CG

This ^^ is the most sensible thing I've read this week. 👏 Wouldn't there also be some value in shorter sweep times but lots and lots of repeats to capture possible random events? There may be none, but you never know until you look...

I think there's more to it than just that... The error correction and encoding schemes are pretty sophisticated, too. However, I get your point. In general, copper Ethernet sure is convenient and sure is cheap. But, it creates that storm sewer situation I suggested before. Somewhat OT: For better isolation, it's often worth trying battery powering system components. Not entirely convenient, but it sure is great for isolation and makes for an interesting test. (I'll keep my own results of this to myself.) A common opinion is that batteries often spoil this sound aspect or that. One way of looking at that is that perhaps the designers of the gear that you (generic "you") compared as powered by the AC mains versus by a battery may have actually applied the effects of dirty power and common mode loops to create the sound they liked. (See "diode switching noise due to reverse recovery characteristics" as an example.) Is it any wonder that system performance is all over the place?

I suspect a solution like this is better suited for network connected devices via Ethernet. One of the problems that network connected devices have is that the network is usually not only physically large, but has many, many connections. Connections not only to the network, but to other devices both through network connections but also the AC mains. That makes for very large and complex current loops that are not only hard-wired together, but couple electromagnetically. Kind of like a storm sewer. Think of what happens when it rains. Except, instead of water and junk, you have the desired signal flow and noise all mixed together. Ethernet transformers help isolate some of this, but not as well as everybody would like. There's a reason that error correction technology was built into the Ethernet protocol in the first place. Anyway, if you have a direct USB connection there's other ways to isolate the DAC that aren't as complex as your solution, because the loops are just just smaller overall and better defined. Not that much attention is really paid to this, you probably understand... 🤷‍♂️

Who cares? Since you're using DSD in this design, the "converter" could be a simple logic device, followed by filtering and some suitable analog amplification and/or buffering. Presumably you could use an FPGA large enough that it would also do all the digital filtering and other processes. You could even add a USB interface to the FPGA for the people who don't feel the need to be network connected. Even SPDIF could be accommodated, right? In principle you'd end up with something like a PS Audio Direct Stream DAC that has maximum isolation between the digital and analog domains. Perhaps you should give them a call... One other comment - the SFP's I've used haven't all been entirely RFI/EMI tight. You'd probably have to take that into account on the converter end. Oops! One more. There's no reason that this needs to be limited to DSD conversion. (I mean the actual converter, not the original file format.) You could as easily use it with other types of converters, too. You sort of imply that, but I just wanted to state the obvious. 🤨

OK. No need to be condescending. (How you were able to determine that to your own satisfaction without those many hours of consulting engineering time is something that I probably couldn't understand even if you explained that aspect of it to me. Please don't tell my boss...) Please note that I didn't even slightly, vaguely hint at whether these cables could change the sound of an audio system. I was just pointing out an example of an engineer from a generally accepted to be good cable company providing a mathematical analysis of what he thinks is the systemic aspect of cabling. Complete with measurements. Isn't that what objectivists usually ask for? So far here's what I've seen here on this "objectivist" forum. Many people are quite willing to dismiss most anything they want to with a brush of their hand. No details, no measurements, no reason. Many people, some of whom happen to be members of the above group, also demand complete juried papers with mathematical derivations and extensive measurements with double blind testing before they dismiss whatever it is with a brush of their hand. This is crazy. At least, to me. But, I won't criticize people if that's their way of thinking. However, this is not for me. I prefer to investigate things before I dismiss them. Or accept them. Note to Chris: I know I may have deviated in this post from the central purpose of this sub-forum. I apologize for that. So, feel free to delete it. I can dismiss myself with a brush of my own hand. Thanks.

Please explain this with some details.

https://www.iconoclastcable.com/story/index.htm Note the links to objective analyses. Note that this a Belden guy. Note that these cables are sold by Blue Jeans. No opinions from me - just passing this along.

This is supposed to be a sub-forum based on objectivity, isn't it? That kind of implies keeping the facts straight. Arm waving isn't allowed. First, RG-59 cable loss is specified for matched source and load impedances. In this case, 75 Ohms, which we pretty much never have in an audio system. The source impedance might be close enough to 75 Ohms, but the load impedance of an amplifier or preamp input is usually much, much higher. Starting at around 10K Ohms, typically. (Many input impedances drop as the frequency rises due to the use of an input filter, various compensation devices, and the actual capacitance of the input devices. But, let's ignore that, for now.) In addition, the loss is rarely measured for these kinds of cables at very low frequencies. Instead, what you get is extrapolated data from the loss curves generated where the loss is more significant. But, if you trust this data, the loss of RG-59 cable at 20 KHz is insignificantly different than it is at 20 Hz. As in, a couple milli-dB for a 10 foot length, for matched source and load impedances. At least according to the cable manufacturers. OK, perhaps this isn't an accurate portrayal. So, why not just consider this short cable to be a simple series resistance and inductance lumped component with the cable capacitance lumped as a shunt element at the load end? It's easy enough to simulate. Oliver Heaviside thought so. When you do this, you get similar results. Try it for yourself. In fact, according to the simulation, the amplitude actually can rise a milli-dB or two at 20 KHz compared to 20 Hz, depending on the source and load impedances. Or, drop, again depending on the source and load impedances. The reasons for this are an exercise left for the reader. (Note: This simple analysis does NOT include the effects of Eddy/Foucault's currents, subsequent skin effects, or a bunch of other details. That might or might not be significant in a cable this length at these frequencies.) But, don't trust me - verify this for yourself. While there, you may want to examine the square wave response for such a network. So, the cable loss, assuming dielectric material that is not intentionally conductive or lossy and ignoring those other little details that may or may not matter, probably isn't an issue. However, as just one aspect of the system analysis, you might want to consider the effect of this lumped RLC on the behavior of the output circuit of the source. And, on the input circuit of the load (preamp or power amplifier.) And on the feedback system of the power amplifier, assuming it uses overall loop feedback - most do.

Two things I meant to post but neglected to. One is a schematic of a simulated loudspeaker. This is the one used in the Stereophile amplifier measurements: https://www.stereophile.com/content/real-life-measurements-page-2 Obviously, different loudspeakers will be different and therefore present different impedances. Plus, it doesn't include various non-linear effects. It's arbitrary, but so what? If you really want to, you can derive the impedance over frequency for any loudspeaker reviewed in Stereophile by examining the impedance plot in the Measurements section of the review. These only go out to 50 KHz, which may or may be important - you decide that. Second is an example of some parameters for a speaker cable. This is a popular cable sold by Blue Jeans cable: https://catalog.belden.com/techdata/EN/5T00UP_techdata.pdf Of course, other cables will have different parameters. It's also easy to break the cable up into shorter sections and add various passive RLC networks at the junction points to see how that affects the results. That would bring us around to the discussion of MIT cables. See for yourself whether they are peddling "snake oil", whether there's something to their ideas, or whether there might be other solutions to the problem statement, whatever that problem statement might be. (Full disclosure - I don't think I've ever actually seen or heard MIT or Transparent cables myself. At least, not to my knowledge. Perhaps at a show, back when I'd go to one on occasion. If I was exposed to them, they didn't leave me with any impression either positive or negative. And, finally, my day job is not in audio - it's in the telecommunications equipment business.) Have fun!

I don't think you even have to drive it harder. There's always perturbations that affect things. If you measure circuits by themselves in ideal lab conditions rather than in a system (by system, I mean a collection of gear intended to work together) and average a lot of samples, you don't always get the full picture. Or, you get a distorted picture. However, this is getting off the topic of this thread, which I don't want to do. There should be adequate information above to give people a starting point for analyzing various conditions that might explain why different cables might cause system sounds to be different. Or, not different. As they always say, that's an exercise left for the reader.

One more thing: Please overlook some of the obvious editing and proofreading errors in the posts above. My mind was trying to put information out far faster than I could compose and type it. (I also learned the statute of limitations on being able to edit one's own postings...) So, if I look like a complete boob, have a laugh. It's on me. My hope is that everybody who wants to can investigate this and other topics for themselves. That way you don't need to wade through the dogma and personal beliefs that everybody has. At least everybody who likes to post on the internet. So, teach yourself to fish.

Finally, want to measure this at home? Here's gadget that is a multi-purpose tool. No need to repeat all the specs here, but I'll just say that with the exception of some very specific audio related measurements, this box will let you make most measurements you'll need. With some add-ons, you can also measure impedance pretty effectively out to at least 10 MHz as well as measure gain and phase margin for amplifiers using closed loop loop feedback. https://reference.digilentinc.com/reference/instrumentation/analog-discovery-2/start If you want make those specific audio measurements, like harmonic distortion spectra, this will do the job for you: https://quantasylum.com/collections/frontpage/products/qa401-audio-analyzer You can certainly use a decent quality A-D/D-A box with suitable software. I like the Focusrite Scarlett 2i2 along with Amici (http://www.w7ay.net/site/Applications/Amici/index.html). But, when you use a system like this, you have to remember a couple things. One is that the gain controls are all highly variable. That's great when you're setting the recording level of your daughter playing the bass guitar. It's not so great for repeatable measurements. Second, the input sections are pretty sensitive and aren't tolerant of the occasional screw-up by the operator. Finally, you're pretty much held hostage by the audio portion of the operating system, which isn't designed for making audio measurements. Things may or may not work as you think they might. Or, just buy an Audio Precision APx555 B with all the trimmings. Hey - maybe you can then start your own web site, too. Seriously, this is the state of the art. But, you might not need that to answer your own questions at home. Here's something to keep in mind. Back in the 80's, and even the 90's to a large degree, the people designing and manufacturing audio equipment even at the highest end did not have the capabilities I just outlined. Computing gain and phase margin for an amplifier was done by using Bode plots, with lots of assumptions. If you measured margin, you used an oscilloscope and a sweep generator. Most audio labs did not have a spectrum analyzer, never mind one that could look out beyond 10 MHz in the land where oscillations often dwell. The design guys had very crude simulators that cost a bundle and required expensive computers. Most audio companies plain couldn't afford either. LTSPICE on a modern computer completely blows whatever capabilities they had completely away.

If you're already good with the basic stuff, here's some more specific information: https://www.analog.com/en/education/education-library/op-amp-applications-handbook.html There's more on that page, too. These two are specific to why amplifiers used to drive imperfect non-restive loads might be unstable: http://audioworkshop.org/downloads/AMPLIFIERS_OSCILLATION_BJT_CIRCUITS.pdf https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved=2ahUKEwiT-5ypvuPoAhXknuAKHXiqBSsQFjAAegQIAhAB&url=http%3A%2F%2Fwww.eevblog.com%2Fforum%2Fbeginners%2Fnoise-on-emitter-follower%2F%3Faction%3Ddlattach%3Battach%3D348962&usg=AOvVaw0YDC6j-OdvVTKAGkZ8yjzs (Some interesting comments are here: https://www.hifisystemcomponents.com/forum/1970s-design-indulgence_topic4543_post60560.html#top) OK, now that you understand that, perhaps you want to simulate what you've just investigated. Personally, I'm a fan of LTSPICE: https://www.analog.com/en/design-center/design-tools-and-calculators/ltspice-simulator.html It's free, well supported, used by a zillion people, and Cordell's book has several chapters on using LTSPICE. You can pretty much learn all you need to know about using LTSPICE for audio applications from Cordell's book: http://www.cordellaudio.com/book/

First, the general background information: If you generally don't feel confident in your electronics background, here is a terrific place to start: https://artofelectronics.net This covers A LOT. It's not all high level math, either. The most advanced are topics covered in high school algebra. (If you want to brush up on that, try here: https://www.khanacademy.org) More specific to audio is this book: http://www.cordellaudio.com/book/ Bob Cordell has been playing with audio gear for a few decades as his hobby. His day job was at very well telecommunications company working on communications systems. He covers amplifiers as well as the use of simulations and measurements.

These topics always fascinate me. You have people who swear that they subjectively can tell the difference between cables. That's not for discussion here. You also have people who are looking for proof that cables and networks possibly associated with cables can matter or that they definitely do not matter. Or, something in between. Finally, you have people who just dismiss the whole subject. I'd like to address that middle group. I'm not knowledgeable enough to discuss blind testing as a way to settle this. So, I won't. But, it's also possible to really dig into the physics and engineering to see what is real, what might be fake, and to what degree. Starting from a different point of view, I wondered if the actual cable characteristics might be affecting the performance characteristics of the associated power amplifier. After all, at audio frequencies you can describe the very simple model of a cable being a series resistor resistor, a series inductor, and a parallel capacitor. The reason you can do this, if you ignore everything to do with skin depth, dielectric and conductor characteristics, and so on is because an electrical wavelength in free space at 100 KHz is 9840 feet long. Most of us don't have speaker cables anywhere near this long. So, the first order approximation of a speaker cable is that simple lumped model. If you find differences in the amplifier performance with the simple model, it isn't going to get less complicated as you dig deeper. So, I started to collect simulations I'd made, build new ones, and redo measurements I'd made previously. Like a lot of people, I don't keep all those measurements around for later discussion. Why bother? So, back to the bench. The enormity of that task and documenting it all sounded like real work. Even though, like a lot of people, I'm pretty much confined to quarters for a while anyway, going through all that didn't seem like fun. Plus, I don't feel as if it would be very productive. Stepping back, I realized that objective discussion requires pretty much everybody involved to have a decent background in the subject and familiarity with the analysis, analysis tools, measurements, and measurement tools. It's not reasonable to expect people on a consumer audio forum to have all that. (Or, is it? If they want to have objective opinions, having some knowledge in the subject area seems reasonable to expect.) So, instead, I collected a few links pertaining to the subject that people can use for specific information. In addition, there's some links to suitable test gear that will get you a long way on measuring this yourself at home. It's not free, but it isn't in the thousands of dollars or more, either. Finally, there's some links to suitable reading material that gives a wider background on the subject. Hopefully this will be of some value. The subject matter here generally is the stability of amplifiers when driving low impedance loads that are not purely resistive. My suggestion would be to read up on that topic. Then, you can build a simulation of various cable schemes and their effect on the amplifier performance. If you want to make measurements, you can do that as well. I'll break this up into a couple section to keep this already way too long post shorter.