Following the previous post, this post will mostly discuss how amplifiers work and just a few words about class-D amplifiers. The software of choice, for the schematic work, was KiCad. This was just because of some arbitrary roam on the web, BTW.
Starting off from the vendors recommended reference design of the LM4808, the design here is a unity gain amplifier. First of all, what is a unity gain? Actually, it means no gain. Assuming an ideal OP-Amp, it is assumed that it has infinite gain. So using a standard inverting amplifier topology (courtesy of QUCS, see my post!).
let us start analyzing it. There are two (dependent) characteristics of the ideal OP-Amp.
1. It (infinitely) gains the difference between the positive and the negative input.
2. The difference between the positive and the negative input always “aspires” towards zero.
Why are these two dependent? Well, intuitively, energy must be conserved. Infinitely gaining a non-zero difference gives the sound mind an unhealthy headache. Hooray then. On to the next step.
As the difference goes to zero, the “equivalent” circuit is obtained
One might think: This is stupid. The current from the source actually goes directly to the ground. Well, this is a way of analyzing the non-linear component known as the Operational-Amplifier. Resolving the voltage distribution obtains:
It is common to refer to 
Looking at the vendors suggested spec:
Found at the TI website, we see that a feedback and source resistors are set to the same value. Hence obtaining a gain between Vout and Vin of 1 (magnitude, ignoring the minus). This configuration is known as a unity gain amplifier. Notice that close to the audio input there is a 2.2uF capacitor. This is mostly used as a DC block. The Direct current sees the capacitor as an open ciruict, namely it shall not pass. How it affects the feedback, however is a different deal.
The reader is more than welcome to check it out for him\herself, but the total gain of the system, in steady state terms, turns out something like this:
Plotting this expression as a function of f shows that this function has a first order zero (the quotient has f in the power of 1) and a first order pole (the denominator as well). This expression allows to calculate the exact frequency of the pole. In this case 18.6 Hz. Namely this is really a DC Block. If this is an audio amplifier, it is important to remember that most people’s sensitivity starts from at least 60 Hz, while most headphones can’t even sound that.
The second trick used here, the most confusing part in my opinion, is what is connected to the positive inputs of the amplifier. Using two 100 kOhm resistors, the positive, non-inverting, input of the amplifier is “floating” at Vdd/2. Why is this? well, this configuration is common in a single-rail amplifier. First thing first. What is a single rail amplifier! It means that the amplifier is connected between the Vdd (such as the battery voltage) and the ground. The better, yet trickier to obtain, version, is to connect the amplifier between +Vdd and -Vdd, giving the amplifier the ability to amplify the whole range between them. So for that extent, the non inverting input is “floated”, thus raising the output signal such that it swings around Vdd/2. So lets build it
Simulating this, however, obtains a surprising result. Where did the gain go!
It seems, for some reason, the the gain dropped in 90dB, namely 1 billion times lower. Well, lets consider this: At the audio input, the DC was blocked. Afterwards, the voltage swing was “floated” to Vdd/2. Then, at the output, it is still swinging over and under the Vdd/2 line! So basically all you need to do is add another DC block (for some reason now called a decoupling capacitor) at the output.
Notice that because of the decoupling capacitor, more low frequencies are now attenuated. This can be resolved, but the only test in case of audio devices is the hearing test. If it sounds okay, keep it the way it is. If you are missing the bass, put in a larger capacitor. One more question remains:
Why a Class-D amplifier
First, It’s of the shelf. That always makes me happy. Seriously, though. Class-D amplifiers are from a family of switch-amplifiers. This amplify by comparing the input signal to a generated alternating pulse (in this case, a triangle wave) and then return the signal to it’s “original” form by filtering out the un-necessary parts. This explanation was so coarse you can grit cheese on it, however if you want to learn more about this you can read about it here, or in any other dark corner of the web. Unlike standard power amplifiers, though, when the input is zero, they do not condut at all. Namely, at these situations, there is no power dissipation. Hence one can say that these amplifiers, although less efficient in terms of signal, they dissipate less power, hence consume less power overall.
Hardcore audio fans will normally toss these aside, as they are impure inbred dogs of amplifiers. however for this purpose of an in-flight device, it is (or might be) perfect. Great! All that is left for this guy, then, is to place it in a schematic editor. Next time I will discuss about the final tweaks for the amplifier circuit and what else is needed for it to work.
Cheers!