SoNic67

Sadly those mean very little. The dynamic performance of a r-2r is not something that can be measured with static sine signals. Sure, on paper the d-s seems that is better, but that is just for a static signal. And even then, the performence delta for static sin wave is actually... unaudible. Since I am listening to music, there is something that those measurements cannot tell me. As for Miska, he is wrong again, a quick glance at one of TI's datasheets would solve his lack of knowledge: http://www.ti.com/lit/ds/symlink/pcm1791a.pdf See page 53 of the above pdf, the Advanced Segment is explained there with a nice diagram. Digital input data via the digital filter is separated into 6 upper bits and 18 lower bits. The 6 upper bits are convertedto inverted complementary offset binary (ICOB) code. The lower 18 bits, in association with the MSB, are processedby a five-level third-order delta-sigma modulator operated at 64 fSby default. The 1 level of the modulator is equivalentto the 1 LSB of the ICOB code converter. The data groups processed in the ICOB converter and third-orderdelta-sigma modulator are summed together to an up to 66-level digital code, and then processed by data-weightedaveraging (DWA) to reduce the noise produced by element mismatch. The data of up to 66 levels from the DWA isconverted to an analog output in the differential-current segment section.

Don't use the optical out from iPurifier, the best option is to plug it's RCA out directly into the DAC coaxial input. Any optical input will add jitter compared to a coax input, especially with longer optical cables, and that includes your AVR optical inputs. Using the coaxial output of iPurifier avoids generating jitter after the device, in the optical converter. PS: I am using independent power supplies for my CCA and the iPurifier, the ones that came with the devices.

Miska, I guess you don't have the language skills to pick up what I said. R-2R was one way of implementing multi bit conversion, that's all and I NEVER said that is the same like DS. That's the strawman attack. The reality is that pure delta-sigma is a failure, that's why TI uses 6 bits of true multibit in the upper significant bits and then a 5 level SDM for the lower 18 bits. Sabre uses 64 "equally weighted elements", who cares, in the end they do the same function - when summed those currents, they will behave like a multibit that's equivalent to an analog 6 bits (each with noise shaping). Sure they will do an averaging to reduce element mismatch, but that's not changing fundamentally what happens. The single element SMD can't work properly alone. But I think there is no point in arguing with you further, since you are rude as always (ad hominem attack at the end).

Well, the title says comparation between R-2R and Delta-Sigma. I just say that modern "delta-sigma" are not just that, they are a evolution of delta-sigma back towards multibit ideea that is represented by R-2R model. If you accept muddying the waters, then... when do you stop? Example: Adding dithering to a R-2R can be also considered? How about filtering and OS? That's not usually part of the R-2R DAC's but then... they are integral parts of a modern D-S DAC assemblies (otherwise they are just switched capacitors). I am not saying that is one better or worse, just that... if he doesn't set the limits well, will end up comparing apples to hubcaps.

Not really fair, Miska. The ESS chip have a multibit modulator architecture (HyperStream), plus they are much more than just a pure DAC chip (filter/jitter-reduction). Not sure about AKM... but I won't be surprised to have multiple switched capacitors on output.

The delta-sigma from TI have the first 5 bits as multi bit so it will not be a relevant comparison between the two technologies. Also, other manufacturers whent to the path of multiple delta-sigma modulators, mainly to compensate for the shortcomings of a pure delta-sigma design, found in the early implementations.

On some existing products it might be provided as an added purchase - Denon offers such an option, based on the serial number. In this way the royalties will be paid by the people who really want that feature, not from that firm's profits (or by future purchasers, added to their costs).

For CCA I am using Bubble UPnP installed on my Android phone - it is streaming 96kHz to my receiver. Also, I have installed their server on my Win10 machine to share my music collection and I noticed that it has an option for Qobuz and 96kHz. https://www.bubblesoftapps.com/bubbleupnpserver/

I was a die-hard supporter of the "CD's need to be played in a CD player" theory, mainly because in this way the signal goes from the pickup to the DAC via i2s signals. Hence the clock is always present, separated from data, always jitter-free. This is opposed to SPDIF that embeds the clock with the data and hence is prone to jitter during transmission and recovery. Well, I just jumped in the Chromecast Audio bandwagon. Bought two of them - one for the living room Denon AVR-3808CI receiver and one for the bedroom that has a new-to-me multibit DAC (PCM63J-K fed by a Yamaha SM5843A SPDIF receiver). Needless to say that the receiver doesn't need any jitter-reduction because of the internal DSP processing (dual processors, lots of RAM). But I tought that the multibit DAC was in need (can bennefit) of a cleaner SPDIF signal than what CCA can provide (Toslink is notoriously prone to jiiter), so... I have ordered the SPDIF iPurifier. Can't wait to hear the results. It is nice that the results of the jitter performance where posted here, I think they should be posted on the iFi website too (I didn't find anything as technical there). LE: I understand that the buffer memory is used in an adaptive mode, based on the incoming jitter, but I was wondering what's the actual size of the used memory?

Foobar - did you use DirectSound? DS will change the bitrate to whatever you have set in Windows. Use WASAPI or ASIO components - they need to be downloaded separately.

That "data" needs to be real-time perfect, because you hear in real time. Unless there is a huge RAM buffer in the DAC, PC jitter matters. You keep saying that is no clock in the signal. That's exactly the problem! The data is send without a clock, so any deviation at source cannot be "fixed" in DAC automatically. Hence the need for PLL loops at DAC level - with all their imperfections. Heck, most of the audiophiles don't even know that their belowed PC's have "spread spectrum" clock enabled in BIOS by default (to alleviate the EM compatibility required by FCC). That is jitter "by design".

Sometimes that's the best approach, it gives you the best flexibility. You need to know what load will be your headphones - 32 ohm (like Grado SR60)? An observation - Analog volume control is a must for 'modern' multilevel Sigma-Delta DAC's.

Funny... You didn't see his company link? Why do you think is posting here if not for advertising his product (nothing wrong IMO)?

What did I make up? Does DSD have any error correction or not? Is that that I don't know? Or is just another cheap insult? You cut off the rest of my post and choose only the part that you fit your agenda. Probably Miska did a mistake in his numbers, I didn't check that. But for sure are zero detectors in DSD DAC's for the same reason. Reality is this (quote again, so you can't attack me): A good idea in 1999, DSD was conceived to improve the quality of music at home over the prevailing 16-bit CD format. By bypassing the down-sampling and up-sampling filters in the CD audio chain, DSD sought to improve performance by shortening the audio signal path: However, DSD has been overtaken by modern technology; A-D and D-A converters have moved away from 1-bit, to far superior multi-bit processes, and the down-sampling and up-sampling filters that DSD sought to bypass have been rendered transparent by the use of higher sample rates and modern algorithms. In fact, DSD is now the quality bottleneck in the modern recording and playback chain. Or this: http://www.weiss-highend.ch/computerplayback/white-paper-on-DSD-en.pdf Or this: https://web.archive.org/web/20071011014242/http://www.smr-home-theatre.org/surround2002/technology/page_07.shtml

There is no metadata in DSD. Think about it - it can be 'decoded' by a passive filter, if it was some metadata, it would sound weird. On the other hand, that simplicity is one of the bad points - you are right, there is no zero detection, any DC correction in the stream is 'made' via noise shaping at recording time. As Miska pointed out, any loss in the data stream translates in amazing errors - it takes 1 full second to get down to -50 dB!