FLTWS
Headphoneus Supremus
LOL, didn't even realize that myself until you pointed it out.
Schiit Happened: The Story of the World's Most Improbable Start-Up
- Thread starter Jason Stoddard
- Start date
-
- Tags
- schiit audio schiit-audio
bigro
1000+ Head-Fier
- Joined
- Dec 18, 2014
- Posts
- 1,056
- Likes
- 538
artur9
Headphoneus Supremus
- Joined
- Apr 25, 2015
- Posts
- 3,817
- Likes
- 10,547
According to Wikipedia
That does not bode well for keeping the critters away. She'll lure them in with her siren song....
http://www.playfulkitty.net/2013/12/09/cats-in-history-freyas-kitties/
Billheiser
Headphoneus Supremus
Never mind. Just grabbed it and forgot I had cannibalized it for another purpose.
bigro
1000+ Head-Fier
- Joined
- Dec 18, 2014
- Posts
- 1,056
- Likes
- 538
Great, I Hope My Dog Does does not think the Freya has cats Inside.
InsanityOne
500+ Head-Fier
- Joined
- Jan 30, 2014
- Posts
- 517
- Likes
- 74
The proper cage for cat owners worried that their felines might touch their tubes
Haha my feline never has never gotten near my Mjolnir 2, and I even have socket-savers on it so the tubes stick up out of the chassis. This could also be because I have my Mjolnir 2 up on my desk on top of its own rack and I don't allow the cat to get on my desk. (Cat hair and keyboards just don't mix.)
- InsanityOne
Jason Stoddard
Sponsor: Schiit Audio
2016, Chapter 16:
Even More Fulla It
Ah, the poor, forgotten Fulla. It’s amazing to me that over 2 years have gone by since we introduced (what we thought was) a giant-killer: a powerful, good-sounding dongle-DAC with an analog volume pot, available at the then-crazy price of $79.
Doesn’t sound so giant-killerish? Well, considering that dongles were still selling for $199 and up at the time, it was a fairly big deal.
But the world moves on. Now, companies like Audioquest have great-sounding dongles for only $99—dongles with their own in-house, low-power USB interface (which makes connecting to smartphones a lot simpler). At the same time, other, less…ahem, upstanding companies are busy ripping off their earlier work and selling it for even less than the original Fulla. And if you move up the chain, there’s no shortage of dongle options that offer even more power and higher performance than Fulla could ever imagine.
In short, it’s a different world, only two short years later. The market has become dongle-saturated. For those who like to add batteries for a portable solution, there are plenty of options there as well—many with screens and storage. In fact, that whole market is moving so fast, it can be difficult to keep up with all the new options.
Call me old-fashioned, or call me non-competitive, but all those options told me one thing: Fulla had to die.
Or, it had to be reborn…as something entirely new.
A Fateful Conversation
About 18 months ago, I had an interesting (and insightful) conversation with a Schiit fan at TheShow Newport. He really liked the Fulla…but he wanted it to be something very different.
“It should have way more functionality,” he told me.
I nodded. Of course, I’d seen dongles with more functionality—preamp outs, gain switches, things like that. But Fulla was already bumping up against costing too much to produce. If we added functionality, it would have to justify a higher cost.
“Like a preamp out,” I offered.
“Or a fixed out,” he said. “So you can use it as just a DAC—a fixed 2V out for your main system, so you don’t have to worry about where the volume is set.”
I frowned. If someone plugged a set of earbuds into a fixed 2V out, that might not end up so well, not even with resistors in line to raise the output impedance.
He saw my uncertainty and added, “Of course, you could have a preamp out too. And maybe an analog in to go with it.”
I laughed. The mental image of a tiny box covered in RCAs wasn’t pretty. “That’s a totally different product.”
“So?”
“So it’ll cost more,” I said.
“If it’s within reason, I’ll pay it,” he told me.
“Yeah, and that’s the problem: if it’s within reason,” I said. “I don’t know if it’ll be within reason. I’d have to do some design to figure out what it would cost.”
“I’m sure it wouldn’t be much more,” the guy told me.
“No clue,” I said, knowing that seemingly small changes on a chassis could cause significant price differences...not to mention the fact that I wasn’t sure that any part of it was a good idea. Even 6 months after launch, we were already seeing price erosion in the dongle-DAC space.
And that’s pretty much where I left it for another year. During that time, dongles got cheaper and better. And Fulla went nowhere, of course. It couldn’t. We had a fixed design, and a fixed number of parts and chassis—a very large number of parts and chassis.
Worse, we began seeing some of the uglier aspects of dongles as Fulla got older. They are tiny products. It’s easy to crush one. Or to break off the small potentiometer. Or have issues with the USB connector. The failure rate wasn’t higher than industry standard, but it was higher than our standard.
But eventually, Alex came to me and said, “It’s time to make a decision on the next Fulla.” The look on his face seemed to say, And I hope it disappears.
Death or Rebirth?
And Alex had a point. A perfectly rational answer to the question of the next Fulla would be, “There’s no next Fulla,” and call it done. Fulla wasn’t lighting up the sales charts, there were lots of options priced near it, and it would take some pretty serious design work to make something that stood out in a crowded market.
And that’s almost what we did. Like I said, it made sense. We didn’t really need Fulla.
But…
Spending time on the forums, reading about people’s experiences with Schiit products (and others), and making sure to pay close attention to people’s first audiophile purchases, I knew there was something we were missing. Specifically, a one-box, flexible product at a price significantly less than a Magni and Modi combo.
Aside: for those of you chuckling about how something should cost less than a Magni and Modi, give thanks that life is good for you. The reality is that $198 is a significant chunk of money for a lot of people. It’s even more if you’re going to school and don’t have a full-time job, or if you are just out of school and staring down the barrel of some serious student loans. In addition, the various connections, cables, etc needed to get a system working can be confusing when you’re first getting into it as well—and it doesn’t take much confusion for a sale to turn into an abandoned cart.
So what price point would make sense for such a product? Significantly less needed to be viscerally less. Really real. As in, not $169 or $149 or even $129.
But $99? Now you’re talking.
And that’s when that conversation came back to me. Not just a $99 DAC/amp, but a $99 do-all DAC/amp. That would be interesting. Especially if it could be quite a bit more powerful than the original Fulla…and use the great new top-line AK4490 DAC.
Aside: people who have followed my blatherings to date know that everything after the “but” in the paragraph above is called “ego talking,” and it’s usually a great way for designers to get themselves in big trouble.
However, there was still one big problem: the fact that one of the reasons dongles had changed was that sources had changed—in that many of them wouldn’t supply the 500mA (or more) that such a powerful device would require.
But I put the misgivings about power away for the moment. Because sometimes the best thing to do is simply to build it, and see how well it worked. Even if it ended up being just a desktop device, even if it had problems with some low-powered USB ports, it was worth exploring. I decided to go ahead, do the design and make a prototype—and sort out the whole power situation later.
And that’s where I was in Spring 2016: with a basic idea (for a new Fulla), a decision on direction (it’s a full-featured desktop product now, damnit, not a dongle) and a looming problem: how to interface it easily with phones and such.
Good enough to start. I figured there would be other problems to work out—and there were…
Decisions, Decisions
Okay, so the second-generation Fulla would be designed for a desktop. But how big would it be? The original Fulla was a tiny product. It had to be bigger than that. At the same time, it should be smaller than the Magni and Modi stack. In the end, I chose a footprint exactly half the size of a Magni/Modi stack—and shaved off a quarter inch of height. So, 2.5” wide, 3.5 deep, and 1” high.
A mockup looked pretty cool. It even could use the same potentiometer and the same volume knob as a Magni, which was a big plus. The potentiometer on the Magni was a lot better than the one in the original Fulla in terms of tracking and durability. And the knob would save us having to do another custom part.
With the size decided, I needed to make some decisions on features. First up was how to get more power out of the product. Fulla’s 200mW output into 32 ohms wasn’t really worth getting excited about. It was plenty for most headphones, yes, but I wanted to do better.
Doing better, though, meant both a better power supply and a more beefy amplifier stage. This wasn’t as simple as it might sound, though, because unlike all of our other amps, Fulla uses a switchmode power supply, and op-amp output stage.
Horrors! Some of you are clutching your pearls and looking for a safe space. How can Schiit, bastion of linear power supplies and discrete design, sully ourselves with such inferior technology?
Well, here’s the super-complex rationalization for such a drastic decision: price.
Yes. It is that simple. Horses for courses. When you’re asked to create a power supply that can deliver a reliable +/-5V from USB’s single 5V rail, and to make the most of the 500mA-1A that USB ports can supply, you’re looking at a switchmode supply. Period. End of story. No magic fairy dust or wishful thinking will get you around that fact.
Now, of course, we could do much better than the previous Fulla’s power supply, which used a charge pump to create the -5V rail, but passed the +5V along from USB (after filtering). This asymmetrical arrangement meant that the -5V rail couldn’t deliver the same amount of current as the +5V rail, and it was the main reason why the original Fulla’s output power was limited. With more space in Fulla 2, we were able to use a symmetrical rail generator with a better, inductor-based switchmode supply for each rail, delivering more than 3X the current of the original Fulla’s supplies. With some filtering, it’s a good, solid, quiet power supply.
Similarly, when you’re asked to create an output stage half the size of a postage stamp that can swing the output from rail to rail, you’re not going to do it with discrete parts. Enter op-amps. Specifically, the same exotic, high-current op-amps we used with the first Fulla, the LMH6643. This fast, high-current, rail-to-rail device has only one downfall—its current output is still less than we’d like. So, in Fulla 2, we simply used two of them per channel to double its output current capability.
Now, we have an amp stage that can deliver 550mW into 16 ohms—much better. Still not a Jotunheim, or even a Magni, but there you go. Sometimes you have to know when to say “When.”
Beyond power output, though, what kind of I/O should Fulla 2 have? I waffled a bit, wondering about having just a preamp out, or not having an analog input, but in the end, I threw them all in: fixed and preamp outs, as well as an analog in. Fixed so you could use it as just a DAC, variable so you could use it as a preamp (it amplifies the analog input, too), and an analog input in case you have a phone or something (that still has a 1/8” output) that you’d like to connect to Fulla.
The catch to all of this I/O? It all has to go in and out on 1/8” jacks. Yeah, I know. But 1/8” to dual RCA cables are easily available.
Bluetooth? Nope. Yes, it’s convenient, and yes, it sounds like ass. Yes, I know, it doesn’t sound completely like ass, but it doesn’t sound as good as a cabled connection. Buy a cable. Don’t be lazy.
Aside: I have been told by friends at very large audio companies that Bluetooth is also still (a) not entirely reliable, (b) quite variable in terms of quality as the data rate changes over distance and direction. I also know how much it costs to implement and to qualify. Fulla 2 wouldn’t be a $99 product with Bluetooth. So, given the choice of adding an ass-sounding input that would cause additional customer consternation and add a lot of cost, we decided to pass.
Aside to the aside: when Bluetooth can reliably pass an uncompressed 16/44.1 stream, we’ll revisit this decision…we’re stubborn, but not crazy.
For headphone output, I went with a full-size ¼” TRS jack, because, let’s face it, most serious full-size headphones still use this connector. And Fulla’s intended to be a serious product. And 1/8” to 1/4” adapters are common. And yeah, I know, I heard from tons of people who said, “But you can do balanced output on TRRS for even more power!” But that’s also a recipe for frying outputs or frying headphones when someone ignores all the warnings in the manual and the big warning sticker and giant plug in the TRRS jack and plugs a TRS headphone into it anyway. Or vice-versa. No thanks. Shoot us for playing it safe in an affordable product.
And that’s where we ended up with the first prototype. It felt pretty good. It looked pretty cool. I threw together a board and sent it out. After it came in, I sent the board and a bag of tiny parts out to our PCB assemblers. And when it came back, I was eager to hear how it sounded.
Unfortunately, it was a bag of crap.
Lessons in Tiny Parts
Designers working in high-end audio have it easy. We get to work with parts that most other electronics engineers would call “rustic.” They’re big. Easy to handle. Lots of them still have leads on them. You don’t have to worry about them blowing away when you sneeze.
Hell, even when Schiit uses surface-mount parts, we’re using big parts—like 0805 resistors and old-skool SOICs. Yeah, I know, you can stop laughing. Because even in our realm, tiny parts are creeping in, together with parts that are simply impossible to solder by hand (parts with pads underneath them, for example.)
And, in the case of Fulla 2, we were still sufficiently size-constrained that it has a lot of 0402 resistors, as well as some exotic stuff like 4x4mm parts with power pads and 24 pins, as well as 100-pins QFNs and…well, let’s just say that it isn’t what anyone wants to put together by hand. That’s why the prototypes go to our PCB assemblers, rather than to me or Dave.
And when you screw up a bunch of traces going to the new QFN DAC, it’s kinda hard to fix it.
In fact, it’s so hard, we gave up. I tried it, Dave tried it, our assembly house tried it…and none of us could de-screw what I had screwed up. So that was the ignoble end of the first prototype: it never worked, and ended up dead in the trash.
While I was fixing my (entirely avoidable) board screw-ups, though, I got to thinking about the USB power problem. Because Fulla 2 would end up drawing even more current than the original. It would be a 100% no-go with iPhones, iPads, many Android phones, and some power-limited USB ports. And that wasn’t ideal.
But now, I thought I saw a way around it. Since we were using a rail-generator chip to do both the +5V and -5V rails, and since it operated down to 3V or so, we didn’t need to have the full USB 5V into it. Which meant we could, theoretically, simply add another micro USB input that provided power only…which meant that the main USB port wouldn’t need to draw any power at all…which meant that our power problems were solved.
Except for one problem: USB power reporting.
You see, USB ports don’t (usually) measure the amount of power a device draws. They rely on the device to tell it how much current it needs. USB devices can request 0mA (no power), 100mA (low power) or 500mA (full power).
Aside: yes, I know, this is monumentally more complex now thanks to USB 3.1, but let’s leave that alone for now. The example I’m using is relevant to audio devices, so roll with it.
Fulla 2, like Fulla, reported as a 500mA device. Which meant trouble with a whole lot of products. Apple iOS devices refuse any device that wants more than 100mA. Some USB ports claim to be able to provide full power, but they really don’t. Some Android phones will provide 500mA for a while, but not when the battery falls below a certain point.
But the problem was, even with a separate power input jack for Fulla, it wouldn’t matter—because it would still report as needing 500mA.
Unless there was a way to customize it to detect where the power was coming from, and report as 0mA if the second USB input was used.
And if we could do that, it would be perfect! You could just use something like a phone charger for power if your USB port couldn’t run the device. Or, even without a compatible phone charger, 5V AC adapters with micro USB out were a dime a dozen.
I knew it could be done. The CM6631A USB receiver we use has many general purpose I/O ports (GPIO). If we could pull one high when the auxiliary power was connected, and if we could alter the firmware to change the reporting, we’d really have something.
But I’m getting ahead of myself. Because I also have to talk about the knob.
The knob. Yes, the most recognizable aspect of Fulla 2. The big vertical knob. At the same time I was fixing the board and figuring out how to do power switching, I was also becoming unhappy with the cosmetics. Because Fulla 2 looked OK, but it wasn’t really all that interesting. And the small knob and shorter chassis made it a little awkward. I wondered if we could do something about that.
A little research, and I found out that yes, we absolutely could. Alps also made a vertical version of the pot we used in Magni, the same exact part with the same exact quality. And the guys who make our knobs also make aluminum shell knobs of the same exact style in much larger diameters. Put the two together, and suddenly Fulla 2 looked pretty cool…and it was exactly in keeping with the original, which also had a knob on top.
With these changes made—but no definitive answer on whether or not we could have the USB-reporting-altering firmware in place—it was time to go for a second prototype.
And We’re Golden
Why go for a second prototype when we didn’t have the power-switching in place yet? It’s simple:
And so, we got a second prototype back…and this one worked great. It even worked when powered by the auxiliary USB jack. Unfortunately, the CM6631A still reported as a 500mA device, even when the power USB was connected.
Could we make the power-switching work? In my mind, this was the difference between a good and great product. Fulla 2 would be plenty good if it moved the cost of entry down, but had problems with low-powered USB ports and we had to tell people to use a powered hub or other workarounds to make it work. But it would be a great product if we were able to say, “Low power USB port? No problem. Just plug in your phone charger and we’re all good.”
Long story short, with C-Media working on the firmware, and a couple of hardware tricks, we were able to do exactly that. Plug Fulla 2 into your computer, and it reports as a 500mA device. Plug a separate power USB into Fulla 2, then plug it into your computer, and it reports as a 0mA device.
Of course, that took a couple of hardware hacks…so with one more prototype under our belt, it was time to move on to production.
As (seems to be) usual these days, the first articles were uneventful. They worked as expected, tested well, and the new firmware performed like a champ. Alex pronounced them good. Tyler (our head of finance and admin…think of him as Hermes from Futurama, a good bureaucrat—hough he has yet to show up in a flying desk) said, “Wow, the bass is really good on this, I’m going to buy one as soon as they’re out.”
That puzzled me a bit. “But you already have, like, a Lyr 2 and a Bifrost and other stuff, right?”
Tyler nodded. “Yeah. But this is tiny. It’s great for a small desk. And it sounds good.”
I grinned. That was exactly the response I wanted. A small, do-all product that anyone could be happy with. Everything seemed good for the launch.
Except…
Except sometimes there are idiots. And sometimes those idiots is me.
I’m an Idiot
About a week before we were scheduled to announce Fulla 2, Alex told me, “Wow, this is a really great device. But I thought the preamp out was supposed to be variable.”
Deep in Vidar guts, I didn’t even look up. “Of course it is. There’s a fixed output too.”
There was a long pause. Finally, Alez said, “Yeah. The outside one. But it’s variable.”
Ah, ****.
“Wait. Are you saying the screen is wrong?”
“Um, well…”
“Check it out, please. And let me know.” Argh. If it was wrong, we had a whole bunch of metal that was screened wrong. Which meant we couldn’t ship…
I realized I could check myself by looking at the PCB layout. A few seconds later, I was staring at a layout that said, Yep, you’re a bonehead, you reversed the screen.
Crap.
What to do? Push the launch, of course. Call the metal guys and see what we could do. And scrap the manuals, because they were done for the wrong screen.
Another long story short, our metal guys proved themselves again. They had a batch of parts in process that they could screen and get to us in a few days. This would hold us until they could refinish the previous run. I got ahold of our printer and got them new artwork for the manual, which we’d have in a few days as well.
Crisis averted, but yeah, sometimes when you get too cocky, you get taken down a peg or two. This was a reminder: stay humble.
(And stay silent—as you’ve already seen, my confident prediction of shipping Freyas in October turned out to be a crock as well. We’ll see about Vidar. As far as the products after that, well, I’m keeping my mouth shut.)
In the end, everything worked. Now, the cost for a do-all desktop system is $99, not $248 (remember, you need Magni 2 Uber for preamp outputs). Hopefully this will bring even more people into the market.
Will the market change significantly in the next couple of years, obviating the Fulla 2? I don’t know. I don’t have a crystal ball.
And, in the end, only one thing is important: what you think.
Even More Fulla It
Ah, the poor, forgotten Fulla. It’s amazing to me that over 2 years have gone by since we introduced (what we thought was) a giant-killer: a powerful, good-sounding dongle-DAC with an analog volume pot, available at the then-crazy price of $79.
Doesn’t sound so giant-killerish? Well, considering that dongles were still selling for $199 and up at the time, it was a fairly big deal.
But the world moves on. Now, companies like Audioquest have great-sounding dongles for only $99—dongles with their own in-house, low-power USB interface (which makes connecting to smartphones a lot simpler). At the same time, other, less…ahem, upstanding companies are busy ripping off their earlier work and selling it for even less than the original Fulla. And if you move up the chain, there’s no shortage of dongle options that offer even more power and higher performance than Fulla could ever imagine.
In short, it’s a different world, only two short years later. The market has become dongle-saturated. For those who like to add batteries for a portable solution, there are plenty of options there as well—many with screens and storage. In fact, that whole market is moving so fast, it can be difficult to keep up with all the new options.
Call me old-fashioned, or call me non-competitive, but all those options told me one thing: Fulla had to die.
Or, it had to be reborn…as something entirely new.
A Fateful Conversation
About 18 months ago, I had an interesting (and insightful) conversation with a Schiit fan at TheShow Newport. He really liked the Fulla…but he wanted it to be something very different.
“It should have way more functionality,” he told me.
I nodded. Of course, I’d seen dongles with more functionality—preamp outs, gain switches, things like that. But Fulla was already bumping up against costing too much to produce. If we added functionality, it would have to justify a higher cost.
“Like a preamp out,” I offered.
“Or a fixed out,” he said. “So you can use it as just a DAC—a fixed 2V out for your main system, so you don’t have to worry about where the volume is set.”
I frowned. If someone plugged a set of earbuds into a fixed 2V out, that might not end up so well, not even with resistors in line to raise the output impedance.
He saw my uncertainty and added, “Of course, you could have a preamp out too. And maybe an analog in to go with it.”
I laughed. The mental image of a tiny box covered in RCAs wasn’t pretty. “That’s a totally different product.”
“So?”
“So it’ll cost more,” I said.
“If it’s within reason, I’ll pay it,” he told me.
“Yeah, and that’s the problem: if it’s within reason,” I said. “I don’t know if it’ll be within reason. I’d have to do some design to figure out what it would cost.”
“I’m sure it wouldn’t be much more,” the guy told me.
“No clue,” I said, knowing that seemingly small changes on a chassis could cause significant price differences...not to mention the fact that I wasn’t sure that any part of it was a good idea. Even 6 months after launch, we were already seeing price erosion in the dongle-DAC space.
And that’s pretty much where I left it for another year. During that time, dongles got cheaper and better. And Fulla went nowhere, of course. It couldn’t. We had a fixed design, and a fixed number of parts and chassis—a very large number of parts and chassis.
Worse, we began seeing some of the uglier aspects of dongles as Fulla got older. They are tiny products. It’s easy to crush one. Or to break off the small potentiometer. Or have issues with the USB connector. The failure rate wasn’t higher than industry standard, but it was higher than our standard.
But eventually, Alex came to me and said, “It’s time to make a decision on the next Fulla.” The look on his face seemed to say, And I hope it disappears.
Death or Rebirth?
And Alex had a point. A perfectly rational answer to the question of the next Fulla would be, “There’s no next Fulla,” and call it done. Fulla wasn’t lighting up the sales charts, there were lots of options priced near it, and it would take some pretty serious design work to make something that stood out in a crowded market.
And that’s almost what we did. Like I said, it made sense. We didn’t really need Fulla.
But…
Spending time on the forums, reading about people’s experiences with Schiit products (and others), and making sure to pay close attention to people’s first audiophile purchases, I knew there was something we were missing. Specifically, a one-box, flexible product at a price significantly less than a Magni and Modi combo.
Aside: for those of you chuckling about how something should cost less than a Magni and Modi, give thanks that life is good for you. The reality is that $198 is a significant chunk of money for a lot of people. It’s even more if you’re going to school and don’t have a full-time job, or if you are just out of school and staring down the barrel of some serious student loans. In addition, the various connections, cables, etc needed to get a system working can be confusing when you’re first getting into it as well—and it doesn’t take much confusion for a sale to turn into an abandoned cart.
So what price point would make sense for such a product? Significantly less needed to be viscerally less. Really real. As in, not $169 or $149 or even $129.
But $99? Now you’re talking.
And that’s when that conversation came back to me. Not just a $99 DAC/amp, but a $99 do-all DAC/amp. That would be interesting. Especially if it could be quite a bit more powerful than the original Fulla…and use the great new top-line AK4490 DAC.
Aside: people who have followed my blatherings to date know that everything after the “but” in the paragraph above is called “ego talking,” and it’s usually a great way for designers to get themselves in big trouble.
However, there was still one big problem: the fact that one of the reasons dongles had changed was that sources had changed—in that many of them wouldn’t supply the 500mA (or more) that such a powerful device would require.
But I put the misgivings about power away for the moment. Because sometimes the best thing to do is simply to build it, and see how well it worked. Even if it ended up being just a desktop device, even if it had problems with some low-powered USB ports, it was worth exploring. I decided to go ahead, do the design and make a prototype—and sort out the whole power situation later.
And that’s where I was in Spring 2016: with a basic idea (for a new Fulla), a decision on direction (it’s a full-featured desktop product now, damnit, not a dongle) and a looming problem: how to interface it easily with phones and such.
Good enough to start. I figured there would be other problems to work out—and there were…
Decisions, Decisions
Okay, so the second-generation Fulla would be designed for a desktop. But how big would it be? The original Fulla was a tiny product. It had to be bigger than that. At the same time, it should be smaller than the Magni and Modi stack. In the end, I chose a footprint exactly half the size of a Magni/Modi stack—and shaved off a quarter inch of height. So, 2.5” wide, 3.5 deep, and 1” high.
A mockup looked pretty cool. It even could use the same potentiometer and the same volume knob as a Magni, which was a big plus. The potentiometer on the Magni was a lot better than the one in the original Fulla in terms of tracking and durability. And the knob would save us having to do another custom part.
With the size decided, I needed to make some decisions on features. First up was how to get more power out of the product. Fulla’s 200mW output into 32 ohms wasn’t really worth getting excited about. It was plenty for most headphones, yes, but I wanted to do better.
Doing better, though, meant both a better power supply and a more beefy amplifier stage. This wasn’t as simple as it might sound, though, because unlike all of our other amps, Fulla uses a switchmode power supply, and op-amp output stage.
Horrors! Some of you are clutching your pearls and looking for a safe space. How can Schiit, bastion of linear power supplies and discrete design, sully ourselves with such inferior technology?
Well, here’s the super-complex rationalization for such a drastic decision: price.
Yes. It is that simple. Horses for courses. When you’re asked to create a power supply that can deliver a reliable +/-5V from USB’s single 5V rail, and to make the most of the 500mA-1A that USB ports can supply, you’re looking at a switchmode supply. Period. End of story. No magic fairy dust or wishful thinking will get you around that fact.
Now, of course, we could do much better than the previous Fulla’s power supply, which used a charge pump to create the -5V rail, but passed the +5V along from USB (after filtering). This asymmetrical arrangement meant that the -5V rail couldn’t deliver the same amount of current as the +5V rail, and it was the main reason why the original Fulla’s output power was limited. With more space in Fulla 2, we were able to use a symmetrical rail generator with a better, inductor-based switchmode supply for each rail, delivering more than 3X the current of the original Fulla’s supplies. With some filtering, it’s a good, solid, quiet power supply.
Similarly, when you’re asked to create an output stage half the size of a postage stamp that can swing the output from rail to rail, you’re not going to do it with discrete parts. Enter op-amps. Specifically, the same exotic, high-current op-amps we used with the first Fulla, the LMH6643. This fast, high-current, rail-to-rail device has only one downfall—its current output is still less than we’d like. So, in Fulla 2, we simply used two of them per channel to double its output current capability.
Now, we have an amp stage that can deliver 550mW into 16 ohms—much better. Still not a Jotunheim, or even a Magni, but there you go. Sometimes you have to know when to say “When.”
Beyond power output, though, what kind of I/O should Fulla 2 have? I waffled a bit, wondering about having just a preamp out, or not having an analog input, but in the end, I threw them all in: fixed and preamp outs, as well as an analog in. Fixed so you could use it as just a DAC, variable so you could use it as a preamp (it amplifies the analog input, too), and an analog input in case you have a phone or something (that still has a 1/8” output) that you’d like to connect to Fulla.
The catch to all of this I/O? It all has to go in and out on 1/8” jacks. Yeah, I know. But 1/8” to dual RCA cables are easily available.
Bluetooth? Nope. Yes, it’s convenient, and yes, it sounds like ass. Yes, I know, it doesn’t sound completely like ass, but it doesn’t sound as good as a cabled connection. Buy a cable. Don’t be lazy.
Aside: I have been told by friends at very large audio companies that Bluetooth is also still (a) not entirely reliable, (b) quite variable in terms of quality as the data rate changes over distance and direction. I also know how much it costs to implement and to qualify. Fulla 2 wouldn’t be a $99 product with Bluetooth. So, given the choice of adding an ass-sounding input that would cause additional customer consternation and add a lot of cost, we decided to pass.
Aside to the aside: when Bluetooth can reliably pass an uncompressed 16/44.1 stream, we’ll revisit this decision…we’re stubborn, but not crazy.
For headphone output, I went with a full-size ¼” TRS jack, because, let’s face it, most serious full-size headphones still use this connector. And Fulla’s intended to be a serious product. And 1/8” to 1/4” adapters are common. And yeah, I know, I heard from tons of people who said, “But you can do balanced output on TRRS for even more power!” But that’s also a recipe for frying outputs or frying headphones when someone ignores all the warnings in the manual and the big warning sticker and giant plug in the TRRS jack and plugs a TRS headphone into it anyway. Or vice-versa. No thanks. Shoot us for playing it safe in an affordable product.
And that’s where we ended up with the first prototype. It felt pretty good. It looked pretty cool. I threw together a board and sent it out. After it came in, I sent the board and a bag of tiny parts out to our PCB assemblers. And when it came back, I was eager to hear how it sounded.
Unfortunately, it was a bag of crap.
Lessons in Tiny Parts
Designers working in high-end audio have it easy. We get to work with parts that most other electronics engineers would call “rustic.” They’re big. Easy to handle. Lots of them still have leads on them. You don’t have to worry about them blowing away when you sneeze.
Hell, even when Schiit uses surface-mount parts, we’re using big parts—like 0805 resistors and old-skool SOICs. Yeah, I know, you can stop laughing. Because even in our realm, tiny parts are creeping in, together with parts that are simply impossible to solder by hand (parts with pads underneath them, for example.)
And, in the case of Fulla 2, we were still sufficiently size-constrained that it has a lot of 0402 resistors, as well as some exotic stuff like 4x4mm parts with power pads and 24 pins, as well as 100-pins QFNs and…well, let’s just say that it isn’t what anyone wants to put together by hand. That’s why the prototypes go to our PCB assemblers, rather than to me or Dave.
And when you screw up a bunch of traces going to the new QFN DAC, it’s kinda hard to fix it.
In fact, it’s so hard, we gave up. I tried it, Dave tried it, our assembly house tried it…and none of us could de-screw what I had screwed up. So that was the ignoble end of the first prototype: it never worked, and ended up dead in the trash.
While I was fixing my (entirely avoidable) board screw-ups, though, I got to thinking about the USB power problem. Because Fulla 2 would end up drawing even more current than the original. It would be a 100% no-go with iPhones, iPads, many Android phones, and some power-limited USB ports. And that wasn’t ideal.
But now, I thought I saw a way around it. Since we were using a rail-generator chip to do both the +5V and -5V rails, and since it operated down to 3V or so, we didn’t need to have the full USB 5V into it. Which meant we could, theoretically, simply add another micro USB input that provided power only…which meant that the main USB port wouldn’t need to draw any power at all…which meant that our power problems were solved.
Except for one problem: USB power reporting.
You see, USB ports don’t (usually) measure the amount of power a device draws. They rely on the device to tell it how much current it needs. USB devices can request 0mA (no power), 100mA (low power) or 500mA (full power).
Aside: yes, I know, this is monumentally more complex now thanks to USB 3.1, but let’s leave that alone for now. The example I’m using is relevant to audio devices, so roll with it.
Fulla 2, like Fulla, reported as a 500mA device. Which meant trouble with a whole lot of products. Apple iOS devices refuse any device that wants more than 100mA. Some USB ports claim to be able to provide full power, but they really don’t. Some Android phones will provide 500mA for a while, but not when the battery falls below a certain point.
But the problem was, even with a separate power input jack for Fulla, it wouldn’t matter—because it would still report as needing 500mA.
Unless there was a way to customize it to detect where the power was coming from, and report as 0mA if the second USB input was used.
And if we could do that, it would be perfect! You could just use something like a phone charger for power if your USB port couldn’t run the device. Or, even without a compatible phone charger, 5V AC adapters with micro USB out were a dime a dozen.
I knew it could be done. The CM6631A USB receiver we use has many general purpose I/O ports (GPIO). If we could pull one high when the auxiliary power was connected, and if we could alter the firmware to change the reporting, we’d really have something.
But I’m getting ahead of myself. Because I also have to talk about the knob.
The knob. Yes, the most recognizable aspect of Fulla 2. The big vertical knob. At the same time I was fixing the board and figuring out how to do power switching, I was also becoming unhappy with the cosmetics. Because Fulla 2 looked OK, but it wasn’t really all that interesting. And the small knob and shorter chassis made it a little awkward. I wondered if we could do something about that.
A little research, and I found out that yes, we absolutely could. Alps also made a vertical version of the pot we used in Magni, the same exact part with the same exact quality. And the guys who make our knobs also make aluminum shell knobs of the same exact style in much larger diameters. Put the two together, and suddenly Fulla 2 looked pretty cool…and it was exactly in keeping with the original, which also had a knob on top.
With these changes made—but no definitive answer on whether or not we could have the USB-reporting-altering firmware in place—it was time to go for a second prototype.
And We’re Golden
Why go for a second prototype when we didn’t have the power-switching in place yet? It’s simple:
- Moving the cost of entry for good sound from $198 to $99 would be a significant achievement, allowing even more people to hear “what the shouting was about.”
- Even if the power switching didn’t work, it wasn’t like we had anything like that on Modi, and Modi did very well.
- I wanted to get the product out this year, and sometimes, you have to fish or cut bait. This was fishing time.
And so, we got a second prototype back…and this one worked great. It even worked when powered by the auxiliary USB jack. Unfortunately, the CM6631A still reported as a 500mA device, even when the power USB was connected.
Could we make the power-switching work? In my mind, this was the difference between a good and great product. Fulla 2 would be plenty good if it moved the cost of entry down, but had problems with low-powered USB ports and we had to tell people to use a powered hub or other workarounds to make it work. But it would be a great product if we were able to say, “Low power USB port? No problem. Just plug in your phone charger and we’re all good.”
Long story short, with C-Media working on the firmware, and a couple of hardware tricks, we were able to do exactly that. Plug Fulla 2 into your computer, and it reports as a 500mA device. Plug a separate power USB into Fulla 2, then plug it into your computer, and it reports as a 0mA device.
Of course, that took a couple of hardware hacks…so with one more prototype under our belt, it was time to move on to production.
As (seems to be) usual these days, the first articles were uneventful. They worked as expected, tested well, and the new firmware performed like a champ. Alex pronounced them good. Tyler (our head of finance and admin…think of him as Hermes from Futurama, a good bureaucrat—hough he has yet to show up in a flying desk) said, “Wow, the bass is really good on this, I’m going to buy one as soon as they’re out.”
That puzzled me a bit. “But you already have, like, a Lyr 2 and a Bifrost and other stuff, right?”
Tyler nodded. “Yeah. But this is tiny. It’s great for a small desk. And it sounds good.”
I grinned. That was exactly the response I wanted. A small, do-all product that anyone could be happy with. Everything seemed good for the launch.
Except…
Except sometimes there are idiots. And sometimes those idiots is me.
I’m an Idiot
About a week before we were scheduled to announce Fulla 2, Alex told me, “Wow, this is a really great device. But I thought the preamp out was supposed to be variable.”
Deep in Vidar guts, I didn’t even look up. “Of course it is. There’s a fixed output too.”
There was a long pause. Finally, Alez said, “Yeah. The outside one. But it’s variable.”
Ah, ****.
“Wait. Are you saying the screen is wrong?”
“Um, well…”
“Check it out, please. And let me know.” Argh. If it was wrong, we had a whole bunch of metal that was screened wrong. Which meant we couldn’t ship…
I realized I could check myself by looking at the PCB layout. A few seconds later, I was staring at a layout that said, Yep, you’re a bonehead, you reversed the screen.
Crap.
What to do? Push the launch, of course. Call the metal guys and see what we could do. And scrap the manuals, because they were done for the wrong screen.
Another long story short, our metal guys proved themselves again. They had a batch of parts in process that they could screen and get to us in a few days. This would hold us until they could refinish the previous run. I got ahold of our printer and got them new artwork for the manual, which we’d have in a few days as well.
Crisis averted, but yeah, sometimes when you get too cocky, you get taken down a peg or two. This was a reminder: stay humble.
(And stay silent—as you’ve already seen, my confident prediction of shipping Freyas in October turned out to be a crock as well. We’ll see about Vidar. As far as the products after that, well, I’m keeping my mouth shut.)
In the end, everything worked. Now, the cost for a do-all desktop system is $99, not $248 (remember, you need Magni 2 Uber for preamp outputs). Hopefully this will bring even more people into the market.
Will the market change significantly in the next couple of years, obviating the Fulla 2? I don’t know. I don’t have a crystal ball.
And, in the end, only one thing is important: what you think.
mattlach
100+ Head-Fier
- Joined
- Mar 9, 2009
- Posts
- 408
- Likes
- 228
This is one of the reasons I have such great respect for you guys. The fact that you both understand the importance of good entry level products to grow the community, but also that just because you don't have a car-priced DAC, doesn't mean that it should be laughed at. Its relatively easy to build low volume parts and charge exclusivity prices for them. it takes way more work and skill to design something that is not only good, but feasible at a low cost in quantity...
...and made in the U.S.
If it weren't for a cost effective amp from Schiit back in 2011 (that sounded, and still sounds great, btw.), I probably would never have gotten in tot the world of high quality audio.
Thanks for what you guys do, and thanks for listening to your customers and being approachable in a forum like this. I'm sure that - in no small part - is responsible for you guys having such a following.
Just reading this bit makes me remember what MEK smells like. I don't miss that part of that job...
KoshNaranek
Headphoneus Supremus
As always, I love reading the story. Thanks Jason.
However, I would like to state for the record that I don't buy Schiit until after I have seen the circuit board porn!
However, I would like to state for the record that I don't buy Schiit until after I have seen the circuit board porn!
lblb
New Head-Fier
- Joined
- Mar 25, 2015
- Posts
- 30
- Likes
- 113
I incorporated Chapter 16 in the ebook, see my signature.
madwolfa
Headphoneus Supremus
Ahh, just 2 months after I've got my original Fulla. I knew something was coming, but, oh well. 
Astral Abyss
1000+ Head-Fier
My wife said it's cute (looks like a little o_O emoji). Congrats, you've won the war.
bigro
1000+ Head-Fier
- Joined
- Dec 18, 2014
- Posts
- 1,056
- Likes
- 538
As User of the Original Fulla for my work computers that will not allow driver installs because the security Nazi's must not allow us little people to do anything. Oddly I spent part of last night looking for a small USB DAC that did not need drivers so I did not need a second computer on my desk just to stream music from tidal or talk shows from the internet while I worked. I wanted a DAC only so that I can route it through a My Magni 2 U, which also is a preamp to my desktop speaker amp. I Just Bought one. It's Not The Goddess of Cats, But It will take the edge off the wait.
