lmitche Posted March 15, 2016 Share Posted March 15, 2016 Regarding the DSC1,2 etc, my current thinking is that the USB connection is the weak link, and while a great deal of work has gone into optimizing it, there seems to be no end to the stream of widgets that are now chained together in search of USB perfection, including boxes and cables that are several hundreds of $$ each. Even entire computers that have been designed with the goal of optimizing the USB connection. Shrug. We can have it all! Jabbr, in theory USB should work fine without any need for enhancements. Ed, with the SGM and T+A DAC 8 DSD are you finding a need to enhance the USB chain? One of the exciting things about the SGM is that the team may have found the minimally complex solution to CA sound quality. There is an old adage that you can't eliminate complexity, you can only move it. The NAA moves it. The SGM proves there is an even less complex solution. Pareto Audio aka nuckleheadaudio Link to comment
jabbr Posted March 15, 2016 Share Posted March 15, 2016 Jabbr, in theory USB should work fine without any need for enhancements. Ed, with the SGM and T+A DAC 8 DSD are you finding a need to enhance the USB chain? One of the exciting things about the SGM is that the team may have found the minimally complex solution to CA sound quality. There is an old adage that you can't eliminate complexity, you can only move it. The NAA moves it. The SGM proves there is an even less complex solution. I have no doubt at all that SGM + T+A DAC 8 DSD is an outstanding sound. If I weren't tinkering with DSC1+ => DSC2 I would be very very tempted to purchase. Regarding the goal of eliminating NAA, for me I'm not sure that reducing complexity -- in and of itself -- is the goal, after all the LP and analog circuit chain is unquestionably simpler! For me, the challenge is to push the limits of what is possible regardless of complexity, but understanding that all else being equal, simpler is better. These quotes are apropos: Clarke's first law: When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is very probably wrong. Clarke's second law: The only way of discovering the limits of the possible is to venture a little way past them into the impossible. Clarke's third law: Any sufficiently advanced technology is indistinguishable from magic. In any case for me, the interest is in building a entirely discrete DAC, pushing the limits of what is possible, and integrated fed by a very modern interface. So we start with the most powerful workstation/GPU and then feed an ARM/FPGA SoC and then feed an entirely discrete DA, with a Borbely inspired Pass inspired, EUVL cascoded jfet I-V and discrete jfet SK filters. Now this is more complex than the opamp DSC1 design, but we will see what is possible. Custom room treatments for headphone users. Link to comment
Jud Posted March 15, 2016 Share Posted March 15, 2016 To me, the most elegant approach would be to use an ARM/FPGA SoC (e.g. the Xilinx Zynq 7000 series) to handle SGMII input (what comes out of the fiberoptic SFP module) and convert to direct DSD output. That is: fiberoptic Ethernet in, direct DSD out -> DSC1 shift register FIR board (and then I-V and analog SK filters). So how would that work? The ARM processor runs Linux and hence networkaudiod (NAA). Do you still consider that another machine? If so why would you accept an SoC to convert USB -> DSD but not to convert Ethernet -> DSD. I think that's even more KISS because there is no USB at all. I like this. As I've outlined, there are some recent techniques that have the potential to be real breakthroughs. I also think room correction kernels will become more desirable, and certainly at DSD512 the GPU coprocessor is necessary. We can have it all! Not so thrilled about this. Room correction means more filters, and filters have inherent flaws that need optimizing. Filters may require different optimization for room correction than for D/A conversion with fewest audible artifacts. For example, I prize phase linearity, while many (most?) room correction filters are minimum phase. I presume there's a reason for the latter, but still, I don't like mixing two tasks. My dad used to run a hardware store (nearly the same as designing digital filters), and he used to say "Never buy anything that does two things, because it won't do either of them well." Anyway, I suppose I don't have to use them if I don't want/need them. If I weren't tinkering with DSC1+ => DSC2 I would be very very tempted to purchase. Regarding the goal of eliminating NAA, for me I'm not sure that reducing complexity -- in and of itself -- is the goal, after all the LP and analog circuit chain is unquestionably simpler! For me, the challenge is to push the limits of what is possible regardless of complexity, but understanding that all else being equal, simpler is better. In any case for me, the interest is in building a entirely discrete DAC, pushing the limits of what is possible, and integrated fed by a very modern interface. So we start with the most powerful workstation/GPU and then feed an ARM/FPGA SoC and then feed an entirely discrete DA, with a Borbely inspired Pass inspired, EUVL cascoded jfet I-V and discrete jfet SK filters. Now this is more complex than the opamp DSC1 design, but we will see what is possible. What appeals about EUVL's cascoded design vs. other possibilities? Why Sallen-Key? As I noted, I like phase linearity. Being second order (or at least so I've read), can Sallen-Key be phase linear? One never knows, do one? - Fats Waller The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature. Link to comment
jabbr Posted March 15, 2016 Share Posted March 15, 2016 I like this. ... Not so thrilled about this. Room correction means more filters, and filters have inherent flaws that need optimizing. Filters may require different optimization for room correction than for D/A conversion with fewest audible artifacts. For example, I prize phase linearity, while many (most?) room correction filters are minimum phase. I presume there's a reason for the latter, but still, I don't like mixing two tasks. My dad used to run a hardware store (nearly the same as designing digital filters), and he used to say "Never buy anything that does two things, because it won't do either of them well." Anyway, I suppose I don't have to use them if I don't want/need them. Glad you like approach:) As you mention room kernels are optional. Considering system convolution/deconvolution, the DAC, Amp, Speakers and Room can each have their own kernel if desired and you can optimize each individually. This is really good math:) HQPlayer being able to do this in the SDM domain is really special (and worth a good GPU in my mind! What appeals about EUVL's cascoded design vs. other possibilities? Described in this paper. Take the SEN circuit, and use BF862 ... which are cheap and available ... and cascode with J111 ala Borbely's I-V design (but simpler). The advantage is that you don't need the difficult to obtain P-channel 2SJ74s. Very low noise (can parallelize jfets to further reduce noise). Disadvantage is that this is an SMD design -- advantage of the SMD design is that all the jfets share the same heat sink hence improved temperature stability. Why Sallen-Key? As I noted, I like phase linearity. Being second order (or at least so I've read), can Sallen-Key be phase linear? SK is Miska's design. Pole at 100 khz, so shouldn't matter. Its actually fourth order. Pass also uses SK in his B4 active x-over. I'm taking the DSC1 design and implementing the analog side discretely ... well also balancing it Custom room treatments for headphone users. Link to comment
bogi Posted March 15, 2016 Share Posted March 15, 2016 To me, the most elegant approach would be to use an ARM/FPGA SoC (e.g. the Xilinx Zynq 7000 series) to handle SGMII input (what comes out of the fiberoptic SFP module) and convert to direct DSD output. That is: fiberoptic Ethernet in, direct DSD out -> DSC1 shift register FIR board (and then I-V and analog SK filters). So how would that work? The ARM processor runs Linux and hence networkaudiod (NAA). Do you still consider that another machine? If so why would you accept an SoC to convert USB -> DSD but not to convert Ethernet -> DSD. I think that's even more KISS because there is no USB at all. I'm trying to compare the above 'fiberoptic LAN to DSD NAA + DSD DAC box' with classical USB solution. The described fiberoptic to DSD approach needs to be integrated into DAC. That's currently nonstandard solution, so you won't have wide range of DACs to choose. USB is avoided, so you won't need to pay for Intona etc., but fiberoptic isn't standard part of computers like USB, so you would need to pay for that. Advantages of the fiberoptic to DSD approach are a) no additional boxes and LPS needed - no Regen/Intona/RUR box + LPS and no extra NAA box + LPS as NAA is integrated with DAC b) fiberoptic interconnection could be run over longer distances, so audio player PC could be placed in other room. It wouldn't need to be quite quiet. One audio player PC could be used with DACs in more rooms. But that would require optical network installation within a building. Wouldn't be possible to perform optical isolation only on the side of the the proposed 'LAN to DSD NAA + DAC box', so it would be possible to use standard computer network interface and standard LAN cables? Wireless but not noisy solution would be more convenient for multi room usage. i7 11850H + RTX A2000 Win11 HQPlayer ► Topping HS02 ► 2x iFi iSilencer ► SMSL D300 ► DIY headamp DHA1 ► HiFiMan HE-500 Link to comment
Miska Posted March 15, 2016 Share Posted March 15, 2016 Why Sallen-Key? As I noted, I like phase linearity. Being second order (or at least so I've read), can Sallen-Key be phase linear? What would you use? The filter I designed for DSC1 has -10 deg phase at 10 kHz and -25 deg phase at 20 kHz. And group delay is flat to 20 kHz. As a result, this combined with the non-ringing linear-phase conversion stage filter gives very good step response: Squarewaves from DACs - Blogs - Computer Audiophile Signalyst - Developer of HQPlayer Pulse & Fidelity - Software Defined Amplifiers Link to comment
jabbr Posted March 15, 2016 Share Posted March 15, 2016 I'm trying to compare the above 'fiberoptic LAN to DSD NAA + DSD DAC box' with classical USB solution. The described fiberoptic to DSD approach needs to be integrated into DAC. That's currently nonstandard solution, so you won't have wide range of DACs to choose. USB is avoided, so you won't need to pay for Intona etc., but fiberoptic isn't standard part of computers like USB, so you would need to pay for that. Internally the DSC1 accepts a direct DSD signal fed by the Amanero USB interface. An SFP interface can accept either a fiber optic or copper module. Advantages of the fiberoptic to DSD approach are a) no additional boxes and LPS needed - no Regen/Intona/RUR box + LPS and no extra NAA box + LPS as NAA is integrated with DAC b) fiberoptic interconnection could be run over longer distances, so audio player PC could be placed in other room. It wouldn't need to be quite quiet. One audio player PC could be used with DACs in more rooms. But that would require optical network installation within a building. Wouldn't be possible to perform optical isolation only on the side of the the proposed 'LAN to DSD NAA + DAC box', so it would be possible to use standard computer network interface and standard LAN cables? you can use an inexpensive FMC box to convert copper to fiber optic Ethernet. You can also use copper Ethernet (rj45) SFP module so: Yes! Custom room treatments for headphone users. Link to comment
Argopo Posted March 15, 2016 Share Posted March 15, 2016 you can use an inexpensive FMC box to convert copper to fiber optic Ethernet. You can also use copper Ethernet (rj45) SFP module so: Yes! I live in an apartment. I have runs of Cat5e underneath the floorboards from the office bedroom to living room. I can't replace those cables without tearing apart my apartment. So I clearly understand: Current proposed setup: Control PC -> Copper ethernet -> Router -> Copper ethernet (under floorboards) -> Gigabit Switch -> Copper ethernet -> NAA/Renderer -> USB -> DAC Fiber optic setup: Control PC -> Copper ethernet -> Router -> Copper ethernet (under floorboards) -> FMC Box -> Fiber optic ethernet -> Fiber optic NIC PCI-e card in NAA/Renderer -> USB -> DAC If part of the connection between NAA and control PC involves copper ethernet, does this not defeat the benefit of using a partial run of Fiber optic? Ideally, it would be: Fiber Optic NIC PCI-e card in Control PC -> FMC -> Router -> FMC -> Fiber Optic NIC PCI-e card in NAA. Software: HQPlayer | JRiver | Fidelizer Pro | Roon | Qobuz Music Server: i7 6700K (Windows 10) | DAC: T+A DAC8 DSD, Marantz SA 14S-1, Schitt Yggdrasil | Preamp: DIY AMB alpha24 Fully-differential line amp | Amp: DIY M3 Balanced or DIY Tube Amp (2A3-300B) | Headphone: Shure SRH-1840, Audeze LCD-X, AKG K-501, Sennheiser HD600, HD800 | Speakers: Klipsch Heresey III Link to comment
jabbr Posted March 15, 2016 Share Posted March 15, 2016 We are discussing putting an SFP input on the DSCx DAC. So: FMC -> fiber optic Ethernet -> DAC Custom room treatments for headphone users. Link to comment
lmitche Posted March 15, 2016 Share Posted March 15, 2016 We are discussing putting an SFP input on the DSCx DAC. So: FMC -> fiber optic Ethernet -> DAC Or PC > fiber NIC > fiber > fiber DAC? Pareto Audio aka nuckleheadaudio Link to comment
ericuco Posted March 15, 2016 Share Posted March 15, 2016 Ideally, it would be: Fiber Optic NIC PCI-e card in Control PC -> FMC -> Router -> FMC -> Fiber Optic NIC PCI-e card in NAA. Ideally, Fiber Optic NIC PCI-e card in Control PC -> Fiber Optic switch/router -> Fiber Optic NIC PCI-e card in NAA However, your router is probably not fiber optic but you can use the FMC box(es) to connect the router to the fiber optic switch. Here is what I use for a switch: Dual SFP Slot Media Converter with Dual 1000M Ethernet Ports At the NAA end (Cubox-i), I have to use one of the FMC boxes because it is copper only. Eric Audio System Link to comment
bogi Posted March 15, 2016 Share Posted March 15, 2016 jabbr answered to my question: Wouldn't be possible to perform optical isolation only on the side of the the proposed 'LAN to DSD NAA + DAC box', so it would be possible to use standard computer network interface and standard LAN cables? Would you expect any difference in sound quality between a) and b) ? a) the whole path between control PC and NAA is fiberoptic b) optical isolation is performed in the NAA+DAC unit and all cables are metallic i7 11850H + RTX A2000 Win11 HQPlayer ► Topping HS02 ► 2x iFi iSilencer ► SMSL D300 ► DIY headamp DHA1 ► HiFiMan HE-500 Link to comment
Jud Posted March 15, 2016 Share Posted March 15, 2016 And group delay is flat to 20 kHz. Ah, nice. I think that takes care of most of my concern. I'm mostly ignorant of what is going on, so I only know to be concerned about the timing for everything remaining consistent through the speakers, which are (at least in the words of the advertising) "time and phase correct," i.e., sound that occurred at the same time on the recording is supposed to arrive from each of the drivers at the listening position at the same time and in phase. It uses first-order crossovers to help achieve this. So that's all I know about filter order, phase, etc. One never knows, do one? - Fats Waller The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature. Link to comment
YashN Posted March 15, 2016 Share Posted March 15, 2016 To me, the most elegant approach would be to use an ARM/FPGA SoC (e.g. the Xilinx Zynq 7000 series) to handle SGMII input (what comes out of the fiberoptic SFP module) and convert to direct DSD output. That is: fiberoptic Ethernet in, direct DSD out -> DSC1 shift register FIR board (and then I-V and analog SK filters). So how would that work? The ARM processor runs Linux and hence networkaudiod (NAA). I think that's even more KISS because there is no USB at all. I agree with and approve of this approach There is room for implementing simple buffers on the CPLD/FPGA itself as well. That would be in lieu of Linux/NAA if possible, if not, the most minimal Linux + NAA hardware + software config will have to do. The simpler the implementation near the filter, the better, which then leaves the PCB, noise profile/power supply and any remaining isolation if necessary (vast subject) to be optimised for our subject at hand. I checked the couple of boards you chose, and they look great, you'll definitely get a prototype going with these if you can code the FPGA. Dedicated Line DSD/DXD | Audirvana+ | iFi iDSD Nano | SET Tube Amp | Totem Mites Surround: VLC | M-Audio FastTrack Pro | Mac Opt | Panasonic SA-HE100 | Logitech Z623 DIY: SET Tube Amp | Low-Noise Linear Regulated Power Supply | USB, Power, Speaker Cables | Speaker Stands | Acoustic Panels Link to comment
YashN Posted March 15, 2016 Share Posted March 15, 2016 I can't replace those cables without tearing apart my apartment. Nothing can stop an audiophile. Dedicated Line DSD/DXD | Audirvana+ | iFi iDSD Nano | SET Tube Amp | Totem Mites Surround: VLC | M-Audio FastTrack Pro | Mac Opt | Panasonic SA-HE100 | Logitech Z623 DIY: SET Tube Amp | Low-Noise Linear Regulated Power Supply | USB, Power, Speaker Cables | Speaker Stands | Acoustic Panels Link to comment
YashN Posted March 15, 2016 Share Posted March 15, 2016 Would you expect any difference in sound quality between a) and b) ?a) the whole path between control PC and NAA is fiberoptic b) optical isolation is performed in the NAA+DAC unit and all cables are metallic There should be a big difference: any time you're transporting an analogue signal over long distances, you will lose signal integrity and SQ. The reason to use optical is that it is simple and digital even during transportation, so that the actual transmission is very robust (one of the main advantages of digital). So, any long distance => avoid protocols and implementations like USB which normally uses an analogue signal over a cable to encode the digital signal (unless you use a Corning optical USB). Instead, prefer optical. Dedicated Line DSD/DXD | Audirvana+ | iFi iDSD Nano | SET Tube Amp | Totem Mites Surround: VLC | M-Audio FastTrack Pro | Mac Opt | Panasonic SA-HE100 | Logitech Z623 DIY: SET Tube Amp | Low-Noise Linear Regulated Power Supply | USB, Power, Speaker Cables | Speaker Stands | Acoustic Panels Link to comment
YashN Posted March 15, 2016 Share Posted March 15, 2016 IThe described fiberoptic to DSD approach needs to be integrated into DAC. That's currently nonstandard solution, so you won't have wide range of DACs to choose. Doesn't matter: we won't need a DAC nor a DAC chip ideally, just a rather simple filter and a couple of additional simple circuits (buffer, flip-flops, IV or other). This is simple enough to build ourselves. Dedicated Line DSD/DXD | Audirvana+ | iFi iDSD Nano | SET Tube Amp | Totem Mites Surround: VLC | M-Audio FastTrack Pro | Mac Opt | Panasonic SA-HE100 | Logitech Z623 DIY: SET Tube Amp | Low-Noise Linear Regulated Power Supply | USB, Power, Speaker Cables | Speaker Stands | Acoustic Panels Link to comment
YashN Posted March 15, 2016 Share Posted March 15, 2016 with a Borbely inspired Pass inspired, EUVL cascoded jfet I-V and discrete jfet SK filters. Tubes for me, and ideally more than two DSD channels (already processed/filtered server-side) so that I can multi-amp at least 4 drivers directly from the amp (no cross-over in the speakers). Dedicated Line DSD/DXD | Audirvana+ | iFi iDSD Nano | SET Tube Amp | Totem Mites Surround: VLC | M-Audio FastTrack Pro | Mac Opt | Panasonic SA-HE100 | Logitech Z623 DIY: SET Tube Amp | Low-Noise Linear Regulated Power Supply | USB, Power, Speaker Cables | Speaker Stands | Acoustic Panels Link to comment
EuroDriver Posted March 15, 2016 Share Posted March 15, 2016 Originally posted by lmitche Ed, with the SGM and T+A DAC 8 DSD are you finding a need to enhance the USB chain? We tried the popular USB devices and found that with the SGM mobo power supply, which weighs over 8 Kgs, we could hear added noise from all of the ones we tried. So somewhat embarrassingly we have no outboard USB card or USB cable device Clarke's first law:When a distinguished but elderly scientist states that something is possible, he is almost certainly right. When he states that something is impossible, he is very probably wrong. Clarke's second law: The only way of discovering the limits of the possible is to venture a little way past them into the impossible. Clarke's third law: Any sufficiently advanced technology is indistinguishable from magic. In developing the SGM's components we have taken a very empirical approach. Trying stuff out, if it sounds better push further, do more. The problem is we have no solid theoretical basis for explaining the gains from applying analog power supply techniques and audiophile practice to a digital platform. However we can hear the results consistently, and repeatedly, and that's the sound of realism we want to share with the Computer Audiophile community Sound Test, Monaco Consultant to Sound Galleries Monaco, and Taiko Audio Holland e-mail [email protected] Link to comment
lmitche Posted March 15, 2016 Share Posted March 15, 2016 Thanks Ed. Pareto Audio aka nuckleheadaudio Link to comment
jabbr Posted March 15, 2016 Share Posted March 15, 2016 Tubes for me, and ideally more than two DSD channels (already processed/filtered server-side) so that I can multi-amp at least 4 drivers directly from the amp (no cross-over in the speakers). No problem. Filters can be implemented with tubes as can the I-V module, e.g. Broskie I-to-V. There should be no practical limit to the number of DSD channels beyond what the server side can process. I am already backed up building projects so I will leave the details of the tube version to you Custom room treatments for headphone users. Link to comment
YashN Posted March 15, 2016 Share Posted March 15, 2016 No problem. Filters can be implemented with tubes as can the I-V module, e.g. Broskie I-to-V. There should be no practical limit to the number of DSD channels beyond what the server side can process. I am already backed up building projects so I will leave the details of the tube version to you Hehe Dedicated Line DSD/DXD | Audirvana+ | iFi iDSD Nano | SET Tube Amp | Totem Mites Surround: VLC | M-Audio FastTrack Pro | Mac Opt | Panasonic SA-HE100 | Logitech Z623 DIY: SET Tube Amp | Low-Noise Linear Regulated Power Supply | USB, Power, Speaker Cables | Speaker Stands | Acoustic Panels Link to comment
4est Posted November 29, 2016 Author Share Posted November 29, 2016 I am not sure if I see a need for any active stages on the outputs of the shift registers. I am pretty enamored by using the Saber chip straight out to a VERY high quality resistor/transformer filter, and am thinking of the same for this balanced DSC1 I just assembled. They both are capable of pushing enough current AFAIK. At present, my TVC is acting as part of the filter. No problem. Filters can be implemented with tubes as can the I-V module, e.g. Broskie I-to-V. There should be no practical limit to the number of DSD channels beyond what the server side can process. I am already backed up building projects so I will leave the details of the tube version to you Forrest: Win10 i9 9900KS/GTX1060 HQPlayer4>Win10 NAA DSD>Pavel's DSC2.6>Bent Audio TAP> Parasound JC1>"Naked" Quad ESL63/Tannoy PS350B subs<100Hz Link to comment
andybob Posted December 13, 2016 Share Posted December 13, 2016 Hi all, Am thinking of building (oh alright, having my Chinese supplier build) the DSC1 in order to try out 512 DSD. Has anyone built it and fed it from an mR NAA ? Link to comment
shadowlight Posted December 13, 2016 Share Posted December 13, 2016 Does mR support 512 dsd? I thought it was limited to 128 in dope mode. Can you share the Chinese supplier who is build your dsc? Link to comment
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now