|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
12.218.118.2
In Reply to: RE: Digital Filtering posted by audioengr on August 17, 2014 at 11:40:09
John Swenson shares similar thoughts about most digital filtering. Search for his posts under his name / ID using the keyword 'filter' on:
http://forums.slimdevices.com
http://www.computeraudiophile.com
http://bottlehead.com
Short version... NOS DACs produce a more alive sounding presentation, but will generally be 'dirty' in the higher-frequencies, something who's audibility will be system and person-dependent. DACs with industry-standard filters will be cleaner in the high-freqs, but not have that alive-ness.
His assessment is that this is due to the way digital filters have generally been implemented, particularly by DAC chip designers. Using a DAC chip where one can turn off or just not use the chip designer's digital filtering and instead use a simpler filter tailored to the DAC as can be implemented in computer playback software using digital upsampling /filtering utilities like Sox can provide the best of both worlds.
A useful 'unexpected' find he's recently related is that how DAC chips implement their filtering varies by sample rate, where they use less filtering at higher sampling rates... and that upsampling can improve the sound you get by minimizing the effect of the built-in filters.
Well worth your time to search out and read his thoughts.
Greg in Mississippi
Everything matters!
Follow Ups:
"Short version... NOS DACs produce a more alive sounding presentation, but will generally be 'dirty' in the higher-frequencies, something who's audibility will be system and person-dependent. DACs with industry-standard filters will be cleaner in the high-freqs, but not have that alive-ness."
I think the best implementation for digital filtering is "time resolute" filtering.... Not as flat in FR like standard filtering, but less "HF beat modulation" (what you call 'dirty') than non-OS.
Early Wadia "decoding computers" used "time resolute" digital filtering. As well as some inexpensive players/changers back in the 1990s. (Most notably the JVC XL-Z1050 DAC and Philips CDC-935 changer.)
"A useful 'unexpected' find he's recently related is that how DAC chips implement their filtering varies by sample rate, where they use less filtering at higher sampling rates..."
Agreed.... I believe sample rates of at least 60 kHz don't even need digital filtering. The passband and stopband requirements would be far enough apart to implement with purely analog filtering.
"and that upsampling can improve the sound you get by minimizing the effect of the built-in filters."
Depends on the context of "upsampling".... In the general sense, it's the means to implement the digital filter. But if it's upsampling CD to 24/192, I believe the effect is demonstrably worse than non-OS or synchronous upsampling.
I wonder what's happening in my implementation?
I have tried to elicit more information about the transport and dac I'm using, but to no avail. All I know is this:
The transport upsamples to 96kHz
The transport communicates with the dac via I2S
The dac, it's suggested, reclocks (is that needed with I2S?) and plugs straight into the 4x 1704U-K chips. Presumably, with the I2S delivery at 96kHz, not much digital filtering is needed? Output stage is FET-based, balanced, class A, no negative feedback and independently powered (as is the digital stage and transport).
There is a digital filter in the dac but the manufacturers have anonymised it. (Normally the DF1704 is used with these chips - is it something else?).
The resulting sound is tremendously engaging and, above all, noticeably 'muscular' and resolute. It feels like the sound is sculpted in space, with no loss of tonal shading or timbre subtlety. The icing on the cake is that I consider the overall cost for the sound I have, to be, frankly cheap.
I wish more on here could hear it.
big j.
"... only a very few individuals understand as yet that personal salvation is a contradiction in terms."
"There is a digital filter in the dac but the manufacturers have anonymised it."
The original poster stated the DAC does NOT have a digital filter.
"I wonder what's happening in my implementation?
"I have tried to elicit more information about the transport and dac I'm using, but to no avail. All I know is this:
"The transport upsamples to 96kHz"
The transport upsamples to 96 kHz?
This is unusual for a transport.... Most transports merely send the output at the native rate. Any upsampling would take place at the DAC.
"The transport communicates with the dac via I2S"
It's supposedly a desirable link option.... But otherwise don't have much opinion of that.
"The dac, it's suggested, reclocks (is that needed with I2S?) and plugs straight into the 4x 1704U-K chips."
An upsampler is supposed to have a *very* precise output clock. (Because if it doesn't, the imprecise timing of 24/96 upsampling would result in larger noise artifacts introduced into the signal.) If it does, reclocking wouldn't gain anything.....
"Presumably, with the I2S delivery at 96kHz, not much digital filtering is needed?"
96 kHz upsampling would already filter the signal. So no additional digital filtering would be necessary. Regardless of the link used between transport and DAC. (Although at the higher rate, the bandwidth of the link becomes more critical.)
"Output stage is FET-based, balanced, class A, no negative feedback and independently powered (as is the digital stage and transport)."
If the DAC has a nice analog output stage, it deserves a cleaner conversion in front of it.........
"There is a digital filter in the dac but the manufacturers have anonymised it. (Normally the DF1704 is used with these chips - is it something else?)."
As stated earlier, if the transport upsamples to 24/96, a digital filter wouldn't be needed in the DAC.... The DAC sees a 96 kHz input, the only filtering is to attain full stopband at 48 kHz, easily attainable with analog post filtering. (The upsampling at the transport would provide full stopband at 20 to 22 kHz for the CD playback. Although the output is 96 kHz sample rate, there should be no information above 22 kHz in that signal.)
"The resulting sound is tremendously engaging and, above all, noticeably 'muscular' and resolute. It feels like the sound is sculpted in space, with no loss of tonal shading or timbre subtlety. The icing on the cake is that I consider the overall cost for the sound I have, to be, frankly cheap."
If you like how it sounds, what I stated above may not matter at all.... The only thing I will say is you might prefer a setup that does not upsample to 96 kHz......
(I might criticize a design technically, but if a listener likes how it sounds, what I say doesn't mean a thing.)
"I wish more on here could hear it."
If you're in the Phoenix, AZ area, I'd be happy to..... [-;
But I'll be honest, I've never heard a CD source that upsampled to 96 kHz that I liked.
Neither have I. I tried it out of curiosity and simply haven't wanted to revert to the SPDIF interface. I2S it is, 96kHz or not. I surprised myself.
big j.
"... only a very few individuals understand as yet that personal salvation is a contradiction in terms."
Me too - as in wanting to be one of those who could hear it.
Give me rhythm or give me death!
Edits: 08/20/14
All that for $85!
Give me rhythm or give me death!
Great exchange of info guys. Thanks! for sharing.
This echoes my experience. No digital filter usually sounds better than a pedestrian digital filter implementation, while an good digital filter implementation can sound better than no digital filter.
_
Ken Newton
Edits: 08/17/14 08/17/14
Where would you consider the BB DF1704 in this scheme of things? Pedestrian or good?
Objectively, I would consider the DF1704 to be rather pedestrian. If I correctly recall, it is an typical multi-stage half-band hardware based implementation, which means it is only 6dB down at the Nyquist frequency, and chosen to minimize the required hardware resources (less chip cost). Also, the stop-band attenuation of the DF1704 is good, but not great. However, there are more implementation factors which matter to sound quality than simply those two.That said, I can't speak to the subjective sound quality of the DF1704, as I'm uncertain whether I've auditioned any DAC utilizing it .
_
Ken Newton
Edits: 08/18/14
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: