On Thursday, Dec. 3rd, I attended an interesting event at Goodwin's High End, Waltham, Mass., USA. Steve Silberman from Audioquest and John Quick from dCS Audio were there to present.
dCS Rossini DAC
The evening started with a presentation of the dCS Rossini DAC by John Quick which interested me in particular since I am considering a potential upgrade to that DAC. He started by explaining the dCS philosophy, which includes their own algorithms for digital processing and using all their own components, without pre-fabricated converter chips. Basically digital processing all from scratch. He was explaining the proprietary Ring DAC, and the noise shaping technology. He said the algorithm packages the noise in such a manner that it can be cleanly filtered out. The analog output stage of dCS DACs is also discrete, and uses opamps only as buffer components. He showed demonstration models of the internal boards of the DACs (I think mainly the Vivaldi DAC) and explained how components were spaced apart in order to minimize influence of radiation upon one another. Since I sat in front, at some point I was able to grab these demonstration boards and look more closely; they seem impressively built indeed. I asked about the philosophy to put the clock for the DAC into a discrete chassis, and he said that this was also done with the intention of minimizing radiation influence (when an an external clock is connected to the DAC, the internal clock of the DAC is automatically set to be just a passive receiver if I remember correctly). He said there is no perfect solution and everything has its advantages and disadvantages. Some manufacturers (like Berkeley, personal remark) swear by putting the clock right next to the DAC in order to maximally reduce jitter, but that in turn may have the disadvantage of increasing radiation influence of the clock on the DAC. John said that in the experience of dCS it paid sonic dividends to separate the components, and it became clear that dCS really is interested in the, in their mind, least compromised solution. A number of people have complained that dCS is just out to make money by separating all the components, but at least for me concerns regarding the philosophy of the company were laid to rest. In a private conversation after the official event John also pointed out that the separation of components is usual in pro audio, where the roots of the dCS thinking seem to be. Indeed, these days most studios have their own dedicated external master clock for analog/digital conversion.
Computer Audio
After the dCS presentation, Steve Silberman from Audioquest took over for the part that consumed most of the evening, computer audio. John Quick stayed around for it, also in support and for answering certain questions. In fact, the two presenters seemed to be in broad agreement with each other on all the points discussed. Steve started with a smile, having great fun quoting responses to a presentation that he gave on the topic that is posted on Youtube and where he points out that RAM and CPU are important. As quoted by Steve, in the comments section one person said: "This guy is full of sh*t!", and another one claimed "he has no clue what he is talking about". So Steve said this suggested that he apparently needed to work further on the presentation to better bring his point across. He started by talking about the focus of his Youtube presentation, the importance of RAM and CPU; the computer should not have to spend most of its resources to serving the musical data. In particular he said that there must be enough memory so that the computer does not always shuffle packets of data back and forth which may affect timing of the data. The requirements on RAM and CPU for great audio seemed to be quite severe, I don't remember the exact numbers. He demonstrated the difference between a 'regular' computer and a very fast one with generous memory and CPU playing the same file, and it was obvious. Steve pointed out that the computer should do nothing else than play music when used for audio, no idling web browser in the background or the like. He also talked about the different ports, and suggested that USB is really not the optimal one, he preferred Thunderbolt; according to him all port options will be superseded by USB-C in the near future which hopefully will be superior. Steve also mentioned the Mac program BitPerfect which appears to help, and that you should never stream the data of your music wirelessly; wireless was only suitable for command connection points. For streaming you should also buy your own router; there are affordable ones that are superior to what the cable provider gives you. He also delved a bit into iTunes and how you can make it much better by modifying the default settings, including sampling rate. In particular he suggested that during data transfer from a disc error correction should always be on in order to obtain a bit-by-bit copy. It only takes considerably longer.
Steve then explained the different programs for servers, JRiver and Roon, and while he liked JRiver for the best connectivity of different sources, he said that Roon is a much more user-friendly interface and provides much more information. For example, the program could find that a certain musician was active on such and such number of his filed recordings, which was in the hundreds. Roon also allows for downloading entire CD booklets and other background info on the artists. If I remember correctly, he expected that in the future the connectivity of Roon should rival that of JRiver, and that at that stage it should be the superior solution. One person in the audience was asking the question that I had in mind as well, which is how a Baetis server (the current Goodwin's favorite) compared to all the regular computers used for digital audio. Steve said that this was the best computer that you could have (personal remark: the server also plays everything like a regular computer would, including Blu-Ray and cconcerts from Youtube, for example). In any case, it became very obvious that there are tons of things to consider when it comes to computer audio and myriads of potential pitfalls, and if I ever will go down the route of playing files I will do that over a Baetis server.
USB and Jitterbug
Steve also talked about USB cables and that there is a misconception that what is transmitted are just 0s and 1s. This is not true, the signal is in fact a series of square waves, and the shape of the wave is compromised by inferior cables and connections. He showed the reference cable of a German company for ethernet connections, and it was massive with elaborate connectors. I remarked that I also had seen on the web presentations about measured square waves over different ethernet connections, and that the quality was different, visibly so in the graphs showed -- that was certainly no audiophile voodoo, since audiophilia was not even addressed in applications, it was just about technical measurements. Just as the mentioned German ethernet reference cable was not even for audiophile purposes. Steve then demonstrated the difference between a cheap USB cable and one that cost 700 dollars, and the difference was easily obvious, even from my compromised seating position outside the left boundary of the left speaker (Rockport Cygnus speakers were driven by Simaudio monoblocks with the Rossini DAC as the source over a Simaudio pre-amp). The regular USB cable sounded flat in comparison, with the more expensive cable sounding much more expressive.
Then Steve proceeded to a demonstration of Audioquest's Jitterbug. He began by addressing the claim that there are no measurable differences, a claim made by a variety of people including John Atkinson from Stereophile. Yet he pointed out that usually people don't have the right equipment, and that in order to measure the differences you need a Tektronix USB analyzer, an instrument that does nothing else but costs $ 35,000. A publication like Stereophile simply cannot afford such equipment, and neither can most who do not engage in USB analysis as a main focus of their work. Finally the Jitterbug device was put to the audio test. We first heard a track from file through the Rossini DAC, and then from the same computer (the powerful and fast one) with the Jitterbug plugged in. Even though I was not in an optimal listening spot (see above) the difference was loud and clear. While there was also some improvement on overtones of acoustic guitar with the Jitterbug in place, the biggest advancement was heard on the voice -- notably more transparent and expressive in its inflection. People seemed to agree. One person asked why there was no blind test performed, and Steve said that blind tests do not work because they put your perception under stress. I certainly could agree that the effect of the Jitterbug was an easily audible improvement. In order to gauge the reaction of John Quick from dCS I asked him on the side, so the Jitterbug improves the signal that goes from the computer to the Rossini DAC? He nodded his head in agreement. Apparently he had no problem with the concept that a $ 50 device could improve the datastream to a $ 24,000 DAC which would benefit the sonic output of the latter.
A conversation after the event
After the official event was over, I had a conversation with John Quick, pointing out my interest in the Rossini DAC. Since I am only interested in Redbook CD as digital medium, I asked about transport vs. server. He agreed with Steve from Audioquest that the Baetis server would be the best server that I could get, but he also pointed out that great transports still often have the edge. He said that he had compared several servers (not the Baetis I believe) with his Scarlatti transport, and that each time the transport either came out on top or was at least as good as the server. He said the Baetis server should be able to deliver a high-quality signal, but only if the implementation was good. It requires an external drive, and this configuration must be done right in order to obtain great results. It only confirmed my suspicion that I have had for several months now, that the idea that computer audio is by definition superior to a CD transport is just simplistic, and that computer audio really requires the utmost care and dedication in implementation to sound right, a point emphasized throughout in the presentation by Steve from Audioquest as well. I asked if my $ 2K Simaudio transport would be a good source, and John said that it is a very good transport and confirmed that a combination of that transport with a Rossini DAC would indeed make sense. He suggested that a Rossini CD player would also be a possibility.
I am interested in exploiting the potential of Redbook CD, where all the music is, to the fullest within my budget, something of far greater interest to me than adding hi-res with its severely limited musical output. If I would have a Rossini DAC at some point and would decide about further digital upgrades, I might forego a Baetis server and rather, for the same expense, buy an external Rossini clock instead, with the Rossini combo still driven by my Simaudio transport. Or alternatively, buy a Rossini CD player. Unless of course I want to get into the digital room correction thing, on top my acoustic room treatments, which makes it a different story.
***
It was a fascinating evening where I learned a lot. Goodwin's High End was a great host for the event, and the food was delicious. Really nice sandwiches and at some point I had a piece of some of the most amazing cheese... It was good to see Alan and Paul again.
EDIT 12-12-15: Corrections about the USB analyzer (see discussion posts #8, 11 and following), the type of Rockport speakers used, and the Baetis server. Comments about the Rossini CD player are added as well.
dCS Rossini DAC
The evening started with a presentation of the dCS Rossini DAC by John Quick which interested me in particular since I am considering a potential upgrade to that DAC. He started by explaining the dCS philosophy, which includes their own algorithms for digital processing and using all their own components, without pre-fabricated converter chips. Basically digital processing all from scratch. He was explaining the proprietary Ring DAC, and the noise shaping technology. He said the algorithm packages the noise in such a manner that it can be cleanly filtered out. The analog output stage of dCS DACs is also discrete, and uses opamps only as buffer components. He showed demonstration models of the internal boards of the DACs (I think mainly the Vivaldi DAC) and explained how components were spaced apart in order to minimize influence of radiation upon one another. Since I sat in front, at some point I was able to grab these demonstration boards and look more closely; they seem impressively built indeed. I asked about the philosophy to put the clock for the DAC into a discrete chassis, and he said that this was also done with the intention of minimizing radiation influence (when an an external clock is connected to the DAC, the internal clock of the DAC is automatically set to be just a passive receiver if I remember correctly). He said there is no perfect solution and everything has its advantages and disadvantages. Some manufacturers (like Berkeley, personal remark) swear by putting the clock right next to the DAC in order to maximally reduce jitter, but that in turn may have the disadvantage of increasing radiation influence of the clock on the DAC. John said that in the experience of dCS it paid sonic dividends to separate the components, and it became clear that dCS really is interested in the, in their mind, least compromised solution. A number of people have complained that dCS is just out to make money by separating all the components, but at least for me concerns regarding the philosophy of the company were laid to rest. In a private conversation after the official event John also pointed out that the separation of components is usual in pro audio, where the roots of the dCS thinking seem to be. Indeed, these days most studios have their own dedicated external master clock for analog/digital conversion.
Computer Audio
After the dCS presentation, Steve Silberman from Audioquest took over for the part that consumed most of the evening, computer audio. John Quick stayed around for it, also in support and for answering certain questions. In fact, the two presenters seemed to be in broad agreement with each other on all the points discussed. Steve started with a smile, having great fun quoting responses to a presentation that he gave on the topic that is posted on Youtube and where he points out that RAM and CPU are important. As quoted by Steve, in the comments section one person said: "This guy is full of sh*t!", and another one claimed "he has no clue what he is talking about". So Steve said this suggested that he apparently needed to work further on the presentation to better bring his point across. He started by talking about the focus of his Youtube presentation, the importance of RAM and CPU; the computer should not have to spend most of its resources to serving the musical data. In particular he said that there must be enough memory so that the computer does not always shuffle packets of data back and forth which may affect timing of the data. The requirements on RAM and CPU for great audio seemed to be quite severe, I don't remember the exact numbers. He demonstrated the difference between a 'regular' computer and a very fast one with generous memory and CPU playing the same file, and it was obvious. Steve pointed out that the computer should do nothing else than play music when used for audio, no idling web browser in the background or the like. He also talked about the different ports, and suggested that USB is really not the optimal one, he preferred Thunderbolt; according to him all port options will be superseded by USB-C in the near future which hopefully will be superior. Steve also mentioned the Mac program BitPerfect which appears to help, and that you should never stream the data of your music wirelessly; wireless was only suitable for command connection points. For streaming you should also buy your own router; there are affordable ones that are superior to what the cable provider gives you. He also delved a bit into iTunes and how you can make it much better by modifying the default settings, including sampling rate. In particular he suggested that during data transfer from a disc error correction should always be on in order to obtain a bit-by-bit copy. It only takes considerably longer.
Steve then explained the different programs for servers, JRiver and Roon, and while he liked JRiver for the best connectivity of different sources, he said that Roon is a much more user-friendly interface and provides much more information. For example, the program could find that a certain musician was active on such and such number of his filed recordings, which was in the hundreds. Roon also allows for downloading entire CD booklets and other background info on the artists. If I remember correctly, he expected that in the future the connectivity of Roon should rival that of JRiver, and that at that stage it should be the superior solution. One person in the audience was asking the question that I had in mind as well, which is how a Baetis server (the current Goodwin's favorite) compared to all the regular computers used for digital audio. Steve said that this was the best computer that you could have (personal remark: the server also plays everything like a regular computer would, including Blu-Ray and cconcerts from Youtube, for example). In any case, it became very obvious that there are tons of things to consider when it comes to computer audio and myriads of potential pitfalls, and if I ever will go down the route of playing files I will do that over a Baetis server.
USB and Jitterbug
Steve also talked about USB cables and that there is a misconception that what is transmitted are just 0s and 1s. This is not true, the signal is in fact a series of square waves, and the shape of the wave is compromised by inferior cables and connections. He showed the reference cable of a German company for ethernet connections, and it was massive with elaborate connectors. I remarked that I also had seen on the web presentations about measured square waves over different ethernet connections, and that the quality was different, visibly so in the graphs showed -- that was certainly no audiophile voodoo, since audiophilia was not even addressed in applications, it was just about technical measurements. Just as the mentioned German ethernet reference cable was not even for audiophile purposes. Steve then demonstrated the difference between a cheap USB cable and one that cost 700 dollars, and the difference was easily obvious, even from my compromised seating position outside the left boundary of the left speaker (Rockport Cygnus speakers were driven by Simaudio monoblocks with the Rossini DAC as the source over a Simaudio pre-amp). The regular USB cable sounded flat in comparison, with the more expensive cable sounding much more expressive.
Then Steve proceeded to a demonstration of Audioquest's Jitterbug. He began by addressing the claim that there are no measurable differences, a claim made by a variety of people including John Atkinson from Stereophile. Yet he pointed out that usually people don't have the right equipment, and that in order to measure the differences you need a Tektronix USB analyzer, an instrument that does nothing else but costs $ 35,000. A publication like Stereophile simply cannot afford such equipment, and neither can most who do not engage in USB analysis as a main focus of their work. Finally the Jitterbug device was put to the audio test. We first heard a track from file through the Rossini DAC, and then from the same computer (the powerful and fast one) with the Jitterbug plugged in. Even though I was not in an optimal listening spot (see above) the difference was loud and clear. While there was also some improvement on overtones of acoustic guitar with the Jitterbug in place, the biggest advancement was heard on the voice -- notably more transparent and expressive in its inflection. People seemed to agree. One person asked why there was no blind test performed, and Steve said that blind tests do not work because they put your perception under stress. I certainly could agree that the effect of the Jitterbug was an easily audible improvement. In order to gauge the reaction of John Quick from dCS I asked him on the side, so the Jitterbug improves the signal that goes from the computer to the Rossini DAC? He nodded his head in agreement. Apparently he had no problem with the concept that a $ 50 device could improve the datastream to a $ 24,000 DAC which would benefit the sonic output of the latter.
A conversation after the event
After the official event was over, I had a conversation with John Quick, pointing out my interest in the Rossini DAC. Since I am only interested in Redbook CD as digital medium, I asked about transport vs. server. He agreed with Steve from Audioquest that the Baetis server would be the best server that I could get, but he also pointed out that great transports still often have the edge. He said that he had compared several servers (not the Baetis I believe) with his Scarlatti transport, and that each time the transport either came out on top or was at least as good as the server. He said the Baetis server should be able to deliver a high-quality signal, but only if the implementation was good. It requires an external drive, and this configuration must be done right in order to obtain great results. It only confirmed my suspicion that I have had for several months now, that the idea that computer audio is by definition superior to a CD transport is just simplistic, and that computer audio really requires the utmost care and dedication in implementation to sound right, a point emphasized throughout in the presentation by Steve from Audioquest as well. I asked if my $ 2K Simaudio transport would be a good source, and John said that it is a very good transport and confirmed that a combination of that transport with a Rossini DAC would indeed make sense. He suggested that a Rossini CD player would also be a possibility.
I am interested in exploiting the potential of Redbook CD, where all the music is, to the fullest within my budget, something of far greater interest to me than adding hi-res with its severely limited musical output. If I would have a Rossini DAC at some point and would decide about further digital upgrades, I might forego a Baetis server and rather, for the same expense, buy an external Rossini clock instead, with the Rossini combo still driven by my Simaudio transport. Or alternatively, buy a Rossini CD player. Unless of course I want to get into the digital room correction thing, on top my acoustic room treatments, which makes it a different story.
***
It was a fascinating evening where I learned a lot. Goodwin's High End was a great host for the event, and the food was delicious. Really nice sandwiches and at some point I had a piece of some of the most amazing cheese... It was good to see Alan and Paul again.
EDIT 12-12-15: Corrections about the USB analyzer (see discussion posts #8, 11 and following), the type of Rockport speakers used, and the Baetis server. Comments about the Rossini CD player are added as well.
Last edited: