|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
74.74.162.169
An over $400 DAC, I see high level of high frequency jitter (+ / - 3 kHz).
Would you return it because of the jitter?
Measurement set up.
Sound source:
Pierre Verany Digital Test CD 10kHz 0dBFS sinusoidal tone track
played on iTunes (volume max) on Mac mini.
Analyzer:
Focusrite Scarlett 2i2 line input, Arta software on Toshiba laptop with Windows 8.1.
When this over $400 DAC (Styleaudio Sapphire LE) receives digital signal from Mac mini via optical cable, there is plenty of high frequency jitter. When the anlalyzer is fed from Mac mini's analogue output, the jitter is significantly reduced. To check whether the optical cable is defective, I connected MuxLab 500081 (a $90 dac for those who have TVs on which choosing 2ch pcm output is impossible, and want to connect TV to 2ch stereo system.) via optical cable. The jitter level is much lower than the over $400 DAC.
I will post 3 pictures.
Follow Ups:
Hi,
I have no interest to comment on the specific products you mention, however given that the CS8416 seems to crop up a fair bit, I thought it is worthwhile mentioning that it has some peculiar "Features" (as in "it's not Bug/Bugfix, it's a feature").
You can read Cirrus Logic's own take on this here:
http://www.cirrus.com/en/pubs/appNote/AN339REV1.pdf
A more indepth discussion is here:
http://www.diyhifi.org/forums/viewtopic.php?f=2&t=1879
The upshot is simple.
The CS8416 has a operation mode where it emulates in effect the CS8414, including the limitation of the sample rate to 96kHz. In this mode the random noise-like jitter from the CS8416 VCO itself is suppressed very well (by an action that is in effect negative feedback), but any jitter from the source passes through to the DAC unattenuated, unless it is above the audio band.
This is the classic SPDIF receiver behaviour and is replicated by earlier Cirrus Logic, AKM etc. parts. It is the cause for much of the stories of digital transports and digital cables making huge audible differences, well, because they did, measured as well as audibly.
The second mode of the CS8416 is needed for > 96kHz sample rates and is one of those things that "seemed a good idea at the time". It is in effect capable of rejecting all source jitter that does not cause the receiver to unlock (I tested that using AP2 jitter generation feature).
The downside, it does very little to attenuate the noiselike jitter generated by the CS8416 VCO. And the levels of this noiselike jitter are directly linked to the sample rate, the lower the samplerate the more of this jitter. And there is a lot of it, enough to FUBAR the noise floor of most DAC Chip's.
So, you choose, massive sensitivity to source jitter or massive noise jitter. I am sure there is a lesson in this for the wisdom of using the CS8416 in high end audio products, to me the lesson was "use something else"...
Ciao T
At 20 bits, you are on the verge of dynamic range covering fly-farts-at-20-feet to untolerable pain. Really, what more could we need?
I changed the sample rate of the output from the optical output of mac mini. It was done by changing the setting on Auio Midi on the mac OS.
Audioengine D1 with CS8416 shows the same noise floor around 10kHz signal. The left graph is with 44.1kHz input and the right graph is with 96kHz input.
For Styleaudio Sapphire LE,I tried various sample rates: 44.1, 48, 88.2, 96. Only at 44.1, my Styleaudio Sapphire LE showes higher noise floor around the 10kHz tone. The left graph is with 44.1kHz input and the right graph is with 96kHz input.
I would return it if it didn't sound good.
The analog noise floor looks to be a little high. I see what looks like a couple of low level sidebands poking out of the noise floor at +/- 1.5KHz and 3KHz. Are these consistent or do they vary from FFT to FFT? Seems like an odd frequency for jitter. There is also a spike down around 400-500Hz, which I assume is random. But on the positive side, it looks to be free of low frequency noise (e.g. power supply) compared to the Mac Mini and Audioengine D1. And I don't see a fourth harmonic.
The measurements aren't great, but aren't terribly alarming either. If it sounds good enough, it is good enough.
Also, there seems to be a 34 KHz noise source in your playback or measurement system that is appearing to varying degrees in the different measurements.
The pictures were with FFT 65536 on ARTA. With FFT 4096, I can see the +/- 1.5, 3, 6 kHz sidebands are fluctuating.
When I connect Styleaudio Sapphire or Audioengine D1 (both have CS8416), there is 34kHz steady tone.
When I connect MuxLab 5.1ch to 2ch DAC, there is 38kHz steady tone.
"When I connect Styleaudio Sapphire or Audioengine D1 (both have CS8416), there is 34kHz steady tone
"When I connect MuxLab 5.1ch to 2ch DAC, there is 38kHz steady tone."
I am speculating what the problem might be, but I do have one question.... Is there also a "10 kHz steady tone" in your measurements?
10kHz tone is the test tone I used. I played 10kHz 0dBFS track on Pierre Verany Digital Test CD.When there is -4.0 dBFS on the screen capture, it means that the 10kHz sinusoidal tone was fed to A/D converter at the level of -4.0dBFS.
Edits: 05/17/15
"10kHz tone is the test tone I used. I played 10kHz 0dBFS track on Pierre Verany Digital Test CD."
It looks like the 34 kHz "steady tone" you were getting is the first alias (frequency reflection) of the 10 kHz tone about 22 kHz (half of 44.1 kHz sample rate).... The 38 kHz "steady tone" is the first alias of the 10 kHz tone about 24 kHz (half of 48 kHz sample rate).
(You might see other steady tones, at 54 kHz and 78 kHz if you're sampling at 44.1 kHz. And at 58 kHz and 86 kHz if you're sampling at 48 kHz.)
This suggests that no filtering is taking place. None whatsoever. Digital or analog. Most DACs have filtering to block out alias artifacts above half the sample rate frequency. Some non-oversampling DACs that also don't use any analog filtering would have these very same "steady tones" present.
So if your DAC is a non-oversampling DAC that doesn't use any other filtering, these "steady tones" you're getting might actually be normal.
The +/- 1.5kHz, 3kHz, and 6kHz side bands are consistently reproduced from Mac mini optical out and Styleaudio Sapphire analogue out. When I measure with different measuring equipment, I still see those sidebands.If I change the source equipment producing the optical out, I have different high frequency sidebands (e.g. +/- 4kHz, etc), but again, Styleaudio Sapphire has higher level of high frequency sidebands than other DACs I have.
In the included pictures, with 3kHz to 17kHz window, the source is Mac mini's optical output.
Edits: 05/17/15 05/17/15 05/17/15
I read the literature page (link).... The verbiage is confusing, I'm not sure whether the DAC upsamples to 24/192 or uses "8x oversampling".... It mentions both, but can't use both, unless it's a selectable option.If the DAC does upsample to 24/192, the issue is likely with the output clock. Either defective or poorly designed/implemented. If the DAC does NOT upsample to 24/192, I can picture a DAC having jitter issues if the PLL isn't robust enough to handle signals from a PC (as opposed to a disc transport). Or if no PLL is used at all.
I would try other digital signal sources (transports rather than PCs) before jumping to any conclusions.
Edits: 05/17/15
I think Styleaudio Sapphire does not upsample. 24/192 is just the DAC's capability in the DAC's catalogue. The 8X oversampling is done inside the DAC chip.
It uses Cirrus Logic CS8416 (no ASRC function there) digital interface chip. It is supposed to attenuate some high frequency jitter.
Audioengine D1, which was one of the DACs for comparison, has the same CS8416.
I thought of the possibility is that Styleaudio Sapphire's PLL is optimised for good quality source, while Audioengine D1 and MuxLab 5.1ch to 2ch DAC are optimised for both good and poor quality sources. But this should be very wrong. Styleaudio Sapphire is marketed for computer audiophiles. The product was designed with jittery computers in mind.
There could be many other reasons than just the DAC itself from the power presented to it to cabling and so on. Do you use a nice linear external PS or did it just come with a (switching) wall wart? Their site says 50 hours break in too, I don't know how old it is.
It does have a transformer on the coaxial digital in which is nice can you try that input? Sorry it isn't doing well for you. Does it sound worse than the other DAC that measures lower or are you just comparing measurements?
E
T
Now I am just comparing the measurement.
They recommend 200 hours of brake in. I am not sure whether such crude measurement (my measurement equipment is a $149, far from what JA at Stereophile uses.) result will change with burn in.
Now on easier to see 4k - 16kHz window.Far less +/- 3kHz and +/- 1.5kHZ jitter sidebands from this DAC for TV markt (MuxLab 500081)
Edits: 05/16/15
Now on easier to see 4k - 16k window.See the +/- 3kHz and +/- 1.5kHZ jitter side bands from my Styleaudio Sapphire LE.
Edits: 05/16/15
The spectrum from this $90 TV DAC is much cleaner than that from the over $400 high end DAC.
Note that there are also other spurious tones such as 400Hz tone
Again, +/- 3 kHZ high frequency jitter component is much lower level than from Styleaudio Sapphire LE
This $90 DAC shows less high frequency jitter than the $400 DAC.
Far less jitter.
The optical out and the analogue out share the same socket on mac mini.
plenty of high frequency jitter
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: