MacBook Air M4 internal DAC

[trondav]

I have recently purchased the incredible beauitful blue MacBook Air M5 15" laptop. It is totally awesome. The best part .... the internal DAC. It sounds incredible good. I have had the Audioquest Black, and Red, Queststyle M12i, Chord Mojo, Chord Mojo 2 and iFi Hip DAC V1 and V2. Because of the hassle of charging the DAC/amplifier device I now only stick to the Audioquest Cobalt. More and more it is more than good enough to use the internal DAC in the MacBook Air M4 and send output directly to the 3.5mm jack - Via Roon it will send up to 96/24.

Anyone with knowlededge of techinal specs of the DAC Apple are using here?

It is for sure the best sound I have ever had in any laptop straight out of the box.

 

[nerdstrike]

As the "tool for the artist", Macs have long been tolerably well specified for headphones.

I tried to send audio to both internal DAC and a USB DAC on an M1 laptop. It didn't go well at all, but that's pretty niche.

 

[Roland68]

As the "tool for the artist", Macs have long been tolerably well specified for headphones.

I tried to send audio to both internal DAC and a USB DAC on an M1 laptop. It didn't go well at all, but that's pretty niche.

What was the problem?

 

[AnalogSteph]

Anyone with knowlededge of techinal specs of the DAC Apple are using here?

They do tend to measure quite well... up to 2.5 Vrms of output, dynamic range 116 dB(A) or something, SINAD 90ish dB. Plenty serviceable for onboard audio. As of the M1 the codec used was a CS42L84A but I don't think there's a public datasheet for that one (annoying).

I tried to send audio to both internal DAC and a USB DAC on an M1 laptop. It didn't go well at all, but that's pretty niche.

Combining multiple audio devices into one that don't have a common master clock is always risky business. There's a reason why studios use wordclock sync.

 

[voodooless]

I tried to send audio to both internal DAC and a USB DAC on an M1 laptop. It didn't go well at all, but that's pretty niche.

I use this every day, and it works just fine: desk speakers over USB, subwoofer over headphone output, filtering via Element.

Combining multiple audio devices into one that don't have a common master clock is always risky business. There's a reason why studios use wordclock sync.

In practice, this isn't really an issue. MacOS will resample one of the two outputs to match the other. I'm pretty sure it will do just fine for things other than audio production.

 

[AnalogSteph]

In practice, this isn't really an issue. MacOS will resample one of the two outputs to match the other.

Resample as in convert between 44.1 and 48 kHz etc., or does the audio stack feature an ASRC like VLC? I don't consider syncing up two independent DAC a particularly easy task, you typically don't know they're drifting apart until one starts to develop buffer underruns / overruns.

 

[voodooless]

Resample as in convert between 44.1 and 48 kHz etc., or does the audio stack feature an ASRC like VLC?

No, it actually compensates for the clock domain drifts between the various devices. So even if they are at the same rate, it will still use a kind of software PLL to compensate for the exact clock differences and resample audio to compensate.

I don't consider syncing up two independent DAC a particularly easy task, you typically don't know they're drifting apart until one starts to develop buffer underruns / overruns.

I don't know how they measure this... This kind of stuff is indeed not trivial

 

[Roland68]

Resample as in convert between 44.1 and 48 kHz etc., or does the audio stack feature an ASRC like VLC? I don't consider syncing up two independent DAC a particularly easy task, you typically don't know they're drifting apart until one starts to develop buffer underruns / overruns.

On a Mac, bundling multiple output devices is done using the Audio-Midi-Setup app. It's possible to select one device as the clock source and also enable drift correction for the other devices.

I've used this several times and haven't noticed any latency or issues with the same or similar DACs over USB, even after several hours.
But that will depend on the individual case and will of course also be device-dependent.

 

[AnalogSteph]

No, it actually compensates for the clock domain drifts between the various devices. So even if they are at the same rate, it will still use a kind of software PLL to compensate for the exact clock differences and resample audio to compensate.

Interesting.

I don't know how they measure this... This kind of stuff is indeed not trivial

I suppose you would pick one device as your reference and then watch your buffer levels on the others. If one always runs low, the PLL (or FLL) needs to speed up, while if the buffer threatens to overrun, it needs to slow down. It is obviously crucial to only act on long-term averages, and the PLL/FLL bandwidth needs to be kept fairly low as we only want to be compensating for small variations of playback speed (like <0.1%?).

EDIT:

On a Mac, bundling multiple output devices is done using the Audio-Midi-Setup app. It's possible to select one device as the clock source and also enable drift correction for the other devices.

So pretty much like was I was expecting then.

 

[nerdstrike]

What was the problem?

A cable splitter was an acceptable workaround... but yeah. I recall disintegrated audio coming out of one DAC. Ghastly noises and stuttering. From a general Unix point of view, it ought to have worked well enough (ignoring audio synchronising challenges), but I was following an approach that was written for Intel-based Macs so who knows? The transition to Apple silicon was lumpy in many ways, so I gave up.

Apologies to @trondav for this diversion!