Using a "non-audio" oriented DAC IC as an audio DAC

Question

I'm starting a project that consists of a simple embedded 16-bit PCM audio player, where no actual DSP takes place, only plays PCM and outputs something you can then amplify. I am currently searching for my DAC IC, and I'm hesitating between these two references from Texas Instruments :

• The DAC7642, a "non-audio" applications-oriented R-2R DAC, parallel data input.
• The PCM1773, an audio-applications-oriented sigma-delta DAC, serial data input

The thing is, even though the DAC7642 isn't meant for actual audio applications but rather for industrial/motor control applications, it uses the R-2R architecture which I highly prefer over the Sigma-Delta one, and it has parallel data input which I find way more convenient than the PCM1773's serial data input.

My question is: Are there any big drawbacks that should prevent me from using the DAC7642 as an audio DAC (Increased noise in audio output? Mediocre performance at clock rates needed by audio to output analog voltages at rates like 44.1 kHz?) Or can I just go with it and forget about the PCM1773?

Answer 1

Are there any big drawbacks

1. Price. It looks like DAC7642 is 8-10 times more expensive
2. Nosie performance. The DAC7642 is weirdly specified but it looks like maybe 14-bit are actually usable for audio and the linearity and monocity errors eat up the rest.
3. Clocking: You need to clock out samples are regular rate and audio is quite sensitivity to clock jitter. I'm not sure that the DAC7642 has a settling time that's independent from the data.
4. System integration. For the audio DAC you just pop down a clock chip, hook up a few I2S lines are you are done. For the DAC7642 you need to build a large parallel bus and make sure all your timings are correct.
5. You may need extra anti aliasing filtering.

The PCM1773 is optimized for audio in terms of interface, ease of use, configuration and performance (especially bang for the buck). The DAC7642 is designed as an asynchronous control DAC for a completely different application. It's going to be a lot of work to make it work. Even if you get it running it still won't perform as well as the PCM1773 .

Answer 2

The other chip is a raw DAC. The other chip is specifically meant for audio.

The PCM1733:

• Has a standard serial audio interface to connect to standard audio outputs on MCUs so it can be easily swapped to another DAC if there is need
• Has 8x digital oversampling before sigma-delta DAC, anyway the output rate is high above sampling rate for building a simpler and cheaper audio reconstruction filter
• Has interface to control e.g. output volume internally

The DAC7642:

• Has none of the above things to help you
• Requires custom wide bus with clock triggers, you could use shift registers to make a serial interface for it
• Has no interpolation or oversampling. So it has to be fed with higher rates and you need to make interpolation and oversampling yourself if you want it
• If you don't use interpolation/oversampling then you need awfully steep, complex and expensive DAC output filter
• Has no interface for control. Needs external volume control or then audio data needs to be processed to have lower volume before sent to DAC.

So it depends to what chip you will connect the DAC and how you will generate audio, you might have reasons to choose the raw DAC, but for simple playing of audio streams, it makes no sense to choose the raw DAC.

Answer 3

Probably the biggest difference is that an R-2R ladder is likely to have perceptible non-linearity in certain steps, for example when the low bits change from 011111 to 100000; this can significantly colour an audio signal in a bad way. Sigma-delta devices tend not to do this.