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

[kidster18]

Thanks Jason for posting this series - you are an amazing writer. I read all the chapters in 1 go...

 

[theebag]

  Then big guy, when are you opening up the pre-order book ? 

 

 
At least it isn't vaporware .

 

[UmustBKidn]

>> So, a tip of the hat to anyone who can make it as a restaurateur. That’s a helluva business.
 
I love watching all of Gordon Ramsay's shows. Especially Hell's Kitchen. I just wish I could get a seat at that restaurant some day.
 
 
>> Software Business Problems. Software’s a bit different. Your “hard costs” are actually salaries for the programming staff, your facilities costs, your admin costs, etc. These >> aren’t tied to the amount of software you sell. If you spend $50K developing software, it doesn’t matter if you sell 100 copies or 1,000,000 copies at $99 each, because you’re >> going to be distributing it via download or via inexpensive media.
>>
>> But—and here’s the big but—you have to support it. Unlike simple hardware, you have to assume that it won’t install on some systems, it will conflict with other software,
>> many customers may need some hand-holding in using it. And, as an added bonus, the OSes and other software changes from year to year. This means that what worked
>> yesterday may not work today, and vice-versa.
 
Well, in your case, you're selling hardware. Any necessary software is somehow considered part of the hardware. That's mostly just a perception issue.
 
Software is a curious product. Unless you're a software engineer, software is a mysterious thing that somehow ends up making your computer (or other device) work as you expect it to (usually). It is often not viewed as a product unto itself (unless you're talking about Micro$oft Office, Windows, or maybe Flight Simulator). People often think that software is something that makes their computer / phone / tablet (etc) work like they think it should. There is some difficulty divorcing the concept of software from hardware: the average consumer does not view these two things as separate entities. They seem to think that hardware isn't working properly unless they can find software to make it do what they want it to do.
 
So really, your customers view their driver issues as part of the hardware they purchased from you. They don't think of software drivers as a separate product that requires separate maintenance or support - they view it as an integral piece of the hardware they bought. After all, their hardware won't work until the driver issues are resolved. They don't think about what differences there might be, between their computer and myriad others that might require individual tweaks to make some driver work. Oh no, to the average user it's all the same - it's much easier to say "My Schiit Widget didn't work" than to say "The driver for my Schiit Widget doesn't work correctly on my Lenovo Model X, because this miserable computer didn't implement the proper library functions..." etc ad nauseum.
 
>> What it has to do with Schiit is that, with the launch of Bifrost, we became a software company by default.
>> A very, very lightweight software company, yes, but we had new complexities of firmware and drivers to deal with.
>>
>> “Oh, big deal, you get the Giant Baby of the Year Award,” say the hardcore software developers now.
 
Well, that wasn't my reaction. My reaction was, I wish you had called me. LOL.
 
>> And yes, a driver isn’t the biggest deal in the world. But with this tiny piece of software, two things should be noted:
>>

• >> 80% of our tech support is for Windows driver issues.
• >> The have been updated no less than 6 times since we (finally) launched Bifrost in October 2011. The biggest change, of course, came when Windows 8 was released.

 
Really, it's not just you. It's not just device drivers that wreak havoc. It's software in general. No matter how wonderfully you design a Bifrost, a Modi, or what have you, eventually there will be a piece of software necessary to make Widget X work. When you need a piece of software to make Widget X work, all of the pitfalls, shortcomings, weirdness, etc of software come into play. It's not your fault. Really. It's the nature of the beast. Of the big companies I've worked for, those that misunderstand how to make software have the greatest difficulty coping with all the weirdness. It's only when you employ the true geeks who really understand the weirdness, and let the geeks take over your software development in it's entirety, that things start working properly. Hardware engineers and Software engineers are two different breeds. They should not be mixed.
 
>>Mike sighed. “Who’s going to do the tech support?” >>
>>“Me, for now.”
>>
>>“You’re going to want to shoot yourself,” Mike predicted.
 
Mike seems to be right, frequently. You should listen to him.
 

 

[Silent One]

 

I know you have your tongue in your cheek. Still have to say - old science does not equal bad science. I'm pretty sure the SR-71 Blackbird from late '50's or early 60's still holds the speed record.

The publicly released record anyways, top speed is still classified. Even if it's not the fastest the capabilities it has is amazing.

 
 
Developed by "Slide Rule" is equally amazing.
 

 

 

[kothganesh]

Developed by "Slide Rule" is equally amazing.

 

LOL. And using vernier calipers for tolerances! Old science to the fore!

 

[judmarc]

   
The hope is, of course, that others will follow suit if Apple jumps in with high-res. Of course, the same caveats about recording quality will apply (thank you, Jud.)
 
With respect to DSD, Mike's opinion is that it is a mathematically compromised format that may sound different, but not necessarily better, than PCM. Mike's opinion is not necessarily popular, but it is shared by others, such as Ayre, Benchmark, and Linn. Jud and Mike will probably have to discuss that amongst themselves, but the reality (as I understand it) is, even if analog to digital converters (ADCs) used in recording use a sigma-delta modulator (SDM) at the input, they usually output PCM, and mixing and mastering is usually done in PCM, and many DACs that accept DSD streams convert them internally to some multilevel amalgamation of PCM and SDM before output. Which makes the "DSD is a less convoluted playback method" assertion more contentious. Of course, I could be mistaken, so I leave that to Mike.
 
I know that Mike's ideal recording and playback chain would be a true multibit ladder ADC outputting PCM and a true multibit ladder DAC accepting PCM. Of course, there are many challenges to this, not least of which is that reaching acceptable numbers in the "bit wars" (that is, 24 bits) on a ladder ADC is pretty much a no-go. But I know Mike would bet that if we had, say, a 20-bit ladder ADC and 20-bit ladder DAC running at 96kHz as the start- and end-point of the recording and playback chain, the game would be very different.
 
My opinion on DSD? Let me put on my pragmatic hat first. If it ever becomes a significant part of the library of available recordings, sure, you bet, we'll support it across the line, or create a device that makes supporting it seamless--and we'll do it right, with no SDM-PCM conversion amalgamation.
 
That said, in my marketing opinion, DSD only fuels the perception that "us audiophiles are a weird bunch, might as well stick with what iTunes is selling." Fighting over formats is really silly. All I want is good music, I don't care so much about the format. And there won't be any large-scale adoption of high-res until we make it (a) simple, and (b) understandable. This is what Neil Young is trying to do, and it's rumored that Apple may also get in with their own combination of black/white, low-res/high-res simplicity which could make for widespread adoption, or at least awareness. Which could change the rules entirely.
 
Also, when we talk DSD, we need to temper our enthusiasm by considering the large storage requirements, the fact that its filtering is drastically different than PCM (and therefore may create a DAC that is good at one thing, but not another, or at least increase cost considerably) and that no non-standard audio format has ever reached critical mass in terms of the number of recordings available, to become a viable, long-term competitor to the mainstream. As Mike says, "How are your HDCDs doing today?"
 
Again, not a popular position to take. But it does fuel additional discussion, right?

 
If I ever get a chance to discuss any of this with Mike, I hope I'll have the good sense to listen and let him do the talking.  I'm strictly a layperson, not an engineer, who's very, very curious about all this stuff and grateful for the opportunity to learn from folks like you and Mike.
 
- Audio systems are complicated stuff.  There's a problem in classical physics called the "three-body problem."  If you've got two bodies - say a star and a planet - you can model it in a stable way, so the planet will keep on happily orbiting the star in the same old orbit for the model's version of eternity.  As soon as you introduce a third body, such as another planet, there's no way to build a stable model.  One or both of your planets will eventually fly away or fall into the sun, or the planets will crash into each other.  Now think of how many boxes you have in your audio system, and how many different circuits are in each, and the interconnections and power cords and relationship to your house wiring and the electrical supply to your home....
 
To me this says there's no One Solution to Rule Them All.  I've heard music sound good coming from an LP, a CD, a CD rip, a variety of higher resolutions, and DSD.  There are incredibly smart people I respect a lot, like Mike, who think DSD is a load of crap foisted on us by Sony (the worst thing Mike ever said about DSD is probably a whole lot nicer than the nicest thing Charles Hansen of Ayre ever said about it), and there are really smart folks who think it's great.  And sometimes the folks who criticize DSD and those who say it's great are the same person!  Bruno Putzeys, who's been quoted criticizing DSD (though I think it's not quite as critical if read in context), makes one of the most highly thought of ADCs used to make DSD recordings.
 
- I mentioned all the conversions our music goes through to show there are technical considerations regarding high res and DSD that don't boil down to "I want something coming out of my speakers that only my dog can hear."  But Jason's absolutely correct to point out that I oversimplified things as well.  Any DSD recording that's had processing done to it - meaning pretty much all of them - had that processing done in PCM.  And yeah, we are no longer talking about classic "1-bit" stuff coming out of the ADC or going into the DAC.  And though using a DSD-type format throughout may avoid a conversion step or two between ADC and DAC, the reason these types of chips pretty much took over the world of ADCs and DACs wasn't in order to avoid conversions, or because DSD was supposed to sound great, or even Sony's efforts (I think where DSD and SACD are concerned, Sony is still looking for more feet to shoot itself in); it's because the chips were a lot cheaper.  It always comes down to the money.  So there's really not much you've said about DSD particularly or high res in general I would want to argue with, except (here's the part where I go rushing in where angels fear to tread):
 
- Back where I said I don't believe there's One Solution to Rule Them All, I think the corollary is you can find great music in lots of places, and I for one would be sorry to see a possibly premature standardization limit that opportunity.  People like Keith Johnson and Barry Diament (responsible for the best Bob Marley remasters you can find) say "4x" rates (176.4, 192) really begin to sound to them like what they hear from their mic feeds in the recording studio, a different level from lower resolutions.  So what if this year we standardize on "2x" rates (88.2, 96) to avoid screwing around with Windows drivers, and a month later Microsoft finally includes native USB2 drivers in Windows like Apple and Linux have had for years?
 
As far as the perception that audiophiles are a weird bunch - we may as well decide to own that one forever, cats and kittens, 'cause that's never going away, not even if you practice listening to hi-res on great headphones while wearing a University of Georgia cap, holding a beer, and hollering "How 'bout them Dawgs?!" all at the same time.  The folks who think audiophiles are weird couldn't tell DSD from an STD, they don't want to have to care about it, and you can't make them.  So just put on a killer hi-res track at your next party (yeah, they're out there, hopefully more of 'em if Pono takes off or Apple decides to get in), hand 'em a beer, and let 'em be happy.  They don't need to know about sample rates.

 

[baronkatz]

Mike seems to be right, frequently. You should listen to him.

Except for the USB implementation and 24/192 support, without which none of the DAC's would have been as succesful as they are... 90% of my headphone use is at work and my work computer does not have optical output. If I use the audio out on it directly to my headphones, I hear coil wind. This is one of the main reasons I even use a USB DAC and I think many people are in the same boat as me. 
 
Mike is a really smart engineering guy with a great sense of humour, and Jason has the marketing background needed to know what will sell... I don't think that they would be able to do it without each other. It's a kind of balancing act between the two of them. A kind of Yin and Yang without which Schiit would not work as a company.

 

[superjawes]

- Audio systems are complicated stuff.  There's a problem in classical physics called the "three-body problem."  If you've got two bodies - say a star and a planet - you can model it in a stable way, so the planet will keep on happily orbiting the star in the same old orbit for the model's version of eternity.  As soon as you introduce a third body, such as another planet, there's no way to build a stable model.  One or both of your planets will eventually fly away or fall into the sun, or the planets will crash into each other.  Now think of how many boxes you have in your audio system, and how many different circuits are in each, and the interconnections and power cords and relationship to your house wiring and the electrical supply to your home....

Physicists are purists. We electrical engineers cheat

See, instead of modeling EM field and wave equations, we use a circuit model, which is more accurately called a "small circuit model," which just means that as long as frequencies are low enough for the circuits to settle, we can ignore a lot of that complicated math and use a simple model. On top of that, when we aren't using CAD tools, we use approximations on top of the circuit theory, especially for semiconductor devices (example: diodes actually have an exponential relationship between the voltage across them and the current they pass, but we usually just say that they "turn on" at 0.7V and treat them as a short circuit).

So while your complete audio setup might add a lot of variables, we can point to several places and either say "close enough" or "here are the limits/measurements" or "here is the equivalent circuit." That makes things much easier to design and analyze

 

[Xcalibur255]

All the best things in audio just might have been conceived in the 1910's.  

  The 45 triodes in my headphone amp are a testament to that.

 

[cjl]

  The publicly released record anyways, top speed is still classified. Even if it's not the fastest the capabilities it has is amazing.

I know this is a popular viewpoint, but the full flight manual (including performance charts) is declassified and available online. Top speed is limited by a compressor inlet temp of 427C, which falls somewhere around mach 3.0 to 3.5, depending on ambient conditions (primarily the air temperature at ~75-85k feet). Here's the manual - it can be very interesting to look through (and it was definitely an astonishing aircraft, and it would still be even if it were made today, much less in the 60s...)
 
http://www.sr-71.org/blackbird/manual/
 
Here's the speed chart: http://www.sr-71.org/blackbird/manual/5/5-9.php
Here's another chart showing max speed vs ambient temperature (based on the inlet temp capability): http://www.sr-71.org/blackbird/manual/5/5-10.php
 
Note that in extremely cold ambient conditions, it does show a capability for higher than mach 3.3, and it probably could push up as high as mach 3.5 or so if the external conditions were just right, but it isn't something it could do routinely without risking engine damage (the reason for the CIT limitation). In addition, if you do the shock angle analysis for what speed the SR-71 would be traveling when the nose shock starts to encounter the wingtips, you end up with a mach number of around 3.3 or so, which further supports that its true design top speed was in this neighborhood. That having been said, sustained cruise at mach 3.2 is nothing to sneeze at - it's still the fastest manned aircraft ever built that could take off under its own power, and the only other ones that ever got even close could not sustain that kind of speed for very long (as opposed to the SR-71, which could cruise at M3.2 for an entire tank of gas without difficulty).

 

[sludgeogre]

- Audio systems are complicated stuff.  There's a problem in classical physics called the "three-body problem."  If you've got two bodies - say a star and a planet - you can model it in a stable way, so the planet will keep on happily orbiting the star in the same old orbit for the model's version of eternity.  As soon as you introduce a third body, such as another planet, there's no way to build a stable model.  One or both of your planets will eventually fly away or fall into the sun, or the planets will crash into each other.  Now think of how many boxes you have in your audio system, and how many different circuits are in each, and the interconnections and power cords and relationship to your house wiring and the electrical supply to your home....

 
This has to do with physics and the equations for gravity. As far as electronics, it is pretty easy to build a stable model on multiple inputs and outputs. The laws are completely different. You can substitute in a whole lot of stuff that solves for one of your variables, or just solve the system for a small range of variables. It is very easy to then test the model and see if it works then iterate to a solution that provides the required result within a certain tolerance. The last thing is that in a three-body problem with celestial bodies, you don't have control over any of the variables, so you can't set one and see how the others react and then build a simplified system. You can model one where the third body is a calculated perturbation to the two-body system, so you just simplify the third body for a limited set of circumstances.

 

[Serenitty]

I had something vaguely like this a long time ago when I was working at an inductor core manufacturer, I decided one Christmas (I was new and the only guy without vacation) to build myself a 2 way crossover splitting at 100hz 12db slope for my home stereo.  I started doing the calculations and kept having to revise the inductance of the core I needed.  While we manufactured very high value (Q) cores, at 100W assumed they were very quickly magnetically saturated.  By the time I was done I was using two cores per inductor and wound about 96 turns of wire.  The whole crossover, 4 inductors and 8 caps weighs a good 10 pounds.  I don't know how Radio shack can sell that little passive crossover (at the time) and claim it's good for 100W..  Of course I really had no reason to design the crossover for 100W sustained.  I know I've never come near that power.
 
I had a point here...  Oh yeah, every time I ran through the calculations and checked against my "needs" I would find I needed a bigger core, since the numbers failed somewhere.  I learned a lot, which I've completely forgotten at this point...

 

[swspiers]

   
There is a reason why I am so pumped for the Yggy, way more than I am for the Ragnarok, and it's cause of the fact that its going to be a ladder dac and this filter...cant wait to see it! Im thinking it's going to be as advanced as some of the best computer based digital filtering out there, but without needing all the computer horsepower.
 
Now we just need to know if it will have i2s, via either hdmi preferably or rj45 if not...

I totally agree.  The Rag seems like it will be awesome, and I can't wait for real comparisons to, say, the Liquid Gold.
 
But the Yggy is something that I really want. Now.

 

[azteca x]

  Oh, and Yggy's digital filter? 18,000+ taps, running a proprietary algorithm based on a 1917 Western Electric paper on time-domain optimization (yes, nine-teen seven-teen, 1917), perfected by a Professor Emeritus of Mathematics at Iowa State (to get around the divide-by-zero problem) and implemented by a RAND Corp mathematician. 
 
The result? The biggest, baddest digital filter in the world. The only true closed-form digital filter that retains the original samples. 
 
Combine this with PCM and a ladder DAC, and forget "add the music to the noise, then filter out the noise" SDM approaches. 
 
In our opinion, of course...

 
Oh god.  This is what I'm talking about.  Can't wait.  Rag will be great but I don't think I need anywhere near that much power.  the DAC though...lord!

 

[judmarc]

   
This has to do with physics and the equations for gravity. As far as electronics, it is pretty easy to build a stable model on multiple inputs and outputs. The laws are completely different. You can substitute in a whole lot of stuff that solves for one of your variables, or just solve the system for a small range of variables. It is very easy to then test the model and see if it works then iterate to a solution that provides the required result within a certain tolerance. The last thing is that in a three-body problem with celestial bodies, you don't have control over any of the variables, so you can't set one and see how the others react and then build a simplified system. You can model one where the third body is a calculated perturbation to the two-body system, so you just simplify the third body for a limited set of circumstances.

 
Yes, that's why it's so easy to design electronics so systems are immune to elementary problems like ground hum.
 
(OK, my apologies.)
 
But more seriously, have a look at Rocky Mountain Audio Fest videos of presentations by people like Keith Johnson, and think about how he takes a much more (literally) "outside the box" view of common problems with audio systems, focusing on system-wide interactions.  Or consider the age-old argument about whether power cords matter.  You can look at the power cord alone and be stuck in the same old back-and-forth: "But I hear a difference!"  "But you can't from 3 feet of wire!"  Or you can look at it from a system perspective and with a couple of bucks worth of parts you can make your transformer much less subject to "ringing" behavior and thus tremendously reduce or eliminate any sonic difference from power cords.  (Google "snubber circuit," "Hagerman," and "John Swenson.")