I've long been a fan of the Apple TV and have been using some version of it as my main A/V source since 2014. I've made a few scattered measurements of various Apple TVs over the past few years but recently made some systematic measurements of both the Apple TV 3 (A1469) and 4K (A2169) and wanted to start a thread to compile them.
Although the Apple TV 3 has a TOSLINK output, I used the HDMI output of the 3 to make the signal path identical between the two devices. The signal path was Apple TV HDMI -> Monoprice HDMI Extractor -> optical to coaxial convertor -> RCA to BNC Cable -> Neutrik NADITBNC-M Transformer -> AES cable -> Okto dac8 pro. For digital measurements the Okto was used as an AES to USB capture device while connected to a 2015 MacBook pro running REW. Test files were generated using @pkane's multitone analyzer and played via airplay from the MacBook using Quicktime, this allows for 24 bit streaming. Although the signal path is a bit complicated, it is the actual signal path I use everyday with the Apple TV 4K and I wanted to make sure the measurements were as real world as possible.
Let's first look at the Apple TV 3 at 48 kHz.
This result is perfect and shows 24 bit transparency. As this is done at 48 kHz and the AppleTV outputs at 48 kHz there is no evidence of resampling.
J test looks very clean.
Again, very clean, no IMD to speak of.
Now looking at the Apple TV 3 at 44.1 kHz. In this case the Apple TV is resampling to output 48 kHz.
You can now see some resampling artifacts, but in terms of performance you will still be limited by the downstream DAC, not the Apple TV.
J test also shows some resampling artifacts but nothing serious.
Again minor resampling artifacts in the 19 + 20 kHz test.
I then set the AppleTV as the main MacBook audio device which results in more typical 16 bit airplay streaming. For these measurements I only used 44.1 kHz files as I wanted to avoid the additional variable of Core Audio resampling.
Other than elevated noise due to 16 bit airplay, 1 kHz looks perfect.
J test looks a little funky, not entirely sure what to make of this, especially as the Apple TV 4K does NOT show the same issues when using 16 bit streaming.
Like the J Test, 19 + 20 kHz looks weird, with oddly placed spurs.
Now moving to the Apple TV 4K.
A bit more noise than the Apple TV 3, but still will not be limiting compared to downstream DAC. Skirt is a bit wider here which gets more interesting with the higher frequency measurements.
We now have spurs around the fundamental which were completely absent with the Apple TV 3.
Again, we similar spurs.
Moving to 44.1 kHz with the Apple TV 4K.
Pretty much same performance as 48 kHz with a few new harmonics barely poking up.
Similar artifacts as the Apple TV 3 due to resampling but they are a bit harder to see due to higher noise. Like the 48 kHz measurement there are spurs around the fundamental.
Same spurs as other high frequency measurements. Interestingly, apart from the spurs around the fundamental this looks cleaner than the Apple TV 3 which implies they treat resampling differently.
And finally, 16 bit streaming with the Apple TV 4K using 44.1 kHz files.
Similar story, spurs around the fundamental but otherwise actually a bit cleaner than the Apple TV 3.
And same story, spurs around the fundamental but otherwise cleaner than the Apple TV 3.
Not much to say about these other than I doubt there are audible differences. Would be fun to compare files sent to the Okto directly via TOSLINK vs Apple TV 3 vs Apple TV 4K in a ABX test and see if anyone can distinguish a difference. Only other thing of note is I've found that in devices that are doing some sort of rate adjustment (AppleTV, Wiim, CamillaDSP) to bridge two clock domains it causes the FFT to hiccup when this happens. On the Apple TV 4K this happened much more frequently than the Apple TV 3, to the point that it made capturing clean measurements difficult. Not sure if this is inherent to the design of the devices or a specific issue related to my devices (i.e. Apple TV 3 clock is closer to MacBook pro clock and requires less adjustment).
Michael
Although the Apple TV 3 has a TOSLINK output, I used the HDMI output of the 3 to make the signal path identical between the two devices. The signal path was Apple TV HDMI -> Monoprice HDMI Extractor -> optical to coaxial convertor -> RCA to BNC Cable -> Neutrik NADITBNC-M Transformer -> AES cable -> Okto dac8 pro. For digital measurements the Okto was used as an AES to USB capture device while connected to a 2015 MacBook pro running REW. Test files were generated using @pkane's multitone analyzer and played via airplay from the MacBook using Quicktime, this allows for 24 bit streaming. Although the signal path is a bit complicated, it is the actual signal path I use everyday with the Apple TV 4K and I wanted to make sure the measurements were as real world as possible.
Let's first look at the Apple TV 3 at 48 kHz.
This result is perfect and shows 24 bit transparency. As this is done at 48 kHz and the AppleTV outputs at 48 kHz there is no evidence of resampling.
J test looks very clean.
Again, very clean, no IMD to speak of.
Now looking at the Apple TV 3 at 44.1 kHz. In this case the Apple TV is resampling to output 48 kHz.
You can now see some resampling artifacts, but in terms of performance you will still be limited by the downstream DAC, not the Apple TV.
J test also shows some resampling artifacts but nothing serious.
Again minor resampling artifacts in the 19 + 20 kHz test.
I then set the AppleTV as the main MacBook audio device which results in more typical 16 bit airplay streaming. For these measurements I only used 44.1 kHz files as I wanted to avoid the additional variable of Core Audio resampling.
Other than elevated noise due to 16 bit airplay, 1 kHz looks perfect.
J test looks a little funky, not entirely sure what to make of this, especially as the Apple TV 4K does NOT show the same issues when using 16 bit streaming.
Like the J Test, 19 + 20 kHz looks weird, with oddly placed spurs.
Now moving to the Apple TV 4K.
A bit more noise than the Apple TV 3, but still will not be limiting compared to downstream DAC. Skirt is a bit wider here which gets more interesting with the higher frequency measurements.
We now have spurs around the fundamental which were completely absent with the Apple TV 3.
Again, we similar spurs.
Moving to 44.1 kHz with the Apple TV 4K.
Pretty much same performance as 48 kHz with a few new harmonics barely poking up.
Similar artifacts as the Apple TV 3 due to resampling but they are a bit harder to see due to higher noise. Like the 48 kHz measurement there are spurs around the fundamental.
Same spurs as other high frequency measurements. Interestingly, apart from the spurs around the fundamental this looks cleaner than the Apple TV 3 which implies they treat resampling differently.
And finally, 16 bit streaming with the Apple TV 4K using 44.1 kHz files.
Similar story, spurs around the fundamental but otherwise actually a bit cleaner than the Apple TV 3.
And same story, spurs around the fundamental but otherwise cleaner than the Apple TV 3.
Not much to say about these other than I doubt there are audible differences. Would be fun to compare files sent to the Okto directly via TOSLINK vs Apple TV 3 vs Apple TV 4K in a ABX test and see if anyone can distinguish a difference. Only other thing of note is I've found that in devices that are doing some sort of rate adjustment (AppleTV, Wiim, CamillaDSP) to bridge two clock domains it causes the FFT to hiccup when this happens. On the Apple TV 4K this happened much more frequently than the Apple TV 3, to the point that it made capturing clean measurements difficult. Not sure if this is inherent to the design of the devices or a specific issue related to my devices (i.e. Apple TV 3 clock is closer to MacBook pro clock and requires less adjustment).
Michael
Last edited:
