Today I will discuss the choices made as I looked to design the Audio DAC portion of the project.
To design a PCB that can take I2S data as input and output analog audio.
As mentioned in my previous post, I initially wasn’t sure whether I wanted to drive a speaker or headphones. While not nearly as important as the amplification stage, deciding which I intended to use was important, because the output of this stage will feed directly into the amplifier. I did indeed decide to drive headphones, so I moved forward with that plan.
At a high level, a DAC is a fairly simple to understand: it turns a digital signal into an analog one. Like most other engineering problems, the devil is in the details. Prior to this project, I had essentially only worked with basic DACs prior to this (or at an abstraction level that made them quite simple to work with).
The DACs I was most familiar with would output a voltage scaled across a given range based on the input. For instance, a 10-bit DAC (like the TLC5615) would be capable of 2^10 = 1024 distinct voltage levels. Depending on the DAC, this would generally be scaled across the [V_min, V_max] range. For instance, a DAC capable of outputting 0-5V would use (5V - 0V) / (1024) = 0.0049V steps, so an input value of 512 would result in an output voltage of 5V * 512 / 1024 = 2.5V.
Current-Output DACs are similar, except that instead out outputting a voltage, they induce a current. This was new to me, and at first glance seemed less useful. I found this helpful document from Texas Instruments, and based on that (and some other reading that I forgot to make note of), it appears that at their core, DACs indeed first convert a digital signal into a current, and the Voltage-Output DACs then have an integrated I-V (or current-voltage) stage to convert the current to a voltage.
In audio applications, however, an engineer is very picky about aspects of this conversion that other applications may not care about. In high-end audio circuits, then, it appears that an engineer might choose to do the I-V conversion circuit on their own, based on their design parameters.
Regardless of whether I used a voltage or current output DAC, I also needed to decide on what the audio DAC PCB should output: a single-ended or differential voltage. A single-ended voltage is essentially just one signal that is referenced against ground, whereas a differential voltage is two signal lines, the difference of which represents the signal. A single-ended voltage is easier in the sense that it’s one signal line, but a differential voltage is less susceptible to noise.
I ended up liking the PCM5242 DAC. I liked the differential outputs because I am going off-board to the amplifier stage. I doubt whether that really necessitates differential signalling, but I like the thought of it. I feel that it keeps the signal as clean as possible on the input to the amplification stage, and this step is important. I am not constrained on number of pins or anything, so having extra signalling is not an issue.
I initially considered using a current-output DAC, but decided that I’d rather start with less circuitry where possible. I will still have an amplification stage, and having a separate I-V stage with some amplification seems redundant and an area for more problems. In the future, I may return to this stage, possibly combining the DAC and amplifier stage to keep signals clean. However, for now I will move forward with the PCM5242.
I researched several ICs that can be used to create an audio signal, and decided to go with a relatively simple-to-use one that should allow me to create a relatively clean signal to feed into my amplifier stage.
The next step here will be to put together the schematic and generate the PCB. As with the USB-to-I2S board, I expect that the schematic and board will be relatively simple.