sonicweld

Steve, Naturally, the most pertinent comparisons between our products would be between currently shipping versions, but given that older versions of both are still available on the used market, I think comparisons are still relevant; the older model discontinuations are in the very recent past and are hardly ancient. I can't speak to the upgradeability of any other manufacturer's devices, but I want to point out that the older Diverters (specifically, any with a BNC output) are fully upgradeable to the new version, which includes both better clocking as well as better power supplies. I took great pains to make the new version 100% mechanically compatible with the original chassis design, which was quite difficult given that the new design is considerably more dense than the old. Josh

Hi XAHTP, Regarding your comment about the Diverter being supplied by "dirty USB power" - I wanted to note that I've posted extensively about this subject in the past, both in this forum and others, so I won't burden everyone with a redundant response here. Anyone should be able to find the past threads without too much trouble. Of course if you've already read what I've written and have new points of discussion to bring to light, I'll do my best to respond to those. As for being "secretive about the innards of the device" - what sort of disclosure would you like to see about the design, ideally speaking? I hope it is obvious that a response to this question like "a full schematic, BOM, and gerbers" isn't realistic. I've posted quite a lot about the design as well, so I'm surmising you haven't seen those threads. Best, Josh

No problem. I realized I forgot to answer Barrows' question about the frequency of silver use in high-speed PCB design - I haven't gotten the feeling that it's particularly common; the things that distinguish a high speed PCB design from one that isn't are features like controlled impedance, the grounding scheme, the number and type of vias used, and especially multilayer construction. The plating is more of a manufacturability issue than it is a feature of high-speed design per se. In our particular marketplace, "silver" is a key word that tends to make audiophiles feel safe and happy, so that doesn't hurt, either.

@Barrows - any effect of silver on the propagation speed of the traces is completely swamped by the geometry of the trace to dielectric relationship, and also the type of dielectric material. I chose silver plating mostly because it produces a very flat surface on the board while avoiding some negative tendencies of other, similarly flat plating methods. @G7 - silver sulfide tarnish is self-limiting and does seal off the underlying metal, unlike iron rust.

The upgrade price hasn't been established yet, but it will probably be something close to the cost difference between the old and new models. Regarding the silver PCB plating: Barrows is correct; it is covered with soldermask and is thus sealed. The portions that are exposed (the pads and lands) are the areas where pins are soldered, so those too are immune to the problem. BTW, the tarnishing problem with silver isn't oxidation - it is silver sulfide, caused by exposure to hydrogen sulfide.

Hi Shibboleth - I'm very grateful for your Diverter ownership, and am sorry to hear the news of the new version has caused you any frustration. For future reference: I am always developing new products and improved versions of old ones at Sonicweld, and product upgrades are always a possibility, and I would even say a probability. I'm sorry the 24/192 version wasn't available at the time you made your purchase (and still isn't, actually), but that's just the nature of any technical product's life cycle. There will indeed be an upgrade path to the latest version. As I noted in my previous post, I expended a lot of effort in making the new version mechanically compatible with the old one, which was difficult given that it is considerably more complex. I don't know what upgrade pricing will be yet, but I will make it as affordable as possible. The upgrade will entail a complete PCB swap along with a new machined top cover plate.

@ajay556 - Yes, I expect the new version to offer increased performance at any bit depth and sampling rate. It contains all of the accumulated refinements from past versions, plus some new circuitry I'm excited about.

@Pat - I hear you about the pet peeves and Mr. China! @Chris - Sure, my pleasure; thanks for your kind words. I have to say I kind of liked the Nagra pyramid amps, though I'll admit they're sort of an odd departure from their tried and true formula. I give them props for trying though, and I think everyone should be allowed a flop or two. I have a lot of respect for Nagra; they seem to turn out consistently well-engineered designs that are true to a certain philosophy and aesthetic. @Scot - glad you got a kick out of "maddeningly lilliputian" and I really meant it - I hand place all of the SMT parts on each Diverter. When you're talking about parts as small as an 0201 package size, they're about the size of flecks of ground pepper. Just breathing normally will knock them off the board, to be lost forever in the world of dust mites on the floor. I could send the boards out to a contract manufacturer, but some of the techniques I use on the PCB aren't conducive to conventional batch reflow processes, so I do it all in-house. Plus, its the only way to maintain total control over the quality and process. It's also a huge pain. @Diminishing Returns - thanks for the kind comments. To clarify about the colors you mentioned, Diverters are available in a natural machined (silver) finish, red anodized, and black anodized. In all cases, the top cover plate has the natural machined finish.

Hello to everyone - it has been a while since I’ve posted here. I’d like to address some of the points raised about the Diverter’s design. I feel like I’ve covered this ground before, but I acknowledge there may be new members here who weren’t privy to past discussions, and the fact that the same concerns persist suggests I may not have communicated about them as effectively as I’d hoped. Any product designer must realize (or will quickly and painfully learn) that not everyone will respond positively to their work, so I have no argument whatsoever with those who feel the Diverter doesn’t represent a combination of design choices that suits their tastes, preferences, or needs. I do, however, take exception to any suggestion that the chassis of the Diverter represents some kind of very calculated and deliberate attempt on my part to bilk naive people out of as much money as possible. It seems that the crux of this conspiracy theory centers around the quantity of aluminum employed in the chassis, or perhaps the aluminum volume to circuit board area ratio. My supposition is that these ideas are advanced by people with little or no experience in industrial design and CNC machining, but I recognize that such experience isn’t widespread, so I don’t say that in an effort to insult or belittle. Let me try to explain the basic economics behind CNC machining. It is an unforgiving and meticulous process, and as in any industrial process, time is money. Unless you’re working with a ridiculously expensive material like Vespel, titanium, or one of the nickel-based superalloys, the cost of the raw material itself is generally a small fraction of the overall part cost. The bulk of the expense comes from the number of setups, the programming, probing, number of tool changes, part run time, fixturing, washing, drying, cleaning, inspection, anodization with its myriad steps, and the part-to-part changeover time. Add to that the cost of consumables like tools, inserts, coolant, machine oils, and general maintenance. In the case of the Diverter, the chassis is fabricated in small batches here in the US and not in China, and is done entirely by me in my own shop, on my own equipment. Say, for instance, I used only one ton of aluminum to make a Diverter chassis instead of two tons. The smaller part would be less expensive, yes, but not half as expensive or anywhere close to that; it might be 5% or 10% cheaper. A complete Diverter chassis requires eleven different CNC programs, two dedicated fixtures, and six different setups (seven in the case of an anodized chassis), whereas the cost of the aluminum blanks themselves is maybe $25. Now there are some of you here, I’m sure, that will take umbrage at the idea that a $1300 product is made with $25 worth of aluminum. To me, this is somewhat akin to criticizing a painting because it was done with $15 worth of acrylic smeared over a $20 canvass which is stretched over $10 worth of wood. The value of any well-crafted object is not simply the sum of its component and material costs. You’re free to disagree with the mathematics of my particular value equations; my competitors offer a myriad of alternative choices which feature far less expensive chassis. If we’re having a frank discussion about costs, we should also talk about an intangible that people seem uncomfortable with for some reason: the inherent value of the design itself; the gestalt, if you will. Do I charge for this? You bet I do, because it represents a great deal of time, effort, and original thought on my part. I often get the feeling in these forums that people place little or no value in the design itself, particularly anything relating to aesthetics, and that the idea of paying for it is objectionable and unsophisticated at best or is indicative of fraud and gullibility at worst. If you don’t value aesthetics, or if you don’t care for my particular style of aesthetics, no problem - I respect your point of view. But I have a big problem with the idea that good industrial design and attention to aesthetic details somehow preclude the idea of overall value, or that it is mutually exclusive to good circuit design and performance. It is quite possible (albeit difficult) to create a device that is both beautiful and technically accomplished. Conversely, I think many people in this hobby blithely assume that because a design is packaged in a utilitarian or even ugly chassis, the designer is “honest” because he allocated the bulk of the development funds to the functional innards of the device - money which presumably would have been squandered on silly aesthetics otherwise. Having examined many such devices, I offer the opinion that the opposite tends to be the case: I find that devices that strike me as poorly designed on the outside are often poorly designed on the inside, too. A beautiful design is certainly no guarantee of performance or value, but I think it is unfortunate and narrow-minded that some dismiss such purely because the designer cared about aesthetics. I’d like to disclose a bit of the history of the history behind the Diverter’s design, and why I made the design choices I did. I began with the circuit and board design, because that is the functional core of the device, its most important feature. I freely and proudly admit that the chassis of the Diverter represents a great deal of work and thought, but I want to be clear in saying that the amount of time and effort I spent on its design doesn’t even approach what I have put into the circuit itself and the PCB design. Again, I think there are some misconceptions at work. The PCB of the Diverter is indeed small, and I think there are some who assume that the value of the Diverter is in inverse proportion to the circuit board size. The size was deliberate, yes, but not in an effort to reduce costs; rather, it was in an effort to yield the shortest possible signal paths and the tightest current loops. Performance was my first and only concern in the PCB design, and making it as small as it is was a good deal more difficult and expensive than if I had used a more conventional, larger, lower tech PCB design. I could easily produce a version of the Diverter’s circuit board that is two or three times its current size and populate it with big, pretty, boutique through-hole parts, and it would be far easier (and cheaper!) to assemble than the current design, with its maddeningly lilliputian parts. I find it ironic that some (perhaps many) in this marketplace would feel a larger, more photogenic PCB design represents a far greater “value” than the highly compact, optimized design I’m offering now. There were no initial constraints on the PCB size or cost; its shape and proportions flowed naturally from the optimal circuit design and layout. Once it was completed, I designed a chassis from scratch to complement it and complete the design. As both the designer and the machinist, I was not limited by standard, off-the-shelf extruded chassis designs available from a catalog; I came up with an entirely original design based on my electrical, mechanical, and aesthetic goals. Electrically, I wanted something that offered excellent electrostatic shielding; mechanically, I wanted something well-damped and which had enough mass to stabilize what would otherwise be a very lightweight device, one which would tend to be a the mercy of thick, stiff audiophile cables; aesthetically, I wanted to create a piece of industrial art, one which would reflect the design values of Sonicweld and inspire pride of ownership. I feel like I’ve succeeded in all three objectives, and I’m very happy with the result. The Diverter's success in the marketplace suggests that many others agree, though I don't expect that fact alone to sway anyone currently on the fence. To briefly address some of the other points in this thread: I can confirm that a 24/192, async-based Diverter has been in the works for quite some time. I expect a release soon, but of course cannot say exactly when. I took great pains to make the new PCB mechanically identical to the old one, so the current design can be upgraded via a board swap. Regarding USB power: it is entirely possible to derive audiophile-grade power from the USB bus, just as it is possible to get it from the similarly noisy AC wiring in your home. I’ve written extensively about this in the past here and over at Head-Fi, so I’d encourage ferreting out those posts if you’d like more of my views on the subject.

"But to say that USB cables make a difference -- I just can't see any reason for that. Data transfer over USB is a bi-directional protocol with error correction and retransmission." Not so for isochronous transfers (the type used in USB audio). This type of transfer uses stream pipes, in which the content has no meaning in the context of the USB protocol. Basic CRC checking is performed for error detection, but there is no mechanism for re-transmission or guaranteed delivery as in Ethernet. The emphasis is on guaranteed bandwidth for real-time data delivery, with a tradeoff in error recovery robustness. Incidentally, Ethernet cables can and do affect quality of service on a LAN, and for at least some of the same reasons that USB cables can affect audio quality. When I worked in IT in the nineties, there were several occasions on which I had to pull a new cable because it was badly kinked (which would show up in a TDR plot) or because it was directly on top of a flourescent light fixture in the ceiling (an EMI problem). The protocols are very different, so a direct analogy between Ethernet and isochronous audio data transfer over USB is specious. Bit-level errors in a short USB interface are rare enough to be discounted as a mechanism of degrading sound quality in USB audio practice, but I can think of two other cable effects that do have an influence: 1.) Signal integrity effects arising from cable parasitics. For examples of this, look at cable eye diagrams, or the standards and industry practice surrounding HDMI cables. 2.) Shielding, antenna effects, and ground loops. In my own testing, I've seen considerable and repeatable differences in characterizing USB to SPDIF translation devices for jitter based solely on how the USB shield was terminated. Regarding lossless formats and how they might sound different: I'm in complete agreement with you regarding the actual data content of an uncompressed file versus one that has undergone lossless compression. But I can think of at least one mechanism that could account for a sound quality difference in playback: power supply modulation and/or differing EMI signatures. Granted, the CPU horsepower required to decompress losslessly packed files is generally very small, but it is not zero. Ergo there would be some probability of non-random, correlated noise finding its way onto the USB cable and into the downstream devices. I want to be clear that I think these effects are probably exceedingly small in most cases, but are at least worth consideration. Although we can say that the data is identical in both cases, we cannot say that the conditions of playback and the effects on various error mechanisms are precisely the same with lossless compression vs. uncompressed playback.

I can confirm that Barrows is listening to the latest version of the Diverter. Here's something I found interesting: I recently loaned my iLok to a friend of mine. He's a sharp guy, is computer savvy, and loves music, but isn't an audiophile in the classic sense of the word. After playing with Amarra for several hours, during which he switched it on and off repeatedly (and sometimes blindly, with a partner doing the switching), he reported that he thought Amarra's effect was quite obvious, and described it in terms similar to what many of you have written here (spatial presentation, slight reverb sound, plain iTunes sounding a bit dull by comparison). The thing I found fascinating was that his playback chain included streaming the music through an Airport Express! Some have speculated that Amarra has some effect on jitter, which I have always found implausible, but this little experiment (uncontrolled and anectdotal as it is) seems to support the idea that Amarra's manipulation, whatever it may be, isn't jitter related; any such effect should be hopelessly swamped by the RF transmission, data buffering, and poor clocking an Airport Express. I thought his feedback was interesting and valuable, as it came from one who is observant but lacks the usual audiophile vocabulary, experiences, and agendas. Anyway, this isn't meant to be definitive in any sense, I just thought it was an interesting experience I'd share.

There are a lot of newer tantalum caps that mitigate the highly exothermic failure modes of the older styles, BTW. The polymer caps are usually better behaved, and there are some with a built-in fusing feature, generally termed "surge resistant" or something along those lines. I agree with CG though that the safest policy is to include some kind of current limiting scheme, whether it's something like a PTC element, or some active device. Gordon, I agree that a low-ESR cap can often be a good thing in the input of a device, but there's also a significant danger if the ESR is too low, and this is a perfect example of what Pat and I were referring to earlier, when we were talking about the lowest ESR not being desirable in all applications. The power-on surge or subsequent high di/dt current steps can excite the input cable inductance and capacitance combo into poorly damped oscillation. Most electrolytics have enough ESR to provide sufficient damping in this situation (at least when used singly), but really low ESR tantalums or ceramic caps could both be trouble. Incidentally, relating back to the discussion about outboard supplies earlier, this is another strike against them in my book. In connecting an aftermarket supply to a device through a long umbilical, one never knows if the input capacitance is sufficiently damped or not - I could see a lot of designers who may not be aware of this phenomenon putting scads of very low ESR capacitance at the input (because more is always better, right?), and thereby creating a set of conditions that would make ringing likely. I think this kind of accidental interaction is probably to blame for a lot of otherwise fine products not working well together, and is also a strong argument for systems-level design, IMO. Josh

Thanks for the clarification Pat - that makes a lot more sense, and I suspected it was something along those lines. A lot of the older regulators are sensitive to very low ESR caps.

Out of curiosity Gordon, what sort of low ESR caps did you have bad results with? Electrolytic polymer, ceramic, or something else? Were they bypassed with something else, or running alone? Interesting details on the Naim product - sounds like it could be promising.

"I don't see why some of you insist on "numbers". Any kind of numbers. How do you know what they mean, and whether they are applicable or not? So what if some error amp has 5,000 V/uS slew rate? What does that tell you about it? Nothing. May sound impressive, but it really tells you nothing about it. A picture of what the noise looks like is a lot more helpful. But it only tells part of the story." I wholeheartedly agree with Pat on this point. I had typed out a long discussion about it in a draft of one of my earlier posts, but thought the better of including it as the rest of the post seemed more than long enough already. The specs quoted about the power supply sounded like classic datasheet numbers, which are next to worthless, and generally an attempt by the manufacturer to generate impressive-sounding ad copy. Datasheet parameters can be useful if you're designing something and want to winnow down a handful of candidate parts you want to try in a circuit, but they're no guarantee of performance - any part must be vetted in a real circuit, always! Especially with high-performance opamps; it's so easy (and common) to get lousy performance based on poor implementation (layout, parasitics, unwanted coupling, etc). The root phenomenon is at least understandable: this is a very crowded and competitive market, and people need some comparative basis to sort through the overwhelming purchasing options. Specs are an easy and intuitive means of comparison, but also one that is susceptible to misunderstanding, deception, and misrepresentation. The problem is that people tend to ascribe often unwarranted performance to particular parts or specs independent of any kind of implementation considerations, especially after they have been blessed and sanctified in the forums. This isn't to say crowdsourced information has no merit, only that one should be aware of the need to weigh implementation details at least as strongly as anything else. Another example I see often is with capacitors - much has been written about the merits of low ESR caps, and they certainly can be a very good thing when used the right way, in the right application. But to say that cap X with a .02 ohm ESR is dramatically better than cap Y with a .04 ohm ESR is misleading at best, because the way the capacitor is mounted can swamp much of the benefit the low ESR provides (especially in high-frequency applications). Aside from that, low ESR isn't always desirable; some damping resistance is often a good thing. Thus you'll see many DIYers build some misguided creation with an array of some expensive, boutique capacitor-du-jour that people think is great, and they'll put the array in an outboard enclosure with 10" of flying leads connecting back to their modded power supply and call it a "bypass". This new circuit will certainly have an effect, but it likely will not be the one they were hoping for - they just unwittingly created a tank circuit and a new antenna, both with potentially high Q. I don't mean to sound overly critical of DIYers in saying that - for all I know, such adventures may be the baby steps of someone who will go on to become a legendary audio designer. We all have to learn from mistakes, and if someone splashed pictures across the net of some of the first mods I did 25 years ago, I don't think I'd be proud to claim them. One of the things I've been criticized for in the forums is not providing more information about my products (I suppose this had a part in the genesis of this thread) and I suspect much of that comes from people being conditioned to seeing the usual presentation of cool-sounding parts, figures, and specs. I would never cite such information, however, because it so often constitutes a non-sequitur. I think a paucity of information also has a hand in much of the "pretty boxes" criticism that has also been applied to me. People see a cool-looking product with little or no specs, and assume it must be some kind of scam. In fairness, there are a few cases in which that criticism might actually be merited, but I think many times it is just manufacturers trying to rise above the numbers game. It can be a real catch-22 for a manufacturer, because if you supply few technical specs for your product, it's labeled as snake oil; if you supply too many or ones of specious origin, you're engaging in specsmanship at best and deception at worst. It's hard to find a happy medium.