valenroy Posted March 21, 2019 Posted March 21, 2019 This is my first contribution and I have actually posted this on headfi.org which seems like the base platform that the iFi team is revolving around:- "Hi there iFi team, I recently got hold of this dac and here are some things that you guys can take a look at to improve on beta firmware 2.01. This would pertain to ethernet network streaming using DLNA (which I personally think is the best sounding and most convenient input option). It would seem that for the iDSD to play nicely on the DLNA network (via latest JRiver v24 or Audirvna), the iDSD needs to be connected to a 2.4ghz wireless network (in the same network as the ethernet connection) concurrently as well. And the easiest way to do this is to add a new device in the MUZO app. Both JRiver and Audirvana have a hard time locating the iDSD on the network if this step is skipped. Alternatively, the MUZO app, with the iDSD found in it, has to be running in the background with whatever Remote app that you are using to control the device (I'm using the JRiver and Audirvana iOS remote apps on my iPad Pro by the way). Even though I do not use wireless streaming at all, the iDSD tethers itself to it and there are currently some issues with that. Firstly, I've found that the iDSD adhoc network cannot be password protected, if I do that, it would immediately cause audio dropouts and stuttering, especially on high bitrate files like DSD64 for example. It would seem that the iDSD is trying to connect to the wireless network as a background process but fails to do so. I suspect that the intrusive wireless setup audio prompts (female voice), if setup/connection is unsuccessful, are causing interruptions in the playback chain, causing the dropouts and stuttering. Users who are experiencing these issues can try out this remedy and report back as a community service to everyone else. I would highly urge the iFi team to remove this dependency on the wireless network and let the ethernet connection be truly autonomous. Furthermore, the voice prompts could be a potential area to investigate interruptions during playback. It would be great to include the option of turning the Wifi option off completely to mitigate connection issues and possibly improve the sound quality further by negating spurious wireless signals and noise. As a suggestion, the WPS button could be used to enable the wifi again after you turn it off in the app for example, if not there won't be any other way to enable it again. Thanks for the time but I'm afraid that I have time only to feedback on these issues when there others abound."
valenroy Posted March 21, 2019 Posted March 21, 2019 Interesting fact of the day... if you think PS Audio’s Ted Smith’s internal 10xDSD (640fs) upsampling for all digital inputs is radical in the Directstream DAC ... think again... the Pro iDSD can upsample all digital inputs to 512fs (8xDSD) or 1024fs (16xDSD - similar approach to the EMM Labs MDAC2) with the Xilinx Spartan-6 FPGA , which is the same one used in the Directstream by the way - on the fly... best thing is that it’s user-defeatable to operate in NOS mode with provisions for digital filters such as their Bit-Perfect (filter-bypassed 1:1 upsampling - this seems to be missing in the latest beta v2.01 firmware), Gibbs Transient (iFi's own), Apodizing (preferred by AYRE Acoustics I believe?) and Transient-Aligned (favored by CHORD?) Filters. Even the professional Pyramix DSD workstation performs DSD operations in the DXD domain only whereas the digital engine of the iDSD operates at 2xDXD in real-time. In the Directstream deployment of the Spartan-6 FPGA, it is used for all the digital conversions and operations within the chip itself whereas the Pro iDSD makes use of the FPGA strictly for upsampling and digital filtering functions only. The input handling and digital-to-analog conversion is handled separately with the XMOS and 4xBurr-Brown DSD1793 chips respectively... Which is a better architecture in your opinion? Self-contained all-in-one chip or the divide & conquer approach by iFi?
valenroy Posted March 22, 2019 Posted March 22, 2019 Interesting fact of the day 2! The internal clocking system (GMT) of the Pro iDSD is of a higher precision than femto-grade clocks found in other high-end DACs!
Doggie Howser Posted March 22, 2019 Author Posted March 22, 2019 Interesting fact of the day... if you think PS Audio’s Ted Smith’s internal 10xDSD (640fs) upsampling for all digital inputs is radical in the Directstream DAC ... think again... the Pro iDSD can upsample all digital inputs to 512fs (8xDSD) or 1024fs (16xDSD - similar approach to the EMM Labs MDAC2) with the Xilinx Spartan-6 FPGA , which is the same one used in the Directstream by the way - on the fly... best thing is that it’s user-defeatable to operate in NOS mode with provisions for digital filters such as their Bit-Perfect (filter-bypassed 1:1 upsampling - this seems to be missing in the latest beta v2.01 firmware), Gibbs Transient (iFi's own), Apodizing (preferred by AYRE Acoustics I believe?) and Transient-Aligned (favored by CHORD?) Filters. Even the professional Pyramix DSD workstation performs DSD operations in the DXD domain only whereas the digital engine of the iDSD operates at 2xDXD in real-time. In the Directstream deployment of the Spartan-6 FPGA, it is used for all the digital conversions and operations within the chip itself whereas the Pro iDSD makes use of the FPGA strictly for upsampling and digital filtering functions only. The input handling and digital-to-analog conversion is handled separately with the XMOS and 4xBurr-Brown DSD1793 chips respectively... Which is a better architecture in your opinion? Self-contained all-in-one chip or the divide & conquer approach by iFi? valenroy i think you are confusing methodologies here. The DS DAC doesn't upsample to 20xDSD... it just samples ALL incoming signals at 20xDSD and then pieces together what the original signal is - sans clock. It is a pretty smart idea *to be fair done by Andreas/Ed before him - and it doesn't need to care what the incoming clock of the signal is doing.
valenroy Posted March 22, 2019 Posted March 22, 2019 Just an enthusiast who's excited for such a little technological marvel that's available at such an attractive price point.
valenroy Posted March 22, 2019 Posted March 22, 2019 valenroy i think you are confusing methodologies here. The DS DAC doesn't upsample to 20xDSD... it just samples ALL incoming signals at 20xDSD and then pieces together what the original signal is - sans clock. It is a pretty smart idea *to be fair done by Andreas/Ed before him - and it doesn't need to care what the incoming clock of the signal is doing. It's been a while doggie... thanks for the enlightenment... still holding onto your DS? I almost got the DS Jr. recently, fortunately or unfortunately, Steve ran out of stocks. I think that is what the iDSD is doing in a way too l right? Store all the data in a buffer, stripping away all the source clock signals, perform whatever digital processing, and then re-clock it with the GMT?
pcking1624705747 Posted March 22, 2019 Posted March 22, 2019 I’m a happy and satisfied owner of this DAC. ;D But have never tried its headphone out since day 1. :P
valenroy Posted March 22, 2019 Posted March 22, 2019 I’m a happy and satisfied owner of this DAC. ;D But have never tried its headphone out since day 1. :P Yeah neither did I, didn't buy it because of the headphone amp. Has the DAC been frustrating you? The playback is still glitchy as hell... can't even sit through a proper dsd album without some audio drop outs irritating the hell out of me.
pcking1624705747 Posted March 22, 2019 Posted March 22, 2019 Slight glitches yes but no major issues. I’m mostly using it via USB streaming from DIY Music Server. Yeah neither did I, didn't buy it because of the headphone amp. Has the DAC been frustrating you? The playback is still glitchy as hell... can't even sit through a proper dsd album without some audio drop outs irritating the hell out of me.
valenroy Posted March 22, 2019 Posted March 22, 2019 Slight glitches yes but no major issues. I’m mostly using it via USB streaming from DIY Music Server. I did actually try USB through my macbook as will that but track skips between dsd tracks produces static noise bursts... reminds me of the mytek when it was first launched. I don't know whether you have made the comparisons but the sound quality, to me at least, is best with the ethernet connection. Possibly due to the ethernet input's I2S signal versus the one from the USB I'm guessing.
Guest AndrewC Posted March 22, 2019 Posted March 22, 2019 Lets inject some reality into some of marketing hyperbole in this thread ;D Interesting fact of the day... if you think PS Audio’s Ted Smith’s internal 10xDSD (640fs) upsampling for all digital inputs is radical in the Directstream DAC ... think again... the Pro iDSD can upsample all digital inputs to 512fs (8xDSD) or 1024fs (16xDSD - similar approach to the EMM Labs MDAC2) with the Xilinx Spartan-6 FPGA , which is the same one used in the Directstream by the way - on the fly... Not all Upsampling is the same; each Vendor’s upsampling interpolation algorithm is going to be totally different, so even if Vendor A upsamples to 16xDSD, it doesn’t automatically mean it’s going to sound better than Vendor B who only upsamples to 10xDSD. Interesting fact of the day 2! The internal clocking system (GMT) of the Pro iDSD is of a higher precision than femto-grade clocks found in other high-end DACs! Whether or not the above is true (any 3rd party sources? or you’re just reading the marketing brochure?), the end-result seems to suggest it’s nowhere as good as other comparable systems, especially w.r.t clocking/jitter rejection. See for yourself :) https://www.stereophile.com/content/ifi-audio-pro-idsd-da-processorheadphone-amplifier-measurements … When I tested the iFi Pro iDSD for its rejection of word-clock jitter using undithered 16-bit J-Test data, the odd-order harmonics of the low-frequency, LSB-level squarewave were all at the correct levels (fig.21, sloping green line). However, many low-level spurious tones were present in the spectrum, and the spectral spike that represents the high-level tone at one-fourth the sample rate was slightly broadened at its base, suggesting the presence of some random low-frequency jitter. This behavior was identical when I repeated the test using the USB input (fig.22). … Compared to, say, the Chord Cutest (@24% cheaper than the iFi), without any such special “super” clock ;D https://www.stereophile.com/content/chord-electronics-qutest-da-processor-measurements … Tested for its rejection of word-clock jitter with 16-bit TosLink data, the Qutest turned in superb performance: all odd-order harmonics of the LSB-level, low-frequency squarewave were at the correct levels (fig.14, sloping green line). With 24-bit J-Test data (fig.15), no jitter-related sidebands were present. … ... I don't know whether you have made the comparisons but the sound quality, to me at least, is best with the ethernet connection. Possibly due to the ethernet input's I2S signal versus the one from the USB I'm guessing. You might have some “cross-wires” there… there’s no I2S signal involved in Ethernet streaming. UPnP/DLNA relies on RTSP packet flows.
valenroy Posted March 22, 2019 Posted March 22, 2019 Lets inject some reality into some of marketing hyperbole in this thread ;D Not all Upsampling is the same; each Vendor’s upsampling interpolation algorithm is going to be totally different, so even if Vendor A upsamples to 16xDSD, it doesn’t automatically mean it’s going to sound better than Vendor B who only upsamples to 10xDSD. Whether or not the above is true (any 3rd party sources? or you’re just reading the marketing brochure?), the end-result seems to suggest it’s nowhere as good as other comparable systems, especially w.r.t clocking/jitter rejection. See for yourself :) https://www.stereophile.com/content/ifi-audio-pro-idsd-da-processorheadphone-amplifier-measurements Compared to, say, the Chord Cutest (@24% cheaper than the iFi), without any such special “super” clock ;D https://www.stereophile.com/content/chord-electronics-qutest-da-processor-measurements You might have some “cross-wires” there… there’s no I2S signal involved in Ethernet streaming. UPnP/DLNA relies on RTSP packet flows. Wow, I didn't expect someone to take the time to reply in such detail! I totally agree with you on the implementation of the upsampling algorithm. There's always more than 1 way to do things, like the programs that I write is totally different from how my peers would approach the issue but the end objective is common. Actually, you're correct that the dlna network transmit data in packets, I should have been more specific that it's the I2S signal output from the ethernet audio module. Here's an interesting read on how things are processed on the iDSD. https://ifi-audio.com/how-the-pro-idsd-actually-works/
Guest AndrewC Posted March 22, 2019 Posted March 22, 2019 ... Actually, you're correct that the dlna network transmit data in packets, I should have been more specific that it's the I2S signal output from the ethernet audio module. Here's an interesting read on how things are processed on the iDSD. https://ifi-audio.com/how-the-pro-idsd-actually-works/ Well, according to that doc you linked, the Pro iDSD uses the XMOS XU216 chip which handles all interfaces (with only 512KB of memory). That chip has a native USB 2.0 PHY directly on the chip itself, so suggesting that Ethernet may sound better because it’s module is connected via I2S to the XMOS chip doesn’t make a whole lot of sense… you can’t feed the XMOS data streams any better than from USB. That said, it’s possible the USB input may benefit from a USB galvanic isolation solution, you should give one a try. By the way, the Pro iDSD and PS-Audio DACs don’t use the same Spartan-6 chip; the Pro iDSD uses the LX9, which has basically half the processing power of the LX16 used in the PS-Audio DS/JR. Half the power, for about a third the price, and about 6 years difference in market introduction… Practically speaking, for that kind of differential in Moore’s Law price-performance terms, the Pro iDSD should have been slightly over two times the power of the DS/JR, not half :P
Doggie Howser Posted March 23, 2019 Author Posted March 23, 2019 I have been doing more testing on the Pro iDSD (speaking of which, I am sure I already created one for this product... oh well)... and I reckon it is an amazing headphone amp with streamer DAC built in for my Focal headphones but as a DAC/streamer, it doesn't sound as good as most of my other DACs. The DSJ sounds better, as does the Naim ND5 XS2
Guest AndrewC Posted March 23, 2019 Posted March 23, 2019 I have been doing more testing on the Pro iDSD (speaking of which, I am sure I already created one for this product... oh well)... Yes you did... Mods, please merge :) http://www.xtremeplace.com/yabbse/index.php?topic=273419.0
valenroy Posted March 23, 2019 Posted March 23, 2019 Well, according to that doc you linked, the Pro iDSD uses the XMOS XU216 chip which handles all interfaces (with only 512KB of memory). That chip has a native USB 2.0 PHY directly on the chip itself, so suggesting that Ethernet may sound better because it’s module is connected via I2S to the XMOS chip doesn’t make a whole lot of sense… you can’t feed the XMOS data streams any better than from USB. That said, it’s possible the USB input may benefit from a USB galvanic isolation solution, you should give one a try. By the way, the Pro iDSD and PS-Audio DACs don’t use the same Spartan-6 chip; the Pro iDSD uses the LX9, which has basically half the processing power of the LX16 used in the PS-Audio DS/JR. Half the power, for about a third the price, and about 6 years difference in market introduction… Practically speaking, for that kind of differential in Moore’s Law price-performance terms, the Pro iDSD should have been slightly over two times the power of the DS/JR, not half :P I did try both USB and Ethernet primarily for my listening assessments and have come to the conclusion that in my setup at least, the wired Ethernet connection sound better to me than the USB one (of course with all the other variables like my silver usb cables and ethernet cables out of the way). I'm pleasantly surprised as well. All the inputs in the iDSD are galvanically isolated by the way and this is not limited to the USB input alone. The more I read about about this topic, the more stuff I uncover, like how Ethernet connections are inherently galvanically isolated to begin with, which perhaps makes it a better interface than USB for network audio streaming to begin with. There's a whole big thread on this on Computer Audiophile... and way too much information to digest at one go. It's great that you point out the different Spartan-6 chips out there, I thought that they are all the same! Thanks for doing the legwork! From your perspective, it makes a strong point why such a less capable FPGA was used in the iDSD based on your rationale, which is why I did propose a question in my earlier post on the internal architecture of the DAC. One-size fits all versus divide and conquer. What Ted has done is to basically do everything within the FPGA itself > Input, processing and output. As for the iDSD, it's broken down into 3 more specialised components to reach the end goal. I'm certain each has their merits. i.e. Directstream would have short signal paths within the FPGA itself since everything is housed under the same roof, whereas the iDSD could potentially have better noise isolation from the digital processing by using its FPGA specifically for digital only and shielding it from the analog conversion section. And that would lead us to question whether the lower-end FPGA is justified and adequate in a modern day product. And to be objective, if the FPGA is for the sole purpose of digital processing only, I would think it's more than adequate don't you think? Versus the more powerful Directstream FPGA that needs to basically handle everything. Again... I can't comment at the moment which is a better implementation until i get hold the DS Jr. that would only be available from June. I would think only Doggie can provide his perspective on these two products at the moment since he owns both. Anyway thanks for spending quality weekend time to reply in this thread. I am sure it's going to provide entertaining reading material for some.
valenroy Posted March 23, 2019 Posted March 23, 2019 The ability to switch between input stages is sublime! Tube+ really adds honey to vocal tracks... never knew my ATCs can sound so sweet... still cannot stand the consistent dropouts from DSD streaming... irritating bugger... ruin the enjoyment
wizardofoz Posted March 23, 2019 Posted March 23, 2019 Some xmos implementations have issues playing when rates changes take place.
Guest AndrewC Posted March 23, 2019 Posted March 23, 2019 ... All the inputs in the iDSD are galvanically isolated by the way and this is not limited to the USB input alone. ... You sure about that?? I don't think so... :P Looks like the inputs are not isolated at all, the only galvanic isolation is digitally post FPGA to the analog conversion stage. Two clues that confirm this; They say as much (in the doc you linked to no less! ;D) https://ifi-audio.com/how-the-pro-idsd-actually-works/ From the XMOS the I2S PCM signal or DSD signal is passed to the FPGA for any filtering/conversion etc. and from there through a galvanic isolation barrier to the final re-clocking and DACs. This makes sure that all the noisy ‘digital’ processing and its noise is contained on the digital module and cannot affect the DACs and analogue side. And, if you look at the images of the main digital board on both iFi’s website and the 6Moon review, you can see the “galvanic isolation” labeled chips at the interface between the digital board and the analog conversion board (it’s a Maxim Integrated MAX14850 chip) And, while the Ethernet PHY is inherently galvanically isolated (balanced transformer coupled), parasitic noise can still bridge that gap (hence the probable reason for “sonic difference” between Ethernet cables). Bottom line, there's no galvanic isolation at the inputs contrary to what you may have read on their marketing material. There’s nothing preventing crap from your PC/USB/Ethernet Switch mucking things at the digital stage from the inputs to your Pro iDSD. :)
mingzhen Posted September 21, 2019 Posted September 21, 2019 Does anyone compare the sound quality of Pro iDSD vs Benchmark DAC3 vs MyTek Brooklyn DAC+?
Recommended Posts