|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
73.233.216.42
In Reply to: RE: 3.5dB headroom in off the shelf DAC chips? posted by dave789 on January 31, 2017 at 06:03:21
Not sure what you mean by, "inter sample over even". Possibly, you're referring to how sustained high level signal data can drive digital interpolation filters to clip. The only commercial DAC chip vendor I can recall mentioning that they provide extra headroom for this occurance is Wolfson (now acquired by Cirrus, I believe).
_
Ken Newton
Follow Ups:
"inter sample over" / "even with"
Most converters clip whenever an inter-sample over occurs. - https://benchmarkmedia.com/blogs/application_notes/inside-the-dac2-part-2-digital-processing
If you think this is real and the "only" thing that makes DACs great then there is no other choice. You must buy a Benchmark.There are very, very few recordings that are even near 0db. Benchmark DACs are not considered very "musical or detailed" by those that have heard all the competition and more expensive machines. Benchmark does write good and extensive ad copy.....they got you. DACs are being released at an increasing rate and competition is super strong. They are writing for their life! Hey, the Oppo Sonica just got released today with a 9038 DAC chip in it for $800 with streaming and usb input....of course, it needs tweaking. "But please buy our Benchmark DAC....we are the only ones that know what we are doing".
There are hundreds of things inside a DAC that determine the final sound. Benchmark is not a tweaky company. Mostly nerd dudes. Hey, I like nerds. But artists make things sound good. Of course, you have to have enough nerdism to make it work.
Edits: 01/31/17
... if you listen to much modern pop/rock music on CD, most of it is constantly pushed to 0dB on top of all the compression and hard limiting used to get it as loud as possible, so how the convertor handles those peaks may be a criteria for some since there will be a lot of intersample peaks above 0dB. That said, I do think most digital filters handle peaks above 0dB now without clipping since it has gotten a lot of attention in the last decade, but that wasn't always the case.I know the PMD100 HDCD processor used something like a -2dB attenuation to avoid the problem, and that was back in the mid 90s. I remember old AA member Christine Tham did a little study about 12 or so years ago here and found mixed results, some Sony players tested OK for handling intersample peaks, but she found some others models did show the effects of digital clipping on the analog outputs. Probably still online someplace ... here it is
------------------------
Favorite album of recent times, beautiful and completely mesmerizing: The Revolutionary Army of the Infant Jesus - Beauty Will Save the World
Edits: 01/31/17 01/31/17
I am sure there are some. My Greateful Dead DVD-A surely has peaks above 0db. However, most all output stages in serious DACs can handle multiples of this (not so sure about portables with their 5 and 3 Volt power supplies). The modded Gustard with its plus and minus 15.5V power supply can do about 11VRMS before it clips.....not too many CDs with that amount of juice!....eh?
Ric, this is an issue involving the digital interpolation FIR filter and the scale of the PCM sample values which it attempts to calculate should be sent to the D/A unit, not the post D/A unit analog stage headroom.Some music tracks have their dynamic range highly compressed so that their average level can be boosted to maximally utilize the MSBs of the system's binary data dynamic range for the purpose of maximizing the perceived average loudness level. To see the problem this can present to a FIR interpolation filter I need to touch on how they function.
Digital FIR filters operated by a mathematical process known as convolution. This process is essentially a running 'sum-of-products' calculation. At each stage of the filter, two numbers are multiplied together, the changing sample value and a fixed coefficient value, producing a product. Then, all such products from all stages of the filter are summed to produce a single output sample. This process repeats to produce each sample output from the filter. So, a string of full scale (0dBFS) input sample values can result in output sample values calculated to be greater than system full scale value. The result is that the filter clips (truncates the value) of the output samples sent to the D/A unit, which then produces a clipped analog output signal.
One solution, aside from using proper mastering technique, is for the digital filter to reduce the scale of the input sample values so that no internal filter calculation can produce an output sample result greater than the system full scale value. Apparently, the required scale reduction factor has been determined to be -3.5dB.
_
Ken Newton
Edits: 01/31/17 01/31/17 01/31/17 01/31/17 01/31/17 01/31/17 01/31/17 01/31/17 01/31/17
I'm not sure how you are measuring the CDs, you would have to oversample it to see the "real" peaks, some ripping software will tell you if there are "overs". That is what inter-sample peak means, the samples on the CD are not the actual peak values except at low frequencies. At high frequencies, the sample point can be quite a ways off from the peak. The digital and analog reconstruction filters will of course restore the peaks if they have the headroom, though the problem is seldom with the analog section. For best sounding results, CD mastering engineers should set the peak level using an oversample filter to pick up the real peaks, but they often don't.
------------------------
Favorite album of recent times, beautiful and completely mesmerizing: The Revolutionary Army of the Infant Jesus - Beauty Will Save the World
Edits: 01/31/17
They are not the same. Yes, some highly compressed music (why would you listen to it anyway?) could indeed do what is said. However, does that mean it is really that audible and to whom and when? Did they show an example of a clipped signal? Did they tell you at what point on a certain musical cut it happens and then what happens when listening....when it does? Do you have to go to the hospital after hearing all the distortion from all the "other" DACs that don't do what theirs does? The quote below is from their article:"Most converters clip whenever an inter-sample over occurs. The DAC2 will not clip these overs. This is probably the single most important improvement in D/A technology in the past 10 years! This difference explains 90 to 100 % of the audible differences between the DAC1 and the DAC2. It also sets the DAC2 apart from almost all other D/A converters."
Wow! what a statement. So the fact that they changed DACs and everything else they hype really does practically nothing to the sound? Only this "new invention" that they thought of and are the only ones doing it...is meaningful? Forget output stages, damping, tweaking, etc. Only this thing is meaningful?.....You have got to be kidding!?!
I was just now listening to a just finished modded Gustard and the magic was not there. On purpose, I had not put two pieces of damping material on caps on both sides of the output board because I run the damping material over to the chassis struts and I wanted to be able to remove the board quickly to try a different low jitter clock. Well, I put back on those two pieces of damping material and practically started crying the sound was now so good. Oh, I am sorry, I lied....really I changed the digital filter settings internally so it won't clip anymore and that is why it now sounds so good....after all, this tweak is the most important improvement in DAC technology in the last 10 years.
And before I put that damping material back on I also put a small piece of damping material on the top of the clock and it made a significant audible improvement......well, I lied again...nothing happened as the only thing that makes a real improvement is to lower the digital gain by 3.5db.
They really do know how to write a tale! Commercial nerdistic ad copy at its best....he he. I like my subjective audio rantings better....at least they are what I experience with my ears and my soul. I believe (BE plus LIVE) what I say.
Yes, digital clipping could be audible on certain cuts. How audible and to whom is clearly not why the article was written. They are feeling the heat of competition. The more words you write that make your product sound invincible could result in keeping your sales high. There are DACs coming out every day and some are so good (and relatively inexpensive) they make you cry (modded Gustard Pro, for instance).
Edits: 01/31/17 01/31/17 01/31/17
I didn't read the article or have any interest in their DACs, and agree there are much more important design considerations than how a digital filter handles intersample peaks above 0dB, but in lieu of an industry standard that forces CDs to be mastered correctly, it would be nice if digital filters did handle at least mild overs, which many do.
Good luck with your Gustard mods. Looks like a nice package.
------------------------
Favorite album of recent times, beautiful and completely mesmerizing: The Revolutionary Army of the Infant Jesus - Beauty Will Save the World
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: