Jump to content


  • Content Count

  • Joined

  • Last visited

Everything posted by ray-dude

  1. My understanding is that the spirals are to tune resonant modes. From a DIY perspective, the accelerometers in our phones are pretty damn sensitive. When you have your servers properly mass loaded, may be useful in dialing in and tuning the bases.
  2. ray-dude

    HQ Player

    It can be hard to get started on HQP, just because all the choices can seem so overwhelming. Once you have a working setup that you like, much easier to tweak up different filters and settings to tune to your ear. As Rajiv and others have mentioned, with TT2 you want to target 16fs (768kHz for PCM, 44.1*256 for DSD). sinc-M + LNS15 will give you a sound profile closest to Hugo mScaler. Setting Vol Min and Vol Max to -3dB will force all volume control to be with your TT2, and give headroom for the upsampling so you don't get clipping. A lot of the rest of the parameters are dependent on your streamer and preference. There are MANY other Filter and Dither combinations to try. Once you get sinc-M going, try sinc-L and sinc-S (I find sinc-L to be quite nice) Once you start finding a preference, it gets easier because you can start exploring other filters within a family (MP vs LP filters, etc). I happen to be a hard core LP guy, but others really enjoy what MP brings to the party. I've found that the more your ear gets accustomed to what HQP is doing, it becomes a lot easier to dial it in to your taste and your equipment (alas, there is no optimum configuration, which is why there are so many options!)
  3. "I was a freedom rider When we thought the South had won Virginia in the spring of '61..."
  4. A thousand times this! When we finally unravel all the secret sauce, I suspect a huge part of it will stem from conceptualizing all of these as aspects of a single component in the design. I keep relearning basic metrology - what you measure/observe is only as good as the integrity of your weakest reference. Especially for digital systems, perfect reference power and timing, and a lot of "problems" just go away! And since we're sharing untold stories, somewhere in the garage I "unofficially" have a memory board from a Cray 1S. As a whipper snapper engineer, I learned so much from studying that board (and for programming for the Cray). The entire board design was about clock and power distribution, with a twist of cooling mixed in to keep it from (literally) melting. Complete brute force design, like a lot of Cray designs back in the day, but overwhelming brute force focused like a laser on the key things that mattered most. I was thinking a lot about that board as I studied Emile's design.
  5. The other wrinkle is Emile sharing that high quality clocks actually take away SQ in the Extreme (although he is understandably proprietary about his clock strategy). He is definitely taking the road less traveled when it comes to power supplies and clocking. That being said, that many rails of Sean Jacobs DC4 rails would be something to behold! I only have one Sean Jacobs supply (DC3'ish) but it was custom built/configured for my DAC, so I haven't had the opportunity to hear this bespoke supplies further back up the digital chain. Alas, I suspect that by the time my SR7 custom build gets here, we will have all moved on to Mr. Fusion power supplies
  6. Fantastic Nenon! I did similar back of the envelope math earlier this year, then added in a fudge factor because while I knew I would eventually get to the more premium end of the spectrum, I also knew I would feel compelled to travel the intermediate steps to get there. That's how I got to spitting distance of jumping on the Extreme train and skipping the intermediate 5 steps. It's amazing how much power (and to a lesser extent clock) dominates the calculus. We're basically building power supplies and clock generators with some electronics and storage attached. I continue to have a nagging tingle from my spidey sense that Emile's "mother of all capacitor banks" for filtering will have a LOT more running room on the power side of the equation than the incredible supplies from Sean and Paul et al. If one were to have (theoretically) perfect filtering capacity with an arbitrarily large and fast filter bank, your options on the power side of the bank REALLY open up (who cares if you have a noisy supply if the filter bank takes care of business?). From a DIY perspective, Sean certainly leans heavy into crazy large filters, but even he is an order of magnitude off what Emile is doing. Has anyone seen another power supply maker take a similar strategy?
  7. Thankfully it is Fathers Day, so all dad jokes get a pass (my kids are suffering today too i have the panzerholtz footers on my Extreme, but I did not look into footers for my Chord DAVE. that would be worth a quick listen (I can hear the impact of mechanical changes pretty quickly with the binaural tracks you’ve been sampling Rajiv...that holographic sound stage is very sensitive) Previously I had a stack of an Acoustic Revive platform to Black Ravioli pads to rubber footers to DAVE. When I added roller balls to get horizontal isolation, that had a huge impact, but the BR pads actually made things worse. I got best sound when the roller balls were rigidly coupled to the Acoustic Revive base, and my DAC was rigidly coupled to the horizontally isolated base. The Daiza by itself is pretty good. On my list is to revisit horizontal isolation with the Daiza, but I haven’t gotten there yet.
  8. i have my DAC on a Daiza as well. I have the body of the DAC resting on three small ceramic tiles (vs the rubber feet) to get that direct coupling. Alas, tuning a mechanical system can be an iterative process. For example the weight of the DAC will impact coupling and resonance frequencies as well, so weights on top of the DAC could also have an impact, depending on which vibrational frequencies are having an audible impact in the DAC. Conceptually, the goal is to drain vibrational energy away, and/or shift it to frequencies that are less impactful. For us EE types (vs ME types) it has been helpful for me to think of the mechanical system as a circuit, that is coupling through a transducer (capacitors, clocks, wires, etc) to the electrical system in the DAC
  9. Mark, my apologies! I have no idea how "Paul" settled into my brain, but certainly no disrespect was intended. Thank you Alex for the gentle correction (as well as the info that you shared)
  10. Paul, I have a couple different components that are SFP capable: The StarTech PEX1000SFP2 PCIe optical Network card in my Taiko Audio Extreme (2x) Ubiquiti EdgeRouter X SFP's Sonore opticalModule I recently sold my Uptone Audio EtherREGEN, but I was experimenting with the SFP port on the ER in my Taiko Audio SGM Extreme review (part 4 for those subjective findings) I currently have some TP Link multi-mode SFPs connecting my two EdgeRouter X SFPs. ERX1 is my router/gateway and connected to my ATT fiber ONT. It has a audio subnet that I connect by copper to my opticalModule. I run fiber from the opticalModule to my Extreme. I've run both the Planet Tech SFPs and Finisar SFPs on my audio net. I have the rest of my home network (including WiFi) on ERX2, galvanically isolated from ERX1. ERX2 is configured as a simple switch, but I do have a couple VLANs for my WiFI (guest network, IoT network, etc) With the Planet Tech SFPs, I preferred copper NIC on the Extreme (fiber to the opticalModule+PlanetTech to copper to the Extreme). With the Finisars, I'm back to preferring the optical NIC, esp. now that I have my audio network isolated from my home network, but it isn't a clear preference. FWIW, subjectively, reduced network traffic to the my music server NIC does have an impact. This is contributing to my hypothesis of laser switching-induced electrical noise being the mechanism in play. All of the above is with streaming TIDAL content to Roon to HQP to DAC. The network part of my system is very much in flux right now, because of what I've been hearing. There are a couple of permutations I'd like to try to see if it helps develop a hypothesis for a what a core mechanism could be. Here is my wish list for things to try if I had proper equipment in quarantine with me: * Probe power rails on the SFP modules and look at spectrum of noise generated on same by network traffic. If that spectrum is the same, my hypothesis can filed in the "oh well" cabinet. If different, then the question shifts to whether that difference is audible (and if so, how). * Build an extender and wire to SFP cage outside of NIC/FMC/Routers/etc. Rig so that I can swap between powering SFP externally or via the cage leads. If there is variability with different class power supplies, that would generally support the laser-switching-noise hypothesis. If not, back to the "oh well" cabinet we go. * Get access to OEM data sheets for SFP modules and find the lowest power consuming unit available...give it a listen, get another subjective data point @Superdad without asking you to divulge anything proprietary, did you or John look into any of these things when you were designing and prototyping the ER? (and if that is poking at things that you consider proprietary, my apologies in advance)
  11. FWIW, at the engineering schools I went to the reaction would have been "No way! Let me hear that... Huh... Wow... What the hell is going on? Cool, let's try to figure this out!" This is the exciting stuff! I hope that is the spirit of this thread. Alas, I have very limited access to measurement equipment right now, so I'm limited to offering (somewhat) informed hypotheses and experiences, and feedback on any experiments or interpretation of results that folks do. Wish I could do more, since given what I'm hearing, this is a pretty compelling topic for me.
  12. Thank's for the pointer to this topic Ken. My hypothesis is that switching to the laser is causing a correlated pulsing on the reference voltage and/or ground planes that is having an indirect impact on the DAC. My basis for this is that in previous experiments with coax SPDIF signals to my Chord DAVE, that particular DAC seemed to have a strong susceptibility to parasitic signals in the 2-3GHz band. Copious ferrites on the dual coax connectors to the DAC had a surprising benefit to SQ. IIRC Rob Watts (designer of the DAC) attributed the impact to noise floor modulation in the DAC (I would have look up some old notes to confirm). It is easy to replicate this by having a WiFi basestation configured for 2.4Ghz and bringing it close to this DAC, or having music server on WiFi that is close to the DAC. All above is hypothesis or experiential.
  13. Even for mundane Cat8 cables from Monoprice and Cable Matters, I found that they needed ~80-100 hours before the SQ with my old EtherREGEN settled out. Aside from shock that ethernet cables were audible at all, the fact that they changed over the first couple days was a definite punch to the stomach. The impact was visceral though (literally...the better the cable, more the visceral bass became, and the more tangible space became) I have no hypothesis for break-in, but now I am a true believer in the impact of ethernet cables (running the amazing Sablon Audio 2020 ethernet cable for all the copper in my digital audio signal path) (and thank you to Alex and Chris and all the contributors here for creating the community space where real discussion and exploration can happen)
  14. Emile has written about Linux vs Windows before. His opinion is that he is able to strip/tune Windows to better SQ than he able to get on Linux (and certainly Linux wold be easier to lock and support than Windows...this was a SQ decision). That being said, Euphony has demonstrated that it is possible to get to very good SQ on the Linux side, and their progress over the past year has been most impressive. I often lament that Zelko and team are diluting their (considerable) efforts between SQ (core Euphony) and user experience. I would MUCH prefer that the open up the APIs to the core, and count on the community to advance the user experience, and double down on SQ and the core. As far as weight, I was a bit put back by same, but as I highlighted in Part 2, there seems to have been an engineering/design driven objective for all the big casework decisions (for example, using copper for the CPU heat fin, or the thickness of the aluminum with the ventilation holes since they are designed as waveguides) The extravagance of the casework seems to all be in service of passive cooling, vibration control, and RF control. Of course, there is a decision to be made in any build whether one needs that level of CPU, etc. As a practical matter, I did not enjoying moving and positioning the Extreme at all. With a bit planning and furniture sliders and a lot of huffing and puffing (for the last bit, I literally had to use a crow bar!), I was able to do so without injury, but my motivation to ever move it again is zero. (as an aside, I'm not terribly happy to be back in a Windows environment after 13+ years...I would be VERY happy if Emile is able to figure out how to get Linux to the same level, and I suspect from a support perspective, the Taiko folks would be too)
  15. This will most definitely be my album of the evening June 26! (a new Khruangbin album definitely is welcome news for a Monday!)
  16. Good for you! Can wait to hear what you hear (and hope it doesn't make too much more antsy for my SR7 )
  17. I think it is inevitable that Rob Watts (designer of the DAVE) adds digital equalization to his DACs at some point, but alas, Chord DACs currently do not have this feature. The full signal is sent from each output. With the 9.87 Pi units, they have a natural fall off around 40Hz. The Pi bass units have an active control for gain, low pass frequency, and phase (with a couple other things). I tune these to blend with the natural response of the 4D drivers. The good news is that there are no cross overs on the DAVE to 4D driver path (direct wire connection from DAVE to driver), which gives maximum transparency over nearly all the frequency range.
  18. What's to debate? The speed and stability of the PH supplies is outstanding. You should be able to hear it quite well on the analog outputs stage of DAVE. I'm am (VERY) patiently waiting for my Paul Hynes SR7 to work its way to the front of the custom build queue, so I'm sticking with my Sean Jacobs DC3 supply for now, but I would love to hear a PH supply on the DAVE.
  19. SFP+ is definitely on my "try it" list, but it is a big commitment to move to a 10G network end to end. I strongly suspect there is a similar speed vs power/noise tradeoff with networks as a bunch of us have seen with CPU - faster you drive the system, the better, until you hit a point where it becomes worse. If one puts in better power, that peak in SQ moves to higher performance. Fortunately, it is a lot easier to power a NIC/switch/router than it is to power a server. I do use MM to connect my two EdgeRouters (1m cable). I'll pick up a longer MM fiber and give it a try vs my SM Finisars and Planet Techs.
  20. I have not tested an MM transceivers yet. Have you seen spec sheets where they are less than 500mW consumption? (best I've seen/heard for SM). Alas, we're still in the empirical stages for these transceivers...try them and see if they sound better, then look at specs and try to figure out why things may be different. Related to the network use comment, there is a definite SQ win in isolating my audio network from the rest of my home network. Having my server on an isolated LAN (no broadcast or other traffic from my home network) definitely has a positive effect. Back when I was fiddling with SqueezeLite endpoint buffers, there was also a huge win in increasing the input buffers so that tracks were preloaded into the buffer within a second or so. These are all positive correlations (for me), but nothing conclusive to hang on yet.
  21. If the root cause is induced noise to power/ground plane, things are more insidious. A faster interface would have faster switching components, which could be generating a different noise profile. Depending on the sensitivity of the rest of the system, moving the noise could help or it could hurt SQ. Other than the truism that less noise is better, noise related SQ impact is devilishly difficult to rationalize or connect the dots on (so system/environment dependent).
  22. The correlation I have seen for SFPs seems to be more related to power consumption of the module. I suspect the mechanism is some induced noise from the transceiver. Generally speaking, the lower the power consumption, the better the sound. That being said, the different categories of SFPs (laser type, speed of interface, etc) seem to have different power profiles, so the correlation extends to those factors, but my spidey-sense (gut) says power consumption may be the key dimension for sound quality
  23. This is crazy cool! Alas, I don't think the reconstruction math works that way. Each Mojo is applying 26k taps (~11 bit reconstruction accuracy) to 2 stereo channels. I presume the compute is there to apply to a single channel, but it is not built to do that. The output of each Mojo would still be 26k tap output. For reference, here are the tap lengths for various Chord DACs: Second column reflects the bit accuracy (vs a pure sinc function) of the WTA reconstruction filter. Remember, an infinite length sinc filter will perfectly reconstruct a bandwidth limited (non-impulse function) digitized signal (https://en.wikipedia.org/wiki/Whittaker–Shannon_interpolation_formula ). As the number of taps in Chord DACs increase, the closer and closer the coefficients in Rob's proprietary WTA filter approach an idealized sinc function. For Mojo, they are identical to 11 bits. For mScaler with 1M taps, the WTA coefficients diverge from ideal sinc for coefficients smaller than 16.6 bits. Paradoxically, the more taps Rob is able to get into his DACs, the less secret sauce he has in the reconstruction part of his pipeline. That being said, noise shaping and the output stage have a HUGE impact on SQ (taps only get you so far, you need to do something with that reconstructed signal). Listening to each of these DACs with a 1M tap reconstructed signal highlights how hugely impactful the other pieces are. Taps are a convenient way to differentiate things from a marketing perspective, but I would rather have a DAVE with 20 element pulse array output and 164k taps than a Hugo2 with 10 element pulse array with 1M taps. The Mojo has a 4 element pulse array in the d to a section of the DAC. That is also not additive, unless the DACs parallelize the output so that each Mojo is covering a different part of the signal. You're basically adding 4 identical signals, not having 4 parallel streams that get combined to 4x the resolution (by analogy, there is a big difference between 15 years of experience, and 1 year of experience repeated 15 times). Even if we had access to the FPGA code to try and parallelize the output to get some sort of multiplex additive benefit, the timing precision required to analog multiplex together 4 Mojo's at the performance levels Rob shoots for would be quite something.
  24. I think you'll be shocked at how much tension goes away with a quality supply on the 5V (RF is a bugger). An alternative is to use the stock DAVE SMPS for the 5V and your Keces for the +-15V. The stock supply is quite good, but a Sean Jacobs or Paul Hynes is a LOT better.
  25. As a former Chord Blu2 owner, I think they would have a killer transport with a modest quality CD transport, robust casework, USB in, ethernet in, USB out, and running embedded HQP for upsampling (PCM to PCM sinc-M could be done very cheaply from a compute budget). It would sound extraordinary, and be end game for 99% of the people out there.
  • Create New...