Jump to content

romaz

  • Posts

    749
  • Joined

  • Last visited

  • Country

    United States

Retained

  • Member Title
    Sophomore Member

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. You're absolutely correct, @Dev, and you bring up good points. My assessment is based on the sum of the parts of this build and so I have no way of knowing how good the motherboard by itself is. To be honest, I wasn't expecting much based on the fact that the motherboard was being powered by an inexpensive Seasonic ATX SMPS and that the OS was a standard Windows 10 installation and so based on tempered expectations, I was pleasantly surprised by how good the unit as a whole sounded. Yes, as you know, clocking does nothing for dynamics. When given the option, I specifically asked for the most powerful CPU that Lee could install because even for playback of NOS Redbook files, I find low-power CPUs to sound thin and lifeless in comparison (although you don't realize it until you compare). It had been some time since I had compared a high-power CPU powered by a cheap SMPS against a low-power CPU powered by a really good LPS and so this test re-affirmed for me that you can only do so much with a low-power CPU in a music server. As for the benefits of low phase noise clocking in a music server, as you know, generally, it's a cleaner sound. Sometimes the presentation is more expansive, sometimes more detailed, but there are times when the presentation can also sound a bit mechanical or sterile and so replacing the CPU clock has not always been a win for me. In this particular case, would the positive impact of clocking be enough to outweigh the potential negatives along with the added noise brought forth by the SMPS? Again, I can only comment on the presentation of the sum of the parts of this prototype but against the well-powered, well-tuned NUC and InnuOS, the SOtM prototype not only sounded clean and clear but also tonally richer and more musically satisfying. I would not have guessed this PC was powered by a noisy high-impedance SMPS. So as you suggested, others will have to report on their own experiences with this new motherboard against an off-the-shelf ASUS or Gigabyte with the CPU, PSU, and other peripherals held constant before you truly know how good this board is but based on my brief experience with the prototype, knowing that this board can utilize a CPU with a reasonably high TDP and has multiple PCIe slots, I think this board is definitely worth exploring further. The board I received had no official support for Xeons or ECC memory.
  2. Exactly! I was hoping to try it with the Taiko DIY ATX but I had no ready access to it at the time I had this board.
  3. Yes, there is considerable potential with this board. Could be the missing piece for DIY.
  4. Excellent review as always, Chris! While at The High End Show in Munich in 2018, I had a discussion with Lee, the lead engineer at SOtM, about designing his own motherboard. I had explored this myself with several motherboard manufacturers and I knew it was possible. During the early summer of 2021, he wrote to me and told me he finally did and it was called the sMB-Q370. He sent me an early prototype in June of 2021 as he wanted my opinion of it. He gave me the option of CPUs and I asked for the most powerful CPU possible, preferably an i9, and that is what I got, a 9900K with a max TDP of 95w. While there is more to it than the clock, indeed one of the special features of this motherboard was that it was now being clocked by SOtM's highly regarded sCLK-OCX10 and this clock was sent to me as well along with one of SOtM's highly regarded clock cables. The unit came as a preconfigured system with SOtM's tX-USBexp PCIe USB card and sNI-1G LAN card. These cards and the sCLK-OCX10 were powered by individual sPS-500 PSUs that SOtM provided. The motherboard was powered by an inexpensive ATX SMPS. Lee gave me the option of operating systems and I selected Windows 10. I don't believe Lee optimized the OS or the BIOS in any special way. This was after all an early prototype. I compared this turnkey setup against 3 different servers in 3 different systems: a stock i7-based NUC powered by a Paul Hynes SR7 running Euphony OS, an InnuOS Zen Mk3 (MSRP $2900) which uses a lower power Pentium CPU, and my $25k SGM Extreme by Taiko Audio. In each case, I used Roon for software playback. Compared against the i7 NUC which was well powered by an SR7, the SOtM server, even with the inexpensive ATX SMPS, sounded more dynamic with better transient response. It also sounded fuller while the NUC sounded thinner and more threadbare in comparison. The results were pretty much the same against the InnuOS server. The signature of the SOtM setup was also very clean and clear with very high resolution but it was the bass definition that especially stood out compared to the NUC and InnuOS. To my ears and to the ears of my entire group, the SOtM server outclassed these other servers. In my opinion, this obviously had a lot to do with the high-level clocking that SOtM is well-known for but also the fact that this unit had a much more robust CPU. Compared against the SGM Extreme, not surprisingly, it wasn't really that close. The Extreme sounded better in every way. It sounded faster with even better resolution and dynamics. There was more air with better separation and had a much lower noise floor but it's not really fair to compare an early prototype with a lowly SMPS against a server with dual Xeon CPUs and their 20 cores, a very tightly tuned OS, and a superior ultra low impedance PSU. But if you take the core elements of this server, specifically the sMB-Q370 motherboard, tX-USBexp, and sNI-1G, tie them together with the sCLK-OCXO10 and power everything properly, I think there is wonderful potential here. The Extreme is a tough match up for any DIY server and is probably an unrealistic benchmark but what I heard with SOtM's prototype was SERIOUSLY good!
  5. But for all out "low impedance", it looks like the Dean's ULT might be the best: 0.13 milliohms is nearly an order of magnitude less impedance than the GX-16.
  6. Yes, very much interested to know your findings. With the giant LiFePO4 batteries I have here, without spending a lot of money, you can buy these 6awg DC cables terminated with ring terminals on one end (to connect to the battery) and PowerPole connectors on the other end to connect to the power supply. These connectors are genderless and very securely snap on. They can also be crimped instead of soldered for a more secure fit without the coloration that soldering adds. https://www.impulseelectronics.com/power-cable-75-amp-powerpole-to-eyelet-terminal/ https://powerwerx.com/5900-bk-anderson-power-6awg
  7. Boulder amps use 250V 32A connectors and that connector is just massive: Even though the Neutrik PowerCon 32A connectors can pass 32A of 250V AC, the contact resistance of that connector is only still about 3 milliohms and so with regards to contact resistance for low voltage power supplies like the ARC6, this connector is not really an improvement over the GX-16:
  8. According to Paul Hynes: Gold-plated DIN connector: 5 milliohms impedance Silver-plated XLR connector: 3 milliohms Jaeger: <1 milliohm According to Sean Jacobs and Farad: GX-16: 3 milliohms With the GX16 4-pin connectors, if 2 pins are paralleled, you can cut impedance in half to about 1.5 milliohms. Neutrik Speakon connector (according to Neutrik): ≤3 milliohms (https://www.neutrik.com/en/product/nl4fc) So the Speakon not really better than GX-16. The GX-16 connectors currently being used in the ARC6 prototype are sourced in the U.K. and are 10x the cost of the ones you can buy from China. A potential option yet to be explored are the Anderson PowerPole connectors which report a contact resistance of only 0.6 milliohms: https://powerwerx.com/anderson-261g2-powerpole-contact-pp45 https://www.acerracing.com/collections/powerpole-connectors For reference, the resistance of 16awg copper DC wire is about 4 milliohms per foot. Barrel connectors have a contact resistance of about 30 milliohms each! If you're using 1 meter of DC wire with barrel connectors on each end, then your ultra low impedance PSU is no longer ultra low impedance. With so much effort to get the ARC6 to sub 1-milliohm levels, you potentially throw away the effort if you use long DC cabling and bad connectors and so while the connectors on the final spec ARC6 DC4 has yet to be established, the GX-16 connectors are by far not your worst option.
  9. The Taiko GaN DC-ATX converter can handle any DC input between 16-40V and so as long as it sees voltage within this range, regulation is not important.
  10. Emile tried powering an Extreme using his GaN DC-ATX converter and a giant 24V battery and the results were not positive. Personally, I think I am done with batteries for directly powering digital components. As for a GaN-based inverter, I have not seen one. The problem with DC-to-AC inversion is that the DSP involved introduces distortion, sometimes >10%. I would not automatically assume, for example, that a giant Tesla 13.5kWh Powerwall battery that people can buy for their homes will sound better than AC power from the grid because the inverters used are not audiophile quality. The best off-the-shelf inverters I have looked at, the distortion ratings are around 2% but those are lower power (sub 2500 watt inverters). The very expensive Stromtank S2500 that I tried claims a line distortion of about 2% but the problem I had with that battery-powered inverter is that transients were soft. You really have to use a massive array of paralleled cells to achieve very low output impedance for this to sound good but the expense goes up and so does the size of the unit.
  11. Yes, lots of possible options for DC wire but those high-impedance barrel connectors are instant turn offs and make the whole setup look cheap. I think the connectors are potentially more damaging than the wire.
  12. When we were dealing with batteries that could discharge up to 1000A and 6AWG DC wire between the batteries and the DC4, we looked at a variety of connectors besides the Jaeger. @Nenon liked this one from Amphenol Industrial but we never got around to trying it. With the ARC6, the only connectors we have to worry about are with the umbilical that connects the ARC6 to DAVE but the limitation is the Molex connector which cannot accommodate anything bigger than 15.5AWG wire and so the ARC6 prototype that I have is using the GX16 connectors. Among the worst connectors are the barrel connectors. Those are high impedance.
  13. The cable capacitance you're talking about is a different story. As Dana himself is fond of saying, unlike a power supply, a cable is a passive device and its only job is to get out of the way. A perfect cable that will do no harm will have zero resistance/impedance, capacitance, and inductance but the only cable that would approach this would be a super conductor. Since super conductors also require super cooling (to zero degrees Kelvin), this impracticality makes all audio cables inherently imperfect and so all cables will add something to the signal, whether it be digital or analog. Reducing resistance is fairly easy, just use a large enough conductor. Inductance and capacitance are tougher to deal with and tend to be a zero sum game where construction techniques that decrease one will increase the other. The trick is to balance them well and it seems this is where cable manufacturers get creative. Adding capacitance can sometimes add a warmth or softness that is desired and so playing with these parameters could be useful for sound tuning. With the giant capacitor arrays used in the ARC6, the goal is different. Used pre-regulator, these giant capacitor arrays serve as energy reservoirs to make sure the regulators are never starved of current. This would be one way to effectively lower output impedance.
  14. Steve, the Neotech single-crystal copper that Sean uses as his standard wire is a very competent wire and if it's all you heard with an ARC6, I'm pretty sure you will be pleased and wonder if you really need anything better. This is how I felt. But then @Nenon lent me his Mundorf Silver/Gold (MSG) umbilical and that proved to be an expensive mistake because once I heard it, it was difficult not to go with it. Performance-wise, it offers a bit better resolution but it was really the tonal qualities of this wire that I like. Tonally, the sound is more intense and more alive without sounding too forward or fatiguing. Unlike some silver wire, it does not sound bright or thin and, in fact, it portrays timbres beautifully and accurately (to my ears) but if you're looking for relaxed and laid back, stick with the Neotech. One word of caution, MSG wire is not cheap and so it is not a high value upgrade. My suggestion is if you already own these Shunyata cables, go with what you have and see what you think. I have not experimented with power cords yet for the ARC6 and while I'm confident they'll make a difference, for now, this inexpensive (I paid $200 USD for it years ago), unfiltered TG SLVR power cord (hand built by the late Bob Crump) that I'm using on the ARC6 is very satisfying.
×
×
  • Create New...