly used for switch-mode power supplies in computers etc, and are named Vertical MOS-FETs. Lateral Power MOS-FETs were developed for audio amplifiers during the 1990s. The primary disadvantage of FETs is that they deliver less power than a Bi-polar transistor amp using the same supply voltage. They are expensive, difficult to manufacture and only a few companies supply them. Less amplifier manufacturers use power MOS-FETs.
The above pic shows the difference between transistors and FETs using the same + - 70V supply.
(1) At the peak of the sine the on resistance of a transistor decreases with temperature.
At the peak of the sine the on resistance of a MOS-FET increases with temperature.
(2) At the peak of the sine the transistor Emitter follows the Base within 0.65V
At the peak of the sine the FET Source follows the Gate by approx 12V
The Source (output) will be 12V less than at the Gate. In technical terms a specified MOS-FET has a rated Vds (saturated voltage, Drain to Source) of 12V at full current, which is subtracted from the DC value of the supply voltage. In the above example the amplifier using Power MOS-FETs will deliver 60 Watts less power than the same amplifier using transistors.
To solve the 12V loss problem requires the Gate to be driven 12V above the 70V rail supply at the peak of the sine wave. To achieve this the driver circuit would need to be powered from a separate higher rail supply (90V example), to enable the input signal to reach at least 12V above the 70V amp rail supply. The 6V difference between the peak of the sine wave and the rail supply shown in the above pic could be reduced to a lower voltage enabling greater power. Most amplifiers that use power MOS-FETs do not have this extra circuitry.
wikipedia.org / MOS-FET
Paranormal beliefs exist about FETs sounding like valves, but what are the facts? The only similarity with FETs and valves is that the input Grid of a valve and the Gate of a FET require no current (Amperes) to function. Outside of this single point Valves and FETs have no similarity whatsoever.
Without being previously informed, it is impossible to hear or scientifically test if the output devices in an amplifier are Transistors or MOS-FETs. The $1,000,000 prize offered by the James Randi Educational Foundation for anyone providing proof of the paranormal, should also include any Audiophile who can prove under a double blind (A B comparison) to hear a difference between Transistors and FETs.
RMS power and music compression
From the previous description about the amount of heat generated by the transistors into the heat sink, the question arises - How is it possible for the majority of amplifiers not to be destroyed by overheating?
Music is capable of a 60dB (1,000,000:1) dynamic range. The transients in music are very small in energy but are approx 20dB above the RMS music level. The average RMS power of fully dynamic music can not go above -20dB of the amplifiers full power capacity without the transients clipping the rail supplies. 20dB is 100:1 so therefore a 100 Watt amplifier should not be driven above 1 Watt of RMS music level (over approx 1 minute of time) to avoid transients being driven into rail clipping. A 100 Watt amplifier can only be used at an average of 1 Watt with fully dynamic music. For this reason amplifiers less than 60 Watts should not be considered as audiophile status, but unfortunately many are.
The modern digital recording trend is to dynamically compress music in an attempt to remove all the dynamic range which includes transients. Dynamic compression allows music to be played at higher power without transient clipping. However excessive dynamic compression imposes extreme inter-modulation distortion. Voices and instruments are squashed and mangled together, rendering articulation of voices and instruments so removed from sounding natural that it is often difficult to recognise. The largest problem of this irresponsible recording behaviour in pop recordings, TV programs and films is that it makes it difficult to understand the words being sung or dialogue spoken. The inter-modulation distortion including the removal of articulation caused by dynamically compressing music is so great (approx 30% distortion), that audiophiles and professional sound installers pretending to be concerned about inaudible time alignment differences of speaker driver components on a baffle board is delusion to say the least.
Also dynamic compressed music is already so distorted by the dynamic compression in the recording process that it can be driven into supply rail clipping without being audibly noticed in comparison to the distortion created by dynamic compression. The maximum level an amplifier can be driven with dynamically compressed music before the added distortion caused by clipping into the rail supplies becomes objectionable, is 1/3 of the equivalent energy of a sine wave at full power. Worse still, in most live concerts the music is further compressed so the average RMS power can be taken close to 1/2 full power of the amplifiers capacity. In this condition many professional high power amplifiers will shut down from overheating.
Bridge amp advantage
Single ended is 1 amplifier driving a speaker. Single ended is the most commonly used application. However a speaker can be bridged between 2 amplifiers. Bridging a speaker between 2 amplifiers is one of the least understood concepts about amplifier management. Bridging 1 speaker between 2 amplifiers is commonly used in sound systems for vehicles where the supply Voltage is limited by the 12V battery.
From a 12V DC supply 4V RMS is the maximum that can be achieved from a single ended amplifier.
4V x 4V / 8R = 2 Watt. The reason 8Ω speakers are not used in vehicles.
4V x 4V / 4R = 4 Watt. The reason 4Ω speakers are used in vehicles.
Bridging a speaker between 2 amplifiers and driving one amp in opposite phase 4V + 4V = 8V RMS.
8V x 8V / 4R = 16 Watt. Therefore many vehicle sound systems use bridge amplifiers to power speakers.
A popular belief is that 4 times the power is achieved from bridging 2 amplifiers in comparison to single ended application. This is partially true, but there is no such thing as something for nothing. The above pic shows two 100 Watt amplifiers with + - 30V rail supplies. 20V RMS is the maximum from a + - 30V rail supply.
20V RMS into a 8Ω speaker is 50 Watt.
20V RMS into a 4Ω speaker is 100 Watt.
Bridging two amplifiers 20V + 20V = 40V RMS.
40V x 40V / 8R = 200 Watt.
By paying close attention we can see that 4 times the power is achieved from bridging 2 amplifiers delivering 40V RMS into a 8Ω speaker (200 Watt) if we are comparing it to a single ended amplifier delivering 20V RMS into the same 8Ω speaker (50 Watts). However when comparing a bridge amplifier delivering 40V RMS into a 8Ω speaker (200 Watt) to a single ended amplifier delivering 20V RMS into a 4Ω speaker (100 Watt) then bridge only appears twice as powerful.
The advantage of bridge is that it delivers the same power as a single ended amplifier with only half the rail Voltage. (40V RMS into 8R = 200 Watt) 40V RMS from a single ended amplifier requires + - 60V rails, whereas 40V RMS from bridged amps only requires + - 30V rails. With bridged amps the speaker is powered from both + - V rail supplies at the same time, instead of alternate between supply rails as with a single ended amp. Therefore Bridge amps make more efficient use of the rail supplies. Also the maximum Voltage across the transistors is half by comparison to single ended amp. Bridge is the most effective method to drive a speaker. The only disadvantage is higher cost.
Bridge management Bridging a speaker between two amplifiers is the most effective means to power a speaker. The power is supplied from both + and - V rails at the same time enabling twice the voltage across the speaker in comparison to using a single amplifier. The only disadvantage is cost.
A dual op-amp is often used to create a balanced signal. The first op-amp acts as a buffer with unity gain. The output of the buffer is sent to an inverting buffer to flip the signal 180deg. A perfectly balanced signal is then sent to the power amps.
There is also an alternate method that does not require a dual op-amp to create a balanced signal to be sent to the amplifiers. The output of the first amplifier is sent to the -inverting input of the second amplifier through a resistor that is the same value as R1. The second power amp now acts as an inverting slave. This is the simplest means to bridger 2 amplifiers as it only requires the addition of a single R1 resistor. The only disadvantage is that any distortion in the first amp is sent to the second amp, causing the distortion to be doubled.
sound-au.com/amp design Advanced essential reading
sound-au.com.net/amp sound more advanced essential reading
Class G Class H
There have been many attempts by amplifier designers to reduce the 30% to 50% wasted heat across the output transistors. By keeping the rail supply close to the peak of the sine wave, the heat dissipation across the transistors is kept to minimum. These designs require greater circuit complexity.
Class G has 4 fixed rails. 2 x +V supply rails and 2 x -V supply rails. There are 2 transistors in series for each + -V supply rail. The above pic only shows the +V supply only. The same arrangement is applied to the -V rail supply. At low level, power is taken from the lower Voltage rail by the 1st transistor. As the audio signal increases the second transistor connected to the higher Voltage rail starts to conduct. All the current flows through the 1st transistor to the speaker.
Class H gives a similar result to Class G and is slightly more efficient. The +V rail and a -V rail change voltage and increase when required. Class H requires the circuit to predict when a high transient input signal is about to appear. The rail Voltage must increase ahead of the audio signal for it not to clip. Because this is not always possible transient clipping distortion does happen.
Both Class G and Class H are sometimes used by hi-powered amplifiers that are expected to be used at low power for most of the time, hereby minimising the amount of heat wasted by being able to function alternatively from a higher to a lower rail Voltage. Class G is also used for domestic amplifiers with a small heat sink.
sound-au.com/amp-basics
sound-au.com/amp-design