Schiit Happened: The Story of the World's Most Improbable Start-Up

Discussion in 'Jason Stoddard' started by jason stoddard, Jan 23, 2014.
  1. wink
    This would have to rate as the most lucid post on this thread since the beginnig of the year.........:L3000: :ksc75smile:
    swspiers, 33na3rd, US Blues and 4 others like this.
  2. Pietro Cozzi Tinin
    It would't surprise me if it already has.
    Look at the serial number.
    If it begins with a B for someone here, we know it has.
    US Blues likes this.
  3. US Blues
    Trickle down doesn't work in economics, but it does in Schiit technology.
    bosiemoncrieff, kehorton and MWSVette like this.
  4. MWSVette
    One of the few situations that I would not mind being trickled on...

    US Blues likes this.
  5. Pietro Cozzi Tinin
    The point is:
    Mike will always work on optimizing his technology.
    It would be strange to do that only for the Yggdrasil.
    RCBinTN and US Blues like this.
  6. winders
    I am pretty sure somewhere on these forums someone had shown that Gumby has been shipping with “B” serial numbers for a very long time.

    I wouldn’t assume Gumby can get the same upgrade. Yggdrasil has a lot more room internally for all manner of upgrades that may not fit in Gumby.
  7. belgiangenius
    Mike, we would all still love to hear whatever you're willing to share on what went into Analog II revision.
  8. Scott Kramer
    +1 an analog 2 write-up!
    Last edited: Feb 14, 2018
  9. rlow
  10. Scott Kramer
    Realized that right after posting, fixed, you were too quick! :)
    Last edited: Feb 14, 2018
    rlow likes this.
  11. rlow
    I agree this could be true, but I feel like there is likely still some improvement coming to Gumby along similar lines, although perhaps not exactly the same - there are obviously design differences between the 2 DACs aside from the case that would likely necessitate differences in the analog board anyhow (but that is a layman's guess obviously).
  12. inmytaxi
    Absolutely want to hear how you think the Magni compares to the Vali. I too have a Vali with the Gold LIon e88cc (more and deeper bass and better overall clarity than stock Vali 2 tubes) which I enjoy so much I don't see a reason to upgrade, but the Magni 3 sounds like so much amp for so little money I find it hard to resist.
    aamefford and ScubaMan2017 like this.
  13. Jason Stoddard
    2018, Chapter 2:
    Engineering, Part 1

    Ready for something different? How about a deep dive into engineering of an audio product? A dive that goes into the choices, constraints, and decisions on everything from the initial design goals to the characteristics of the devices used, to the finished product itself?

    Sounds a bit too boring and engineer-y?

    Well, how about if it culminates in a product that can be simply a fun novelty…or a buildable product?

    “Wait a minute,” you’re saying. “What is Stoddard on about now? Is he talking about (gasp) DIY?”

    Well, yes and no. Although I’ll end up releasing schematics, a BOM, and PCBs, it should be noted that we don’t have the infrastructure to support a lot of DIY. Nor do I want to take away from the other companies and individuals that are doing a much better job than we ever could with DIY. And, when you get right down to it, what we’ll be designing is a low-performance product that is more of a fun novelty than anything else.

    So why do it? So we can take a deep look at designing a product:
    • We’ll set engineering goals and expectations.
    • We’ll look at the different kinds of active devices we can use, and what their characteristics are. We’ll be looking at potentiometer tapers and connectors.
    • We’ll touch a bit on packaging. (As of the time of this writing, I just now realized that I could do a 3D-printed case for this. Never considered it before. Hmm.)
    • We’ll also get into measurements and subjective performance (subjective, in this case, means, “Noise level, clicks and pops, interference, and other unexpected behavior, not “singing the praises of how this thing sounds great.”)
    The reason I’m breaking this chapter up into a number of pieces is that it’s pretty big. It might be 12-15K words (plus pictures) when it’s all done. Or more.

    So what are we doing?

    We’re gonna design a small tube hybrid headphone amp. Think, kinda like the original Vali. And we’re gonna make the PC board pretty and round, so you can use it as a coaster.

    Yes. A coaster.

    Like this:


    As in, something you can put a beer…er, I mean, a drink, on.

    Yeah. I know. We’re completely insane.

    How This All Started

    This started on a lark in late summer 2017, when I’d gotten the first PC boards back for our phono motor can. This is one of the simplest PCB designs you can imagine. It has a DC barrel connector on it, places for a couple of large motor-run capacitors, and a couple of holes to screw it down onto the motor can base.

    In fact, only one thing made this very boring PCB stand out: it was round:


    “Huh, nice coaster,” Tyler said, when he came into my office one day, probably to have me sign checks.

    “Coaster?” I asked.

    Tyler picked up one of the motor PCBs and twiddled it beween two fingers. “Coaster.”

    “Ohhh…kayyyy,” I said, doubtfully. “Kinda boring coaster.”

    Tyler shrugged. “It’s shiny. It’s red. It has electronic-y lines on it. It looks cool to me.”

    It still looked boring to me. But it was about the right size to be a coaster. Hmm…

    “Maybe you should cut up our old PC boards and sell them as coasters,” Tyler added, interrupting my thought.

    I shook my head. That wasn’t practical. We didn’t have that much scrap, and, in any case, components would have to come off the boards, and there’d be all sorts of logistics having them sent out and cut up, and not many people would be that interested, and not all of them would look all that cool

    And that’s when a wonderful, terrible idea bloomed:

    What if we made a cool-looking PCB…and sold it as a coaster? That was a piece of Schiit Schwag I could really get behind. It was engineering-y. It wasn’t just another t-shirt or coffee mug. It was something I hadn’t seen before. Hell, we could sell 4-packs of PCB coasters at the Schiitr. Or even online.

    And then I had an even crazier idea:

    What if the PCB was buildable into a product that actually did something interesting? What if it lit up? What if it was a VU meter? Or, what if it did something really useful? Hell, what if it was…a small headphone amp?

    I laughed out loud.

    Tyler looked at me like I was insane. “What did I miss?” he asked.

    “Make it a headphone amp,” I said.

    Tyler frowned. “Make what a headphone amp?”

    “The coaster!”

    Another look of incomprehension. “The coaster…is a headphone amp?”

    “Yes! Or no! You could just use it as a coaster, or you could build it into a headphone amp.”

    My mind was racing. Maybe this was where we could do something simple like a CMOY, maybe using one of the newer op-amps that worked really well for headphone output. But there were a lot of CMOYs. Or maybe this could be something like an op-amp and follower amp. But there were plenty of those out there, too. Or maybe this could be something like the original Magni, a simple Lin-topology amp, so I could talk about discrete design.

    “I don’t get it,” Tyler said.

    I pulled myself back to the present. “It’s simple,” I said. “We make a cool-looking PC board coaster. We make sure it’s safe, too—run it ENIG or something, not HASL—and you can use it like a coaster. And that can be it. It’s a cool coaster.”

    Tyler nodded. “Got that. What about the headphone amp?”

    “That’s what you build the PC board into…if you want. No obligation. You can just use it as a coaster. Or you can build it.”

    “Can you use it as a coaster after you build it?” Tyler asked.

    I laughed, envisioning someone trying to balance a drink on a PCB covered with parts. “Not likely. But hell, I haven’t figured out exactly what it’s gonna be. It should be something fun, something different…”

    And then I trailed off. Because I had the perfect idea: do the old Vali.

    Because, you know, other than the ringing, the old Vali was a cool little amp. We still get fanmail about it. And it was different. A little bitty tube hybrid amp using NOS pencil pentodes run in triode mode. That might be fun enough to build.

    Well, except that the original Vali had some pretty high voltages running around on it (about 80V before regulation), which wouldn’t be such a hot idea for a bare-PCB amp. And, the original Vali used some custom parts, which wouldn’t be easily gettable. And the original Vali was done for robotic assembly, so the spacing of the parts wasn’t so fun. And the I/O would be weird on a round board.

    So, yeah, maybe more like: do something kinda like the old Vali. But safer, and more buildable. On a coaster

    That actually seemed like a pretty good idea. So that’s where I started.

    Yes, I know, I know. Crazy.

    Designing for DIY-Lite

    Now, here’s the thing. I’ve never designed a damn thing for DIY. I probably suck at it. If this first attempt goes off the rails, well, we have coasters. And, again, I don’t want to take anything away from companies already doing DIY. Anyone already doing DIY will do it better than us.

    Wait. Stop. Go back and re-read that paragraph.

    No, wait. Let me put it in bullet points:
    • I have never designed for DIY
    • I probably suck at this
    • Every other company doing DIY is probably better than us at it
    • Any other DIY project will probably be better than this
    • Any other DIY will probably be more well-documented
    • But, you can use the PCB as a coaster
    Okay, got it? Cool.

    Now let’s talk a bit more seriously about DIY, or at least what I think about how it’s different than designing for production. Because it is different.
    • Both require documentation, but in one case you’re talking to external vendors using pick-and-place robots and internal assemblers and test techs handling finished products, and in the other case you’re talking to people who are going to be picking up a soldering iron and doing this all themselves.
    • Both require a PC board layout, but the production one can use 3x3mm chips with 20 pins and a buried thermal pad, and the DIY one needs to have space so that humans can make it without going blind or insane.
    • Both require a BOM, but the production one can freely specify custom or tooled parts, like Alps pots with 10A taper or injection-molded light pipes, and the DIY side needs to look carefully at what’s available off the shelf, and design to that.
    Aside: a BOM, or Bill of Materials, is a list of everything that goes into a product. Yes, I know, not very exciting.


    Above: Schematic Capture


    Above: PCB Layout

    And yet, when all is said and done, there are many similarities. A DIY design will need:
    1. A schematic. This is the starting point for all electronic design. Well, all documented electronic design, anyway. You use schematic capture software to create a schematic. Some schematic capture software allows you to also simulate your design. Simulation is running the design in virtuality, without building a thing. Simulations can provide a very accurate preview of a design…or can have nothing to do with reality. Success or failure of a simulation is dependent on the models you use and assumptions you make. A quick path to failure is not taking into account thermal or power dissipation factors. An input differential amplifier using a LED-biased current source with a 1 ohm emitter resistor might work in simulation, but the 1.2A (that AMPS) it passes will cause small-signal transistors to quickly go up in smoke. We won’t be doing simulations. The schematic capture software I use is Kicad, which I recommend unreservedly. It’s a very robust package, with surprising capabilities.
    2. A PCB layout. From your schematic, you’ll want to produce a PC board. That is, if you expect to make more than a handful of products. PC boards, or printed circuit board, is the basis for all modern electronics. It takes the “wiring” and puts it on a board, making assembly much faster and more consistent. It also allows you to pack much more in a small space. Modern PC boards are usually at least double-sided, meaning they have traces (wires) on both sides. However, many are 4-layer, 6-layer, 10-layer, or even more. Increasing the number of layers helps a designer manage more complex layouts more effectively, to shrink the physical size of products, and to achieve higher performance. However, better be sure your design is right—you’re not going to re-work internal traces! The PCB design software I use is also Kicad.
    3. A Bill of Materials. As I mentioned before, this is a list of everything used in the product. And I do mean everything. From every resistor in every location, to the PCB, to screws, feet, manuals, and chassis parts, a Bill of Materials is all-encompassing. The BOM also specifies which parts go where…so you know that things like a 100 ohm, 0805 size resistor goes in R8,10,104,105,204, and 205. BOMs we usually do in Excel. Google Sheets is also fine.
    4. A prototype or prototypes. A prototype is, by definition, the first model of something. In electronic design, this usually means “a PC board, possibly hand-assembled, for testing and design verification.” We usually do a couple of prototypes (at least) to iterate one or more design ideas, and/or to fix errors, add features, improve performance, or eliminate bad ideas. This little amp (let’s call it Vali Mini) is no different—it went through two prototypes before it reached its final PC board. We sometimes get PC board prototypes through our main PCB manufacturer on the east coast, and sometimes through an outfit called EasyEDA. EasyEDA is inexpensive for short runs, and even provides their own schematic capture and PCB layout software online (which we have not tried).
    5. Documentation and measurements (DC operational points, gain, etc). This is so the builder has a basic idea of any gotchas in the build, and has some test points to verify if what he’s built is working correctly. Many DIY projects have extensive documentation, because they’re aimed at people who are just getting started in electronics. This little amp will have only basic documentation, because if you’re going to be jumping into learning electronics, this isn’t a great place to start. This is a discrete design using surface mount components, with a reasonably high number of parts. Start with something through-hole. Learn soldering. Get comfy with surface mount soldering. And then come back to this. Or, use it as a coaster.
    Are you thoroughly confused? Intimidated? No? Good. It’ll get deeper. Wait until we get into the whys and wherefores of selecting devices, including some basic physics of vacuum tubes, BJTs, and FETs. That’ll be next time.

    In the meantime, let’s move on to something really crazy…where we decide to make a poor-performing, low-power, highly compromised amp.

    Sound nuts? Read on.

    Setting Engineering Goals and Expectations

    Okay. Let’s get this right out there: you can easily create a headphone amp with lower distortion, higher power, and a lower noise floor than the one we’re going to design here. How easily?
    • Want lower distortion and noise? Just take an OPA1688 or similar modern audio op-amp, throw a couple of 9V batteries on it for power supply and use a couple resistors to set gain, and you’ll crap all over this design in terms of distortion and noise. It might lose in absolute power output, but hey, ICs do have current limits.
    • Want more power to go with your low THD+N? Use a higher-voltage power supply, like +/-15V, and add some followers within the feedback loop. Boom. Tons more power.
    • Want even more power? Boost the supplies to the limit of the IC and run an output stage with gain so you can get near the rails. Watch out for oscillation in that case, though. Or go discrete. Or use error correction. There are plenty of cool games you can play to make good objective numbers. Of course, by now we’re well beyond “easy.”
    But you get the picture. If you want to design a headphone amp that produces great numbers, you can.

    Aside: this does not minimize the amount of effort needed to produce exceptional numbers—as in, class-leading single-digit PPM distortion. That’s a whole ‘nother challenge. Just one we’re not taking on. Making a Porsche go fast around the Ring is one challenge, making an S-class quiet and butter-smooth is a different challenge. Both take tons and tons of engineering. And neither invalidates the other.

    So why design something that’s less than optimal? Why make something that might have 0.5% THD at best, even at fairly low output? Why accept a noise floor that may be audible with high-efficiency IEMs? Why settle for power output that won’t be enough to drive HE-6s?

    It’s simple: because that’s what we choose.

    Stop. Go back. Read that again.

    Then, turn it up: because we choose to do it, knowing full well the alternatives.

    Yes. We can choose to design a better-performing headphone amp. But we don’t. We choose to make something that’s well, maybe a bit more interesting. We choose to start with a tube hybrid topology, because a lot of people are curious about tubes. Lots of people who choose an accurate, neutral solid-state amp still wonder, “What if? What do tubes really sound like?” So adding a tube makes it more interesting. Maybe. At least in my mind.

    And, starting with a tube in the design allows us to explore more facets of amplifying devices. How do tubes work, compared to FETs and transistors? How do they compare to op-amps? What are the upsides and downsides of each? How do they interface?

    It also allows us to talk about the trade-offs. How are we limiting ourselves in designing with a tube? What does it mean if we decide to go with no overall feedback? How does a mix of devices interact?

    So, yeah, interesting. And maybe educational. And the finished product will have a couple of small glass envelopes on it, as well as a couple of glowing LEDs. Pretty nifty, if not particularly high-performance.

    And yeah, maybe that’s a point as well: that cool looks don’t mean great performance.

    Okay. Enough blathering about choice. Let’s define our goals and expectations.

    Primary Goal: To design a functional tube hybrid headphone amplifier that fits on a coaster-sized round PC board.

    1. PC board usable as a cool-looking coaster without adding any parts to it.
    2. Documentation provided for moderate to advanced DIYers who want to build a functioning amp with the PCB.
      • Schematic
      • PCB (finished)
      • Bill of Materials
      • Basic measurements
    3. Low cost to build into a functional product
    4. Uses easily available parts
    5. Safe to use and operate
    Now, what’s interesting is how the sub-goals start to affect the features—especially two key words in the sub-goals: available and safe.
    • Available means no custom or unobtanium parts. No injection molds or $600 potentiometers. No parts that exist in catalogs only. No circa-1929 tubes of which only 4 exist in the world.
    • Safe means no high-voltage rails, no 350F hot tubes, and at least basic protection for headphones. You can’t have a 200V rail running around exposed on an open-frame amp. Well, you can, but please keep that to your own design, thanks. Safe also means some kind of protection from turn-on and turn-off transients.
    Now, with those two words, we’ve really cut down on our tube choices. I’ll get into that later in more detail, but you’re really down to subminiature tubes (like the original Vali), or a low-plate-voltage design (but you still gotta deal with the heater…ah hell, let’s get into this later.)

    Okay. Goals defined. Let’s move on to features.

    Features: This thing needs to be a functional headphone amp, but it also needs to fit on a coaster, so let's not go too nuts with features. Maybe something like:
    • One line input
    • One headphone output
    • A volume control
    • As easy to use and safe power connector
    • A power switch (not necessary, but nice to have)
    Yeah, I know, bare minimum. No preamp outs or pass-thrus or input switching or DAC modules or anything fancy like that. Of course, there are still plenty of decisions that need to be made in the future, like what kind of connectors we’ll use for the line input, if the headphone output will be 1/8” or ¼”, what type of volume control, what kind of power connector, what power switch. But we’ll get to that.

    Okay, with those goals and features defined, what can we expect in terms of performance?

    Well, we’re still pretty early on in the design process, but we can begin to get a glimpse of what we’re looking at, based largely on the requirement for the product to be about the size of a coaster, that it has to be safe, and it has to use a tube.

    What? How do we do that? Magic? No, just a basic knowledge of engineering limitations. If it’s gonna be about the size of a coaster, it doesn’t have a lot of space to dissipate heat. So it can’t run hot. You’re talking about something that runs from a wall-wart, and a small one at that. Safe means it might have 30-40V total between the rails, all in. Otherwise you might start feeling a little jolt if you touch it in the wrong place.

    Not much heat dissipation, low rails and a tube means pretty low performance in terms of THD. Period. Of course, it will be “tubey” THD, meaning mainly low-order harmonics, at least until the output stage starts giving up its linearity outside the Class-A operation window, which will be low (remember, low heat.)

    But I’m getting ahead of myself again. So what would I expect, performance-wise, from an amp like this?

    Performance Expectations:
    • 0.5-1% THD+N at 1VRMS out into 300 ohms
    • 300-500mW full-scale output at clipping into 32 ohms with several percent THD
    • 90-95dB S/N ratio, unweighted, referenced to 1V
    Sounds a little disappointing? Maybe not so much. Lots of people liked the way Vali 1 sounded, and its measurements weren’t a lot better (about half that in THD, higher output, about the same S/N ratio).

    Is it “accurate?” No.

    Is it “fun?” Maybe. As with everything, depends on your perspective.

    More next time, in a couple of weeks. This is already getting pretty deep. In the meantime, here's a picture of where we're heading.

    Schiit Audio Stay updated on Schiit Audio at their sponsor page on Head-Fi.

  14. Rowethren
    Sounds like a really cool idea, would definitely crack out my soldering iron for this! :beerchug:
  15. Ableza
    What a cool idea. I love your description of how this was created. So many great ideas come from off-hand comments, or fantastical "what ifs" or flights of fancy. It's just rare to be in a position to turn one into reality. Kudos.

    The one design criteria you have not mentioned yet is also the one I find is many times the most difficult to meet: target price point and how that affects both design and component selection.

Share This Page