iFi Audio Pro iDSD discussion thread

[sdolezalek]

Different DAC's respond differently to upsampling.  Some, like the iFi micro and the T+A DAC8 DSD respond extremely well to pretty much everything upsampled to DSD512.  Others, like some of the Chord DACs are more specifically built to sound well with PCM input.  
 
I don't find that the same files that I have in native DSD256 sound as good when I upsample the Tidal version of that file in 16/44 to DSD256 (which is the highest I can use right now).  It's close, but there is an audible difference.
 
But there is also an audible difference between a file I only have in 16/44 and listening to that same file upsampled in HQPlayer to DSD256.  
 
Those owning the T+A DAC and able to upsample to DSD512 report that it is yet another significant step up from using DSD256 (even with that DAC).  
 
My hope is that AMR/iFi saw some of these new DACs coming out and given that iFi was already focused on the benefits of upsampling, decided to re-engineer the Pro to benefit from some of what others are hearing from the competing (and higher-priced) DSD512 capable DACs already on the market. 

 

[youurayy]

The term "upsampling" here is quite a simplification -- the PCM data is literally rendered into DSD the same way it would be rendered by a PCM DAC.
 
The micro iDSD's PCM DAC sounds very nice btw., I'm not entirely sure why I prefer the DSD conversion, but I do.
 
Here I found some background on the whole thing (I have yet to read it):
 
http://www.emmlabs.com/pdf/papers/DerkSigmaDelta.pdf
 

The driving forces for the use of this technology were pure technical: in the CD era, demands for distortion levels were becoming more stringent, and it proved virtually impossible to create low distortion devices with many (16) bits. Contrary to that, it was much easier to create low-distortion converters using a digital format of 1 bit, which were running at very high sample rates such as 64 or 128 times 44.1 kHz. The conversion of this high speed, 1-bit format to 44.1 kHz/16 CD format can easily be accomplished in the digital domain using filtering and signal processing, which does not introduce any non-linear distortion. This technique has been highly successful, and the so-called ‘oversampling’ and /or ‘bitstream’ technology has dramatically increased the performance of the CD-players in the nineties. In fact, those CD-players were all generating their own DSD internally from the CD source; this DSD would then be fed into a high quality, 1-bit DAC.

 
EDIT: also this is quite revealing:
http://bitperfectsound.blogspot.com/2014/03/real-time.html
 

With a multi-bit output - such as PCM itself - the DAC has to create an output voltage which has a magnitude determined by the bit pattern.  The more bits, the more different possible output levels there are.  With 1-bit, there are only two levels - encoded as 1 and 0 - and from an electrical perspective, all a DAC has to do is switch its output between two fixed voltage sources represented by those numbers.  These voltage sources can be, for example +1V and -1V, and can be controlled and regulated with fantastic precision, and with extremely low noise.  The job of the DAC is then simply to switch the output signal line between one voltage source and the other.  This is something you don’t need a chip to do, and is furthermore something you can employ a lifetime of audio electronics circuit design experience to realizing in the best possible manner.

 

[sdolezalek]

Careful, you are standing on the edge of a rabbit's hole of information in which you could be lost for years...
 
I have found the easiest way to think about it in my own mind is that both processes (PCM and DSD) require further processes (like dithering and filtering) to get from the digital data they represent to the analog music you hear.  The further that the effects of those conversion processes are moved out of the audible range, the less they affect what we hear.  The common misconception is that hi-res or upsampling is about allowing us to hear sounds above 20kHz or to hear steeper transients. Mostly, it is not.  It is simply to prevent other stuff we are doing to the data stream to have any "bad" effects outside the range of our hearing.  
 
Unfortunately, as good as that may be for our ears, it can wreak havoc on equipment that doesn't like dealing with signals that far out of band, so there is always a balancing consideration of whether your equipment can handle the data.  Fixing the filtering process in the DAC has less impact when your pre-amp, amplifier or speaker inserts a 20kHz brick wall filter that reintroduces all the noise you just tried to move to 40kHz. 

 

[sdolezalek]

Look out, here comes the competition (wonder which we will actually see first in the flesh, the new Oppo or the iFi Pro):

The Absolute Sound just revealed the new Oppo Sonica DAC in its latest Buyer's Guide:

"The new Oppo Sonica DAC improves upon the highly regarded audio performance of the Oppo HA-1 and BDP-95/105 players and presents an upgraded DAC chipset, the flagship ESS ES9038PRO SABRE DAC. The Sonica DAC combines the convenience of modern network streaming and the hi-fi sound quality of the traditional audiophile DAC. The Sonica app for iOS and Android allows you to conveniently stream music from your smartphone or tablet. The Sonica DAC supports high-resolution PCM and DSD audio via its USB, coaxial, and optical inputs. Additionally, Sonica DAC is a high-resolution audio player and can decode files directly from a connected USB drive, NAS drive, or network shares. Available in Fall 2016. Price: $799."

http://www.theabsolutesound.com/buyers_guides/35/

Not trying to suggest that it will be the sonic equal of the iFi Pro by any means, but given Oppo's history it will almost certainly do DSD256 and set a low price point for a high level of sonic quality.

 

[EVOLVIST]

Yeah, I don't see how the Oppo is in direct competition with the iDSD Pro. I mean, everyone and their dog does DSD these days, even Chord got in on the game, even though Rob Watts doesn't dig on DSD.

To me, this new Oppo DAC has all of the bells and whistles (and it probably sounds good), for someone who wants a DAC with a small footprint to stream from Tidal and the like.

The iDSD Pro, on the other hand, is a bid for the pro audio market, as a jack of all trades that simultaneously has one foot in the present, while honoring classic designs of the past, which gives it a totally unique class and phylum. That the iDSD Pro also caters to both the solid state and the tube crowd is just another peek to the product.

I guess I'm a sense it's all competition, because every audio team is looking to get out money, but in the comparison between these two, I think you're looking at two different buying crowds who are looking for two different DAC applications.

 

[Maelob]

As far as I knowThere is no IDSD Pro yet. We are still waiting....

 

[occamsrazor]

The Oppo Sonica seems pretty interesting to me, especially as it will use the new 9038 chip. My main question will be what kind of network streaming it will support....

 

[Esprit]

As far as I know There is no IDSD Pro yet. We are still waiting....

This is the only real fact...

 

[technobear]

As far as I know There is no IDSD Pro yet. We are still waiting....

This is the only real fact...

See: http://www.head-fi.org/t/702376/ifi-audio-pro-desktop-line-discussion-thread/1350#post_12783079

Not long now.

Shame iFi won't be at Headroom/Indulgence in London this October but I guess doing that and RMAF would be a stretch.

 

[Esprit]

Nowadays there is nothing and we have nothing: from 1/26/14 (1st post of the thread) we have read a lot of dates...

 

[Jimi Zine]

I'd question the use of Sabre Dac chips for DSD output, I think the best implementations for 'direct native DSD' processing via 

TI/Burr Brown chipsets and chipless designs, http://www.positive-feedback.com/Issue78/dsd.htm

The Oppo Sonica seems pretty interesting to me, especially as it will use the new 9038 chip. My main question will be what kind of network streaming it will support....

 

[Jimi Zine]

The Oppo Sonica seems pretty interesting to me, especially as it will use the new 9038 chip. My main question will be what kind of network streaming it will support....

I'd question the use of Sabre Dac chips for DSD output, I think the best implementations for 'direct native DSD' processing via 

TI/Burr Brown chipsets and chipless designs, http://www.positive-feedback.com/Issue78/dsd.htm

 

[iFi audio]

Hey everyone!
 
We know a lot of you have been anxious to get more impressions of the Pro iCAN as well as more details of the upcoming Pro iDSD. Not to fear! We have some exciting updates that we're excited to share with all of you very soon! Be sure to be on the look out for that.

 

[Esprit]

iDSD Pro price (in Euro...)? 

 

[MLGrado]

  I'd question the use of Sabre Dac chips for DSD output, I think the best implementations for 'direct native DSD' processing via 

TI/Burr Brown chipsets and chipless designs, http://www.positive-feedback.com/Issue78/dsd.htm

 
 
Chipless designs and the Burr Brown implementation are the same thing.  Just, one is done via discrete parts, the other is on chip Process is the same though.  
 
Burr Brown implementation is outstanding, maybe the best?  I haven't had the chance to hear some of the newer 1 bit discrete designs, such as Lampizator, the T+A, Holo Audio, and Miska's open source dac.  
 
The reason the BB is so good is the outstanding time domain performance.  8 bit delay line, as opposed to others who are using a much, much longer delay line.  The increase in taps means better frequency domain performance, at the expense of time domain.  A workaround is to upsample to DSD256 or DSD512 on these DACs with lots of taps... time domain performance will increase as the speed of the signal increases.  But there is no free lunch.  This creates other issues, requires signal DSP, etc, which is probably NOT what you really want to do with DSD.  (Many would disagree with that, though) Anyway, the 8 bit delay line, and unequally weighted grouping of the 64 bitswitches in all Burr Brown Native DSD dacs is an excellent implementation, and two decades after the venerable DSD1700 chip was introduced, is still a standard bearer.  Of course, this is one of the primary reasons iFi uses this chip rather than ESS.  
 
 
For those of you who aren't really familiar with what the ESS chipset does with DSD, here is my best stab at it.  ESS actually keeps what really happens in the chip very, very close to the vest, protecting their intellectual property fiercely.  Not even the confidential data sheet says very much!  (Actually, data sheets for ANY dac out there don't say a whole lot, and often what they do say isn't exactly what is going on...) Anyway, again, here is my best stab at it based on what I have read and been told.  
 
 
ESS does not convert DSD 'natively'.  The closest thing to native conversion we have is applying a FIR filter in the analog domain, like I described above.  That is it.  
 
ESS doesn't filter DSD in the analog domain.  When a 1- bit bitstream hits the ESS chip, it gets routed to an IIR filter.  You can select the cutoff frequency for this filter (50khz, 60khz, 70khz,).  The OUTPUT of this filter is no longer 1 bit DSD.  The samples that come out are now multi-bit.  (No this is NOT mult-bit delta sigma) Now, the filtering that is uses is a slow rolloff that extends up to around 1 mhz, if my memory is correct.  So the sample rate is not decimated to any of the common PCM rates.  Actually, it is probably a really good filter with excellent time domain characteristics.  Essentially it is the digital version of the analog filter used in the Burr Brown chipset.  The filter, and the fact that things are now multi-bit, is NOT the problem here.  The problem is, when this is done in the analog domain, you are DONE!  The DSD signal has been converted to analog.  The problem is, with the ESS, after this filtering, you are JUST GETTING STARTED.  
 
While the DSD signal is now in the intermediate format (I am guessing it also goes though a noise-shaper, so its a lower bit high sample rate signal) 32 bit volume control may or may not be applied, and the ASRC jitter eliminator is applied.  Then after that, the signal is converted back to delta sigma in the Hyperstream multi-stream modulator.  And then that is likely analog filtered, as well, for the final output.  
 
So, what I am getting at, is the ESS chip does a LOT more with the DSD signal.  Lots of DSP and conversions to get to an end result.  The other way is much more simple.  The signal is already in a state where a single analog filtering is all that is required for conversion.  
 
 
NOW don't get me wrong here... the ESS way apparently sound very, very good!  Which way is technically better doesn't really matter.. all that matters is what the listener is hearing.  If it sounds great, so what!  But, on paper, I just can't see the ESS way as being the better solution, and I can't see how you could call it native DSD conversion.  
 
 
Anyway.... sorry about the long post... I AM indeed looking forward to the iDSD Pro DAC.  Saving my pennies.  On my must buy list!