Matching and tuning audio amplifier output stability and sound performance.

This “Product How-To” article focuses how to use a certain product in an embedded system and is written by a company representative.

Audio amplifier products are becoming more and more advanced in termsof product outlook, style, system control and sound quality.Experienced electronic engineers are using different sorts of circuitryin their audio amplifier designs.

Many traditional high-power audio amplifiers with over 100W/ channelare using discrete components circuitry. High-fidelity audio amplifiereven needs to take extra effort on matching and tuning the outputstability and sound performance.

Considerations
There are different approaches to developing an audio system. NationalSemiconductor's LME49810 is optimal for high-end consumer andprofessional audio applications that span powered studio monitors andsubwoofers, A/V receivers, commercial sound reinforcement, after-marketaudio, professional mixers, distributed audio and guitar amplifiers. Itis also well-suited for a wide range of industrial applications, wherehigh voltage and low distortion are required.

Input stage design is a critical consideration in the amplifier. Itgenerates error signal by subtracting the feedback signal from theinput and then drive to the output. The error signal is assumed to besmall to provide high linearity for the amplifier. LME49810 is abipolar input amplifier.

Figure1: To further reduce the output offset voltage, reducing the impedanceboth on feedback and input signal path is a way to do.

The input impedance matching is important. Due to the bias currentfrom the positive and negative input port, input impedance mismatchleads to input offset voltage. The input offset voltage will amplifywith close loop gain and appear on the output.

While LME49810 has a very low input bias current, the offset voltagethat will appear on output will be insignificant for general uses. Forexample, a typical LME49810 bias current is 100nA and 1K-ohms inputimpedance mismatch. Voffset = Ibias x Rmismatch = 100nA x 1K-ohms =0.1mV. For a typical 30x close loop gain in the amplifier,Voutputoffset = 0.1mV x 30.

Due to very low input bias current, the offset voltage that appearson output will be insignificant in general uses. To further reduce theoutput offset voltage, reducing the impedance both on feedback andinput signal path is a way to do. The offset due to input bias currentimbalance can be reduced. But we need to make sure that there is enoughdriving capability from the front stage.

There are generally two common types of audio input design: AC- orDC-coupled. The advantages of AC-coupled input are the following:

1) The DC-offset to the inputof the amplifier from pre-amplifier, filter stage or codec stage isalways zero.
2) There is no need to protectthe DC faults by any DC servo circuitry in the amplifier.

On the other hand, the DC coupled input offers the followingadvantages:

1) No large and expensive AC-coupling capacitor is required.
2) Low frequency distortiongenerated by the AC-coupling capacitor is avoided.3) It reduces thenoise coupled from the AC-coupling RC network.

Figure2: By increasing the value of the capacitor between these two pins, theamount of compensation and the phase margin can be increased. Thesuggested range is from 10p to 100p.

Negative feedback configuration in the power amplifier can providehigher stability and linearity of the system. Using negative feedbackcan avoid the saturation on the amplifier DC offset voltage.

Generally, there is phase shift at high frequency in the amplifier;a higher negative feedback factor can reduce the high frequencyinstability and oscillation.

In a discrete amplifier system, high feedback factor will cause poortransient response or high frequency instability. LME49810 has a higheropen loop gain.

Thus, it has lesser close-loop gain error and power supply rejectionand can maximize the amount of negative feedback in the circuitry tomaximize the linearity. A recommended range of voltage gain is 30-40dB.Negative feedback factor can be easily set by the R input and Rfeedback relationship: AV = RF/RI (V/V).

Compensation
The compensation of the amplifier is to adjust the open loop gain andphase performance to stabilize the system when the feedback loop isclosed. Generally speaking, the amount of compensation is as high aspossible to obtain a higher stability. However, it has a tradeoff onreducing the bandwidth and slew rate of the audio chips.

Lower slew rate will provide a softer sound nature, while higherslew rate can provide a clearer and harder sound nature in general. InLME49810, Miller compensation is implemented by inserting a singlecapacitor between the pin Comp and BiasM.

Figure3: It is not recommended to exceed 200µA current through the mutepin.

By increasing the value of the capacitor between these two pins, theamount of compensation and the phase margin can be increased. However,it is not recommended to be too high or too low.

The suggested range is from 10p to 100p.The compensation capacitorshould be low equivalent series resistance (ESR); it can avoidintroducing a potential zero by the ESR of the capacitor. Usually, aceramic capacitor is better than an electrolytic capacitor.

The mute pin is controlled by the current flow into the mute pin.This is done by connecting a reference voltage to the mute pin througha resistor to control the current from 50 to 100 microAmps as “play”mode and less than 50µA as “mute” mode. It can be calculated by adefault formula: IMUTE = (VMUTE – 0.7V) / (RM + 10k-ohms) (A). It isnot recommended to exceed 200 microAmps current through the mute pin.

Figure4: LME49810 has two dedicated pins (BiasP and BiasM) for Bias setup andprovides a certain output bias current.

Output
LME49810 has two dedicated pins (BiasP and BiasM) for Bias setup andprovides a certain output bias current. The R pot (variable resistor)is for tuning the bias current for the output stage. Lowering the R pot+ Rb1 will result in a higher bias voltage.

The QMULT (multiplier) compensates the bias voltage to prevent thethermal runaway of the bipolar output transistors.

The QMULT must be attached with the same heat sink as the outputtransistors. When the temperature gets higher, the Vbe is decreased tolower the bias voltage – the higher the bias current provided the lesscrossover distortion from the output, but higher current consumption.

The most common output stage in audio power amplifier isemitter-follower (Figure 5 below ).It is usually called double emitter-follower or Darlington pair configuration, where the first followeracts as driver to the output device.

Figure5: The most common output stage in audio power amplifier is emitterfollower.

The large signal's linearity of the emitter-follower mainly dependson the loading. As the loading increases (i.e. loading resistancedecreases), the output current increases and the bipolar junctiontransistor current gain reduces due to the RE (emitter de-generator)and the beta rolloff at high current density.

It may reduce the linearity and increase the distortion at theoutput stage. For higher power, multiple stages output is recommendedto maintain the high current and better linearity. LME49810 audiodriver has around 50mA output current. It may be configured as singleDarlington pairs or parallel transistors output, depending on theapplication requirement.

Voltage rating
Double emitter-follower or Darlington pair usually have a high currentgain factor. The current gain factor should match the currentamplification on both the negative side and positive side to increasethe stability of the output stages.

For parallel transistor configuration, make sure that the mid-powertransistor is in the drive capability. The output current of the mediumpower transistor must be greater than the minimum driving current ofthe high power transistor to prevent overloading on the medium powertransistor stage.

An important consideration in output transistors is the voltagerating. It ensures the stability of the system and prevents it from anydamages. The VCBO, VCEO voltage maximum range must be larger than thesupply voltage rail to rail range. For a ±100V voltage supplyamplifier, the transistor voltage rating should be higher to ensurethat it operates within the specification.

In Figure 6 below , the midpower transistor's base and collector pinwill obtain around two times the Vee or Vcc voltage during operation.So it is necessary to ensure that the voltage rating is enough duringthe selection of the transistors.

Figure6: The mid power transistor's base and collector pin will obtain aroundtwo times the Vee or Vcc voltage during operation.

Resistor RE
In the high power audio amplifier application, the matching, currentbalancing and protection of the output transistor are important to thelinearity of the power amplifier. Resistor RE is recommended to improvethe matching, current balancing and protection of the outputtransistor.

Thus, it is suggested to have such a resistor in real applicationfor high power audio amplifier. While RE connected as series with theoutput transistor will also reduce the linearity of the amplifier, theresistor RE is the main source of distortion in the crossoverdistortion, which occurs at the one side of output transistor.

With this situation, the value of RE should be optimized and kept aslow as possible for reducing the nonlinearity. Thus, reducing theresistance of RE is a way to improve the crossover distortion.

Figure7: The RE is the main component of the open loop output impedance.

For the same equivalent RE, a parallel type output can reduce theoverall RE for improved linearity. At the same time, larger value of REfor each stage can improve the matching and current balancing for theoutput transistors.

RE is also related to the output power loss; the higher outputcurrent will bring a higher power loss on same RE. The value of the REdepends on the number of parallel output transistor and the speakerloading.

Generally, 0.1-0.5 ohms resistor with enough rating can be used. REpower rating is a calculation on the R x I2 = W (e.g. 0.1 ohms x 5A2 =2.5W). The RE is the main component of the open loop output impedance.

The closed loop output impedance is defined by the open loop outputimpedance and the negative feedback factor as follows:

The average open loop output impedance is 200m-ohms, and thenegative feedback factor is 29dB or 28 times. We expect the close loopoutput impedance is approximately 7m-ohms. The close loop impedance isusually recommended to be very low compared with the impedance in thespeaker system to give a higher linearity.

Typically, power amplifier closed loop output impedance is as low as10-50m-ohms. For a certain negative feedback factor, the RE should bekept as low as possible.

Output network
A usual output network called “Zobel” for a power amplifier withtypical component value is shown in Figure 8 below. All kinds of outputnetwork have only one purpose – it is to improve the stability. Thisrecommended network Zobel contains a resistor and a capacitor in seriesfrom the amplifier output to ground. The resistor is for limiting thecurrent on a certain higher frequency to reduce the capacitor ratingrequirement. Approximate resistor value range is from 4.7 to 10 ohms.

Figure8: The recommended network Zobel contains a resistor and a capacitor inseries from the amplifier output to ground.

In most cases, the capacitor is about 0.1 microFarads to reduce theloading effect of high frequency to the output speaker. The currentdraw in the Zobel network increases as the output levels increase.

So the rating of the output network components should be enough atany condition. For example, in a 20Vrms output swing, the Zobel networkdraws 248mA or 0.62W at 20kHz. To ensure that the component canmaintain high power and high frequency content, however, a 3W to 5Wpower rating component is usually recommended for the Zobel network.

One other function of the Zobel network is to protect the amplifieroutput from the inductive reactance in the loudspeaker coil. A realload impedance of a speaker is far more complex than a simple parallelresistor and capacitor.

The frequency response of a real loudspeaker speaker shows unevenimpedance load as the series combination of the ampli- fier outputimpedance and cable resistance and inductance. It could cause highfrequency instability.

Adding a small coil inductor in series with the amplifier outputincreases the instability. It isolates the amplifier from the shuntcapacitor without causing significant loss at the audio frequencies.The inductor value normally ranges from 1 to 7 microHenrys. A suitablevalue can be set to avoid high frequency roll-off at a certain loadingimpedance.

The air-cored inductor is recommended to eliminate the possibilityof magnetic saturation. A damping resistor across the output coil isalso recommended to reduce the Q factor of the output LC network,overshooting and ringing.

Conventionally, a few ohms wire-wound type damping resistor is usedto avoid self-inductance. Lower ohm value will bring less overshoot andringing. As the output current depends on the output power and loadingimpedance, a 1W to 5W rated damping resistor should be chosen.Generally, a 0.1-0.5 ohm resistor with enough rating can be used. A0.25 ohm resistor is recommended for general use on two to three stagesparallel tape driving.

There are different types of configuration to design an audioamplifier. The design consideration based on both theory andpracticality can enhance the performance and stability on the audiosystem. As the market leans toward hi-end audio systems, professionalamplifiers need more output power delivery, higher linearity and higherstability.

Kenneth Lee is Senior Product Application Engineer at National Semiconductor Corp.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.