# What determines if a particular DAC is “good” for audio?

I want to generate low-end audio (sinewave, sampling rate 8kHz). How to determine if a particular DAC is okay for such application? Is it only the settling time (of course apart from resolution, interface, voltage etc.)?

For example the AD5601 has a maximum settling time of 10µs. At a rate of 8kHz a sample takes 125µs, so I would guess that this particular DAC could handle the sample rate. Is that correct?

The AD5601 allows data bits being sent at 30 MHz, each packet takes 16 bits.

$\frac{30 MHz}{16}=1.875MHz$.

That means that you can theoretically make waveforms up to the Nyquist frequency which is $\frac{1.875 MHz}{2} = 937.5 kHz$.

$937.5 Khz > 8kHz$

The slew rate is 5 V/µs, that means that if you got a signal bouncing between 0 and 5 volt, a sine wave of 1Hz of amplitude 2.5 V has a maximum dv/dt at sin(0) => dv/dt = 1*2.5V => 2.5V, so a 2Hz sine wave got a dv/dt of 5V at t = 0.
5V/µs => X/s = 5*10^6/s, so 2Hz*10^6 => the slew rate supports a whopping 2 Mhz sine wave.

So yes, you are correct. That DAC is good enough for making sine waves sampled at 8kHz. What you said, 125 µs and 10 µs, together with the data speed & slew rate => why I said you were correct.

But let's reverse the formula, say you want to output what you sampled.

$8 kHz × 2 = 16 kHz$

$16 kHz × 16 = 256 kHz$

This means that you need to send one bit every $\frac{1}{256 kHz} = 3.9 µs$.

• To put it in more precisely: You can make square waves of arbitrary height at 937.5kHz. If you want a sawtooth with only one intermediate step, you are down at 470kHz, a triangle with only one intermediate step can be output at 235kHz and so on. The frequency goes down very quickly with accuracy. – Janka Jul 15 '17 at 10:09