Marantz SA8005 measurement and review (CD/SACD player, DAC)

[Luca Sorrentino]

Hello all, inspired by some readings of other reviewers here on ASR, and with the help of the Test CD send me by @NTTY, I decide to make some tests to my Marantz, a ten years old player I most use as a USB DAC with 44/16 material. I've employed a Motu M4 as ADC for measurements, I know not the most affordable instrument but enough to give me a good representation of how it works.

Let's start with a 0dB 1KHz sine:

We have a very clean spectrum at low frequency, with no power supply spurious at all, nice to see, a negligible jitter around the fundamental and a recalculated ENOB of 15.6 bit. Not bad, inaudible distortion too, this is left channel, same for the R.

This is an almost completely flat freq response in audio band.

Multitone signal shows almost 110dB of spurious free spectrum, very good.

Distortion vs frequency @ -12dBFS, good.

And now something I didn't like, not so good attenuation of ultrasonic aliasing of audio band, white noise spectrum reflect the triplet tone spurious.

Here is the magic, same signal but passed by USB input with 4x oversampling made with Foobar, instead of CD playing! We have a steep out of band attenuation of 90dB, was this circa 2015 player made with in mind a good DAC instead an already obsolete CD player???

Wow, same white noise signal but fed through different way, a test CD, by USB 4x oversampled, and DSD128 converted as the Marantz accept this format too.

Jitter seem not to be an issue, fundamental slightly modulated by 50Hz and near harmonics at very low level, a lonely spike at 460Hz I can' t imagine the cause.

In my undestanding plenty of headroom in intersample over test, why noise floor of 11khz signal is so high?

Police - Every little... an energic song. This is the whole song spectrum, peak and average value, probably the not perfect attenuation out of band is not a real issue, I don't know.

In the next few days I will complete this test entering the SDPIF input performance, I remember jitter is clearly visible.

This is my first measurement on a digital player, so please suggest, comment, I can edit and improve. I think I can't be precise measuring HR performance just with my Motu interface, and definitely I use to listen 44/16 material and to be honest, I can't hear any difference with higher resolution formats, my age (54)? Probably, or probably not.

Special thanks to the authors of REW and Multitone that I know populate this forum, and another time to @NTTY for the test signals and some instruction about using.

Luca

 

[NTTY]

Let's start with a 0dB 1KHz sine:
View attachment 405930

Nice! Was it 1kHz dithered?

You can increase the gain at the input of the Motu to get a better direct reading of the ENOB. But not really necessary, it’s just that noise is wrongly calculated because of the 10dB headroom you have at the input.

We have a very clean spectrum at low frequency, with no power supply spurious at all, nice to see, a negligible jitter around the fundamental and a recalculated ENOB of 15.6 bit. Not bad, inaudible distortion too, this is left channel, same for the R.

View attachment 405933
This is an almost completely flat freq response in audio band.

Impressive considering the zooming here.

View attachment 405943
Wow, same white noise signal but fed through different way, a test CD, by USB 4x oversampled, and DSD128 converted as the Marantz accept this format too.

Different filters are in action because of the different formats and sampling rate input. Noise shaping technique shows at 40kHz+

View attachment 405951
In my undestanding plenty of headroom in intersample over test, why noise floor of 11khz signal is so high?

You can calculate the THD+N removing low pass filter in distortion settings of RTA when running this test.
It seems you have a good headroom for ISO indeed. 11025Hz shows saturation of the oversampling filter, hence the raised noise floor and high distortion.

These are good performances!

 

[AnalogSteph]

And now something I didn't like, not so good attenuation of ultrasonic aliasing of audio band, white noise spectrum reflect the triplet tone spurious.

They must be using internal upsampling, as already seen on other D&M players... the CS4398 by itself ought to have roughly 100 dB stopband rejection up to fs at least, so running it "barefoot" would have yielded better results.

BTW, could you switch all the graphs to a dBr vertical scale consistently? That would make things easier to read.

Here is the magic, same signal but passed by USB input with 4x oversampling made with Foobar, instead of CD playing! We have a steep out of band attenuation of 90dB, was this circa 2015 player made with in mind a good DAC instead an already obsolete CD player???

You are aware that the filter slope you're seeing is from the software upsampler, right? When using 4x oversampling, you are feeding the player with 176.4 kHz, and fs/2 is at 88.2 kHz. You are not seeing anything of the filter response up there, just shaped noise from the DAC.

 

[Luca Sorrentino]

Nice! Was it 1kHz dithered?

It's yours "C_Sine_999.91Hz_0dB_-1dB_-3dB_-6dB_NoDither_L"

You can increase the gain at the input of the Motu to get a better direct reading of the ENOB. But not really necessary, it’s just that noise is wrongly calculated because of the 10dB headroom you have at the input.

Motu M4 has two mic inputs with variable gain, but this had far worst performance in terms of distortion, so I had to use in 3 and 4, no variable gain but better overall behavior.

Different filters are in action because of the different formats and sampling rate input. Noise shaping technique shows at 40kHz+

Yes I suppose, and completely different data stream as it was oversampled with RetroArch integrated in Foobar.

11025Hz shows saturation of the oversampling filter

Interesting, I didn't realize that! Since this is a high res player, I was thinking that 0dB@16bit didn't arrive at the MSB of the DAC or of the the oversamplig filter.

Luca

 

[Luca Sorrentino]

They must be using internal upsampling, as already seen on other D&M players... the CS4398 by itself ought to have roughly 100 dB stopband rejection up to fs at least, so running it "barefoot" would have yielded better results.

Thank you, Interesting.

You are aware that the filter slope you're seeing is from the software upsampler, right?

Yes, yes, I used RetroArch included in Foobar, I suppose it changes all the data? Or maybe is a spline type? Maintain actual data adding 3 more sample in between, I don't know. The steep filtering is made outside the player this time.

BTW, could you switch all the graphs to a dBr vertical scale consistently? That would make things easier to read.

Changed two plots as per your suggestion, I didn't save all the data, I'm not able to replot all of them.

Luca

 

[NTTY]

It's yours "C_Sine_999.91Hz_0dB_-1dB_-3dB_-6dB_NoDither_L"

Ah ok, so, slight deviation as the measurement shows 999.95Hz. You can use file #33 of the test CD which is pitch error test (19997Hz) to check further.

Motu M4 has two mic inputs with variable gain, but this had far worst performance in terms of distortion, so I had to use in 3 and 4, no variable gain but better overall behavior.

Ah ok, got it, I did not know. Well, that changes only the noise calculation in REW. And it’s good to have lots of headroom at the ADC input, especially with the IS over tests.

Yes I suppose, and completely different data stream as it was oversampled with RetroArch integrated in Foobar.

Yep, resampling with Foobar or Audacity requires filtering which is performed by the resampler in digital domain. So you get a filtered output, which will be seen at 4x the initial sampling rate by the Player and so a different filtering will be activated when it will process the file.

Interesting, I didn't realize that! Since this is a high res player, I was thinking that 0dB@16bit didn't arrive at the MSB of the DAC or of the the oversamplig filter.

Luca

Yeah, there’s a long thread here about testing resistance to intersample-overs. Some DACs will have few dBs headroom but that means they sacrifice them on SINAD and ENOB.