IFi Nano (and Micro)-GAME CHANGERS

WestGrooving, can you be more specific on your test and which format you used with which playback format?
Ex. Source recorded is WAV 24bit 96khz => Nano via PCM
or=>Nano resampled as DSD
Or DSD256 => Nano DSD native

Thanks

 

Source is old vinyl. Recorded to Korg MR-1000 in WAV 24/96 & then cue up the vinyl again and record in DSD X 2. Before I had the Nano, playback was done on the Korg unit (it plays both PCM & DSD natively) and I thought the DSD recording sounded a bit more lively (ambience) than PCM. Now playing those files back in JRiver to the Nano (in their respective formats), still think DSD sounds more ambient. Playing back the PCM sampled to DSD X 4, the PCM has that DSD ambience, but, IMO, there's a bit more texture in the DSD X2 played natively. So to me, DSD played natively sounds most appealing, then PCM upsampled to hi-rate DSD and finally PCM played as PCM.

Also, with this Nano and PCM recorded files, I've found the most appealing sound to me is taking the PCM file and converting it to a DSD X 2 file (my software is limited to converting to DSD X 2) and then playing it natively in DSD. Before the Nano, I honestly couldn't tell much difference using my laptop's decent sound card. So, yeah, for me this Nano has been a digital game-changer.

 

Yes, the mathematics first needed to be worked out, and then it was down to the software/hardware designers to dream up efficient structures to implement the required actions, minimizing things like noise build-up, latency, multiplications, etc., etc. The hope was that as more people attacked the problem, and published their findings, the knowledge base would grow, and improved architectures would come to the fore (which is why PCM signal processing is so ubiquitous).

I have tried to find some publications on the subject, but they all seem to be behind pay-walls (AES, etc.) or are now dead links. However, I did find this paper which, hopefully, will give a little insight to the possibilities:

http://dafx04.na.infn.it/WebProc/Proc/P_372.pdf

Sony's Oxford group did develop working DSD processing cards which could handle mixing, EQ, dynamics, flying faders, etc. They were stuffed full of Xilinx FPGAs, as I recall. You can see the mix card in this article:

http://www.digitalprosound.com/2003/03_mar/news/sonic_aes.htm

Sadly, the image is tiny, so it is hard to see what is on it, beyond the Xilinx chips. If you had enough DSP horsepower, you could do the same in software, or go the whole hog and make custom ICs.

When exactly did Microsoft give away IBM PCs in order to get MS-DOS established?

As I wrote earlier, by the time of DSD/SACD Sony was essentially out of the digital audio workstation business. Sonic Solutions, SADiE, ProTools, etc. had all come to the fore. Yet, Sony and Philips did work with the various companies to help develop DSD processing tools. The first stereo editing systems were made by Sonic. Sony and Philips had teams of people in the US, Europe and Japan who were constantly traveling to studios with loan equipment for engineers/producers/artists/labels to try DSD for themselves. Sony Oxford developed mixer, EQ, and dynamics processing. However, these small companies could not react in the way giants like Philips and Sony can, and, in the end, Sony set up the Sonoma (multi-channel) development in order to have a reference/fall back option. Sonoma was never intended to be a product, since the goal was for DSD processing to simply become part of the tools studios were already using (ProTools, etc.).

Between them, Philips and Sony had many millions of dollars invested in recording systems (all in custom flight-cases), including huge numbers of dCS and Meitner ADCs and DACs. These systems were loaned (with/without support from a Philips or Sony team member) to all the biggest names in the industry. It was not financially feasible to give away such systems to every recording or mastering studio (even the most famous). They were also prototype in nature, so needed specialist support.

Sony did NOT lock the files into a physical format, that was one (of a list) of requirements set out by the music industry. There would have been ZERO label support if SACD (or DVD-A) had been an open standard. The encryption schemes in SACD remain intact, so if the music labels had gotten on-board they would have had a secure physical format now. Irony of ironies, SACD had to compete with other divisions within Sony like Playstation for funding, and it was a failure in Playstation that finally allowed SACDs to be ripped, but it is Playstation that is hacked, not SACD (which is why there are no PC-based SACD rippers).

Once you do any processing in Pyramix the signal is converted to 8 Fs PCM (aka DXD), so any standard PCM DAW (e.g., ProTools) working at the same rate would achieve the same results (ignoring differences in implementations of fade profiles, filters, etc.).

 

Thanks Westgrooving for the thorough walk thru.
Could it be that the Nano and even the Korg are not very resolving in playback of PCM, whereas the Nano excels at DSD native? Guess what I'm trying to get at is that other PCM native playing DAC's may excel better at this format, thus the upsampling or conversion of PCM and played DSD is just a gimmick, I suspect. But it is at the moment your best sounding playback for PCM it sounds like.

 

That's a good question regarding DSD just being implemented better in the Nano. With the Nano, I recall reading the DAC was cherry picked as Burr Brown Japan's highly regarded best chip. The clock is military/telecom industry grade using AMR jitter-free technology. I assume both PCM and DSD are built into the dac chip. However, just listening to Redbook files played natively on this DAC, it's easily the best 16/44 playback I've ever heard (so clear and focused). I once had a very old Denon DCD 620 20bit CD player that sounded better than any other similarly priced cd player I had back in the day, but, honestly, this Nano's standard redbook playback just sounds in a different league.

I've never been wowed with the playback of the Korg unit... it's ok... The nano has really impressed me though.

I'm curious to hear from those who have the Nano plus another top performing PCM DAC, if redbook file converted to a DSD128 file and then played back natively DSD (say via JRiver Media Center) sounds much more life-like and engaging than the redbook file played on the other PCM DAC.

 

Thanks for the PDF link. I have a much better idea of what is possible now for DSD processing.

With some luck we may finally get some consumer level software that can process DSD. It will take a while, and a lot of work. But with the increasing interest in DSD playback using computers the software development of DSD effects may finally get the push needed to make it happen.

 

I heard this obesrevation about prefering upsampling to DSD with the Nano from other sources as well, although with the Micro DSD most seem to prefer PCM played back as PCM.

I personally don't think that these observations are related to the different playback formats as such, as these are based on PCM converted to DSD (and not native DSD). Most likely the difference in playback experienced is casued by difference in design of DSD and PCM architecture in the different DAC's:

Nano: DSD architecture > PCM architecture
Micro: PCM architecture > DSD architecture

Please note that above holds only for PCM converted to DSD and then played back by the two DACS and not for playback of native DSD.

 

Interesting and more incentive for me to get the micro DSD. Would definitely love to hear better PCM.

 

Check out Computer Audiophile and Headfi forums. They have long threads with user experiences as well.

 

I am finding both PCM and DSD sound good on the Micro.

 

But do you prefer PCM played back as PCM or converted to DSD (by Jriver or other software) and played back as DSD? Most Nano users apperently prefer the latter whereas most Micro users prefer PCM played back as PCM.

 

I don't have a Nano and in the Micro it does native DSD.

 

From what I can find online, it seems like both units use the same DAC chip (the Burr Brown DSD1793):

http://www.ti.com/lit/ds/symlink/dsd1793.pdf

The Nano uses a single chip for stereo, while the Micro uses two chips in a dual-differential configuration (you can see how a single chip can be used for monaural use on Page 33 of the above PDF, and this delivers a better dynamic range). We know for sure that the Micro uses the DSD1793 from this image:

http://audio-gizmo.com/wp-content/uploads/2014/07/idsd_micro_teardown-9.jpg

which is part of this: http://audio-gizmo.com/ifi-idsd-micro-tear/

I have not been able to find a similar image for the Nano (if anyone knows of one...).

From the PDF file, it should be clear that both the Nano and Micro are NOT using the DSD1793 in its specified modes. Whether they have uncovered hidden modes, or worked with Burr Brown, I have no idea, but they are definitely doing something non-standard in order to support higher than 24/192 and DSD64 rates with this chip. It is also impossible to say whether they use the same or different code in both devices, so I am unable to comment on any difference in architectures.

A few minor quibbles, I do not know why they refer to 'dual-core' for the Micro when they simply mean dual DAC, i.e., both DACs in a stereo converter being used for a single channel. NOTE: the chip itself is already differential by design, hence my use of the term 'dual-differential' when using a single chip per channel.

They also use the phrase 'All native decoding, no internal hardware conversion' for all forms of PCM and DSD, and yet the 1793 is a 'hybrid' converter, as shown on Page 43 of the PDF. Unlike some chips which handle both PCM and DSD, it does not seem to have a direct path for DSD signals (it does not show one, in any case), so I assume that PCM and DSD are processed as described (6 MSBs in standard PCM converter, remaining bits in 5-level sigma-delta modulator). This means neither signal is processed natively!

This last point is not something to get worked up about, provided the units sound good. I just get the impression that manufacturers are playing the 'jargon game'. Some make a big thing about not using sample rate converters, only for there to be one in the front-end of the DAC chip used. So, while it may be true to say that the designers did not include one, it is a little disingenuous. I feel the same about the use of native processing with the iFi. Yes, it may not contain an upsampler, etc., but the DAC chip certainly does, and format conversion too.

Back to the reviewers: how do the Nano and Micro compare? Cost/benefit analysis? (Which represents the better deal?) Compared to Benchmark, etc., how much do you feel you are getting for your money? Which filter mode are you using, and why?

 

Micro: So far it is at least on par with myBenchmark DAC1 Pre but there is much break-in to go. I think this is a terrific value. The Nano is very good based on several listens but the Micro is much better in my opinion. For filters I have minimum phase engaged which I like better that "bitperfect" I also like the 3D and XBass filters in "on" position for both the Maggies and the PS500 Grados.

Darren, the importer, told me that BB did some custom work for them on the chip but I don't know the technical specifics.

 

Thanks for the speedy response. I tend to go minimum phase too. I guess I had better go read up on what 3D and XBass do!

I imagine the specifics are proprietary, but it makes sense that they worked with BB. At first I could not believe that the DSD1793 was the chip used, given the published data sheet, but the photos proved it, hence they must have done something non-standard.

 

IMO if you need a portable DAC the Micro seems to be the better value (and power supply). However if this is going in a home kit, I would hold out for the Mini as the chassis size has more room to develop and implement a better analog section that should give it a level of refinement over the travel sized DACs.

 

The Mini will have a much better power supply as well.

 

But that is expected to be twice the price again?

 

The Nano and the Micro are cut from the same cloth. However...

Micro is superior sonically. More spacious, more ambient detail.

For some one needing a DAC for a desktop, for casual use, the Nano is off the charts good, and better than any other portable DAC it competes against under $300. Way better than the Dragonfly et al. I love the volume control and it exudes quality.

IMO the Micro is good enough to use in a very good dedicated system, well away from a desktop or headphone use.

I am using the recommended filter, the BitPerfect. Have no interest in the XBass or 3D.

 

$999

 

As far as when to buy a DAC and what is coming down the road, you'll never get ahead. The market and innovation is moving fast. You just need to stop and buy what sounds the best today for you. Not tomorrow, because if you play that game you'll never buy and never be satisfied.

 

Yes, but I think you are going to get a lot for that money. IMO iFi is an overachiever and give a lot of bang for the buck, so I'm betting the Mini is going to punch above its weight class and start taking jabs at the higher end offerings from other companies (perhaps even their own parent company AMR).

 

Using JRiver and foobar's foo_dsd_asio, DSD64/128 now can be played at DSD256 with the Nano. Works great and I like what I'm hearing.
Still amazed at the sound quality produced for what I feel is a reasonable price.

I did order up a iDSD micro to hear the improvements to both DSD & PCM. I heard the PCM playback is a major huge improvement.
The JRiver/foobar foo_dsd_asio combo can also support up to DSD512... which the iDSD Micro can do. Should be interesting.
Still, the native PCM is where I'll be focusing on... will there be texture/detail in those acoustic guitar string plucks versus just a flat sound(?)

 

Westgrooving, are you saying DSD64/128 sounds better upsampled to DSD256 with foobar or JRiver into the Nano? As opposed to the given sampling rate of DSD64/128 played back at it's given rate?
I've yet to notice any advantage with PCM in any similar upsampling with the Hugo. It's just a much bigger data feed. In fact, I often find little to no difference in Hi Rez to Redbook when under the same mastering.

 

Also I should note, about JRiver, is that any DSD conversions are resampled to PCM then back to DSD, thus lossy. Didn't know if you knew that? Kind of defeats the purpose, no?