rovopio
Headphoneus Supremus
- Joined
- Jan 18, 2014
- Posts
- 1,749
- Likes
- 126
i e-mailed them at 12.48 with some questions. Schiit replied at 12.51. Ordered at 1.15. Got tracking number at 2.15.
how amazing is that?
that is what i call the bomb dot com!
Schiit Happened: The Story of the World's Most Improbable Start-Up
- Thread starter Jason Stoddard
- Start date
-
- Tags
- schiit audio schiit-audio
i e-mailed them at 12.48 with some questions. Schiit replied at 12.51. Ordered at 1.15. Got tracking number at 2.15.
how amazing is that?
that is what i call the bomb dot com!
- Joined
- Dec 1, 2013
- Posts
- 378
- Likes
- 85
You will find that Schiit is quite different than other companies. Buying from them is painless.
swannie007
500+ Head-Fier
- Joined
- Oct 30, 2013
- Posts
- 942
- Likes
- 217
+1 the above. Great people to deal with. Merry Xmas and Happy New Year to all from Oz.
RCBinTN
Headphoneus Supremus
I agree. They have great customer service - some of the best I've experienced. How many times do we bad-mouth a company for their poor service. Let's give Schiit Audio credit for their responsiveness!!
Cheers -
RCB
In the past 10 days I have emailed and called and have not got any responses.
They specifically ask not to call them because they do not respond to the phone calls. Try sending another email it probably got lost.
rmullins08
100+ Head-Fier
- Joined
- May 12, 2014
- Posts
- 469
- Likes
- 98
Jason Stoddard
Sponsor: Schiit Audio
Chapter 42:
One Year, Twelve Products
Okay. Let me get this off my chest now: twelve products in one year is too many. Yes, even when half of them were updates, and two of them variations on the updates.
Twelve? Yes, twelve.
A nice, familiar, comfortable dozen (well, unless you like that easy metric-system stuff, where 10 is a better number.)
Let’s count:
When we introduce this many products in a year, it naturally lights people up with a lot of questions. Why did we go bonkers and introduce so many new products in 2014? Will we do this again? When are the third-gen products coming out?
I’ll answer all of these—and get into the whole whys and wherefores of the Magni/Modi 2/Ubers a bit later. First, let’s start with what we learned.
Hard Business Lessons
If you’re thinking of starting your own business, or running one right now, this is probably the most important part—something that can be boiled down to heeding our own advice about “keeping it simple, stupid.”
Because we end 2014 as our most successful year ever—and, at the same time, asking ourselves pointed questions about the future.
Here’s what we learned.
12 intros in a year is too many with our engineering resources. And probably even with additional resources. We have nobody to blame for this except ourselves. Early in the year, I put up a whiteboard in my office and outlined all of our possible/planned products and numbered them. I came up with 16. Three died in the Valkyrie experimentation, and one missed (Yggdrasil.) Even at that point in time, I knew that an average intro of over one product per month was logistically painful—or perhaps even impossible. We are lucky that only one product had significant issues, and that only one product missed its intro date. So how many products makes sense? I’ll get to that.
12 intros is also probably too much for good marketing impact. If you’re hitting the press with a dozen new products—especially when they are concurrent—it makes it hard to digest. The intro of the second-gen Magni and Modi coming right after Fulla was not ideal, but our hands were forced by production delays. The initial plan was Fulla in August/September and Magni/Modi 2 in mid November. Yeah. There you go. Fewer products, more widely spaced, allows the press to absorb, digest, and comment on the new stuff, before you hit them again. What’s the ideal timing? I’ll get to that, too.
Each product has many more complexities than you expect. Yeah, it’s one thing to have a working PC board, and it’s another thing to be shipping. Even getting to a fully working board can be challenging. But, to get to shipping, you need to:
And this is in addition to all the marketing stuff:
This giant-ass list above applies to updated products as well. Yes, it does. Don’t tell yourself it doesn’t. Updates, real updates, give you a shortcut in only one place: chassis. You may be able to use something very similar, and cut down on the rev cycle. But that’s about it. A revised product is a new product, unless you are so stunningly cynical you’re gonna reprogram it a bit and slap a “2” on it.
The line may now be too large. When you look at it, we have a hell of a lot of options. Which is really cool, because not everyone likes the same thing. But we are getting into “confusion by profusion,” so this is something we’ll have to look at. Don’t panic—we’re not thinking of killing off anything, but we will be trying to make it more clear how the different options stack up…and, of course, we’re going to weigh any additions to the product line very carefully in the future. Again, don’t panic—I think we have a pretty good grasp of where the shortcomings are…and this does not preclude moving into different markets (more “traditional” preamps and power amps…though our evolving ideas are very much non-traditional…)
Harsh? No, realistic. Again, want to start a business? You’re gonna have to be realistic. You need to step back, look at what you’ve done, and evaluate it as Spock-like and dispassionately as possible.
Am I down on Schiit? No, not at all. It’s the most exciting thing I’ve done, ever…and the stuff we do have coming is, well, pretty darn wowie. Looking at it dispassionately, even.
Would I have done things differently, introduced less products this year? Maybe. Not sure. Get back to me on that one later.
The Ideal Scenario
“Okay, so what would be an ideal number of products for the year?” you ask.
Cool. Let me preface this by the fact that I’m not a trained, professional product manager, and I never spent any time at a big company that introduces dozens, or hundreds of products a year. So this applies to niche businesses…maybe only niche audio businesses…or maybe only Schiit.
I think it’s about 4 products per year.
Why?
Engineering-wise, it’s a comfortable rate. At least for us. Our engineering cycles usually start with “what ifs” more than a year before the product is a reality. But once we’ve decided we have a product to do, 3 months between releases allows engineering to concentrate fully on the product, rather than pinballing between several different ones.
Production-wise, it fits very well with typical cycles. Metal has a cycle of 6-8 weeks, and finding all the other parts can take another few weeks, and dealing with out-of-stock parts, first articles, and production ramp-up is another few weeks. Releasing every three months allows you to work through the issues on one product at a time, which is much more sanity-inducing than several.
Marketing-wise, it’s ideal. Think about it. You introduce a new product, and everyone’s talking about it for a month. Another month, and you get the inevitable “it ain’t that great” people coming out of the woodwork. Another month, and it’s settled down into the line well, and the press who were excited about it have moved on to different products…so when you introduce a new product, they’re ready to hear about it again.
“So you’re not going to introduce more than 4 products per year from now on?” you ask. “If you go over that, the world will implode?”
No. Not at all. I’m just saying that from a technical and marketing standpoint, it seems like 4 per year is ideal. The number might even be lower. Or it could be higher.
In short: 2015, we’ll see. We have 4 “definitelys” and a couple of “maybes” on the board now…which is a heck of a lot better than 16.
And…if you think we could precisely space 4 products at 3 month intervals, you have a lot more faith in us than we do. Let’s hope all four don’t ship December 20. 2014.*
*Don’t worry, they won’t. And Yggy will be first.
So, About These Second-Gen Magni and Modis
Why did we feel we had to update the Magni and Modi, after only two years in production?
Well, 2 years is actually a pretty long run on the low-cost side of things. And the market changes over time. And, we’d had plenty of feedback about how a gain switch would be a good addition to the Magni. And, we’d also had plenty of feedback about how a Modi with multiple inputs would be a good thing.
So, all in all, it was time to take a look at them, beginning in early 2014.
Yep, the dev time was that long. Not because it was problematic, but because we intentionally held the products back in order to introduce them later in the year. A little later than we would have liked, yeah, but hey, sometimes things don’t go precisely to plan.
It started with me playing with Magni. Or, more specifically, Magni. The gain switch was a must, of course. But if I was doing that, I figured I might as well take a look at the whole thing and see if it could be improved upon.
And, like many engineering challenges, the question was, “Where do you stop?”
Staying within the original Magni budget meant fairly limited changes, but it still did allow for some pretty significant work. The gain stage had originally been designed as a minimal-parts design. With the addition of a handful of parts, though, I could turn it into a constant-feedback design that promised to sound significantly better than the original.
Aside: so what’s all this constant-feedback blather? It’s constant feedback across the audio band. Now, Cordell disproved the old Otala TIM hypothesis, but in my work, I’ve always noted sonic benefits to having an open-loop stage with bandwidth larger than the audio range. So, I extended the open-loop bandwidth to greater than 20kHz with some parts additions and other tweaks, changed the operating point of the front end (it now runs quite a bit more current, and has smaller resistor values for even lower noise), and, of course, put in the gain switch.
Aside to the aside: yes, this is heavy geek-speak. If you know Cordell and Otala and concepts like thermal noise, this makes sense. Though you might not believe what I’m saying about the subjective side of things. That’s cool.
Aside to the aside to the aside: bottom line, Magni 2 sounds better. And it’s not subtle.
How much better? The Magni 2 was originally called the Magni Uber.
But, while I was doing this cost-constrained version, I wondered, “What would we get if we threw some more money at this?” So, I built another super-over-the-top Magni Uber Squared, which had a whole host of improvements, starting with preamp outputs, better, adjustable voltage regulators and higher rails, a complementary-input VAS to cancel even more noise and distortion from the front end (and yeah, I know Self sez you don’t need no complementary VAS, that’s cool, maybe we hear differently), much bigger power supply capacitors and a giant new wall-wart, and a couple of other little tweaks. This super-over-the-top version I figured we might never build, but I had to know how it performed.
How’d it do? So well that we decided to build both of them. One of our listeners said, “Just stop building everything else and do these,” when hearing the Magni Uber Squared prototype for the first time.
Aside: don’t worry, we’re not going to do that. He’s insanely cheap, and that explains a lot of his comment. It is really good, though.
So, with that decision, we knew that we’d have two new amps. There was some production budget left over, so I drew up a small solid-aluminum knob and added an aluminum top to the Uber to make it look a bit fancier.
Beyond that, prototyping to production was pretty uneventful. Magni 2 and Magni 2 Uber are two of those “boring” products that you really like to have.
Modi? Modi took some more turns.
Why? Partly because the original Modi was a very nice DAC, for a $99 DAC. Mike pretty much thought of every trick he could apply to it. So, that made it difficult to make better.
One obvious thing we could do was to open up higher sampling rates—extend it up to 24/192. Of course, this would require drivers for Windows.
Aside: really, Microsoft? No USB Class 2 support in Windows 10? Reeeaalllly? Android phones do this now, you know. Do you realize how stupid you look to the audio community? Yeah, I know, you don’t care. But you still suck.
To get around the “hey I just got my first DAC and I can’t install the drivers” problem, we added a switch. In Standard mode, Modi 2 is locked down to USB Audio Class 1, and 24/96 max. But you can pretty much plug it into a toaster* and it will work.
*I am kidding. Unless the toaster is also a computer. Which is possible around here.
Put Modi 2 in Expert mode, and you can run 24/192, but then you have to install drivers. But at least there’s one mode that works, and we’re setting expectations that, well, Expert mode is for Experts.
Plus, we tweaked the parts and layout a bit, and there you have it: Modi 2.
“So why U not change the DA chip and do 32/768 and DSD 8X and stuffs?” some people are wondering. Well, it’s simple:
Fun fact: I did a custom version of the Modi firmware that enabled up to 24/384 for laughs. I really wanted to introduce this as a 24/384 DAC, but then be the only manufacturer to say, “And we did it only for the numbers, unlike everyone else, we’ll tell you that this really doesn’t matter.”
But Mike talked me out of it. So, you could have had a 24/384 Modi 2. Which would mean nothing. But it would have a big number on it. Maybe screen it on the front and put racing stripes on it, like they used to put the “5 Speed” badges on the back of Hondas in the 1980s.
Aside: don’t bother asking me for the firmware. Seriously.
In the case of Modi 2 Uber, though, things were very different. Mike and I had talked about doing a Modi with USB and optical input for a while. I’d even drawn up some chassis for it. (This will come in later.) But when we got down to actually talking about a step-up from Modi, Mike opined:
“We should just do the whole thing,” he said. “Optical, coax, USB.”
“Just like a Bifrost,” I said.
“Yep!” Mike grinned.
“Like a mini Bifrost,” I repeated.
“Right!” Mike said, still not getting it.
“Don’t you think that might cut into Bifrost sales?”
Mike laughed. “Who cares? Plus, if they don’t get the whole upgradable thing, or don’t want it, then you have a heck of a stack for $300.”
Hey, I’m not one to argue. So that’s why Modi 2 Uber is pretty much a mini Bifrost. Mike also lobbied for the Bifrost pushbutton on the front, too, so you can thank him for that.
But Modi 2 Uber is a very different product than a Modi 2. Because it has optical and coaxial inputs, it needs a power supply. It can’t rely on the USB input alone for power.
Which gives it a big leg up on Modi 2 right there—a linear, low-noise, multiply regulated power supply. With a wall-wart, yeah, but a linear supply will beat USB power anyday.
It’s also different in another way: we have to switch, and manage, multiple inputs. Like all of our other DACs, Modi 2 Uber doesn’t use asynchronous sample rate conversion to make all the inputs a convenient single bit depth and sampling rate—which means we have to detect the bit depth and sampling rate, and change clock multiples on the fly for multiple inputs.
This means that Modi 2 Uber has a microprocessor. Which has to have its own firmware. So, when we’re producing Modi 2 Uber, we install 2 firmwares, one for the USB input, and one for the microprocessor clock management.
So, yeah—just like a mini Bifrost.
Metal Hilarity
Well, it’s funny in retrospect. While it was happening, not so much. Modi 2 Uber was a bit of a pain on the metalwork side—more so than ever before.
And no, not because of fit and finish. Because of revisions.
Remember, I said that I’d done drawings of a 2-input Modi 2 Uber? Yeah. Those got sent out to quote. Then, when I changed it over to 3 inputs with an aluminum top and button, I sent those out to quote, too.
Then, thinking we might be able to show the products at Can-Jam, I asked for a finished first article of Modi 2 Uber, complete with the rev level of the 3-input product: Rev D.
Then, when the metal guys asked for the silkscreen, the problems started.
“Hey, this screen doesn’t match the metal,” they told me.
Weird, I thought, and checked the dimensions. But it was fine, it should fit. So that’s what I told them.
“It doesn’t fit, like, at all,” they told me.
“What?”
“It’s like it’s for a different product.”
Ah, crap. What had I done? But I checked the PO and the drawing, and everything matched.
A few emails later, and the mystery was solved: they’d built the older “Rev B” drawing by mistake. At least they didn’t build a few thousand of them. But it still set us back.
Then, when we got the PC boards in, again we had problems—they didn’t fit. And again, the metal guys were the culprit. They’d built the Rev D drawings, rather than the Rev E I gave them after we got the (corrected) first articles.
With a mistake of this magnitude, you have two choices:
6-8 weeks, as of mid-November, was a no-go…and we were rapidly running out of Magnis. New boards, on the other hand, could be had in 5 days.
So that’s what we did—went with the older metal rev, and changed the board. Now you know why the Modi 2 Ubers are a little late.
What Will We See Next Year?
Let’s get this off the table first: Gen 3 products? Nope, you won’t see any in 2015.
As I mentioned before, upgrading Asgard 2 to Asgard 3 would be a monumentally different thing than going from 1-2. We’ve addressed pretty much all of the first generation’s limitations within the product budget. So, don’t plan for an A3 next year.
So what will we expect to see? Less products, but more significant introductions. In fact, I think I can safely say that all of our planned intros will be “one and only” products that will really surprise you.
It’ll start with Yggdrasil, of course. Q1 is the target. With any luck, it’ll be mid-Q1. But we’ll see. I don’t think we need a beta test in this case, either, so when they’re available, they’re available.
And that leaves three mysteries.
Mystery 1. Expect to wrap your mind around a couple of new concepts with this one. It’s really hard to say anything more about this…but consider that one component of it will be compatible with some of our other products.
Mystery 2. I’ve said I’ve been working on a new balanced topology, and that will be incorporated into one of the planned mystery products. It does not replace anything we’re currently selling, but it does take us in some surprising new directions. And that’s all I’ll say about that.
Mystery 3. And expect something that nobody is expecting from us. I can’t say more.
And, of course, there are a couple of wild cards, too…we’ll see what happens with those as we get further along in development.
And with that, we are fully up to date, and this book is complete…but I hope you put up with me posting new stories from time to time…about every other week.
One Year, Twelve Products
Okay. Let me get this off my chest now: twelve products in one year is too many. Yes, even when half of them were updates, and two of them variations on the updates.
Twelve? Yes, twelve.
A nice, familiar, comfortable dozen (well, unless you like that easy metric-system stuff, where 10 is a better number.)
Let’s count:
- Wyrd
- Optical Modi
- Sys
- Valhalla 2
- Lyr 2
- Mani
- Ragnarok
- Fulla
- Magni 2
- Magni 2 Uber
- Modi 2
- Modi 2 Uber
When we introduce this many products in a year, it naturally lights people up with a lot of questions. Why did we go bonkers and introduce so many new products in 2014? Will we do this again? When are the third-gen products coming out?
I’ll answer all of these—and get into the whole whys and wherefores of the Magni/Modi 2/Ubers a bit later. First, let’s start with what we learned.
Hard Business Lessons
If you’re thinking of starting your own business, or running one right now, this is probably the most important part—something that can be boiled down to heeding our own advice about “keeping it simple, stupid.”
Because we end 2014 as our most successful year ever—and, at the same time, asking ourselves pointed questions about the future.
Here’s what we learned.
12 intros in a year is too many with our engineering resources. And probably even with additional resources. We have nobody to blame for this except ourselves. Early in the year, I put up a whiteboard in my office and outlined all of our possible/planned products and numbered them. I came up with 16. Three died in the Valkyrie experimentation, and one missed (Yggdrasil.) Even at that point in time, I knew that an average intro of over one product per month was logistically painful—or perhaps even impossible. We are lucky that only one product had significant issues, and that only one product missed its intro date. So how many products makes sense? I’ll get to that.
12 intros is also probably too much for good marketing impact. If you’re hitting the press with a dozen new products—especially when they are concurrent—it makes it hard to digest. The intro of the second-gen Magni and Modi coming right after Fulla was not ideal, but our hands were forced by production delays. The initial plan was Fulla in August/September and Magni/Modi 2 in mid November. Yeah. There you go. Fewer products, more widely spaced, allows the press to absorb, digest, and comment on the new stuff, before you hit them again. What’s the ideal timing? I’ll get to that, too.
Each product has many more complexities than you expect. Yeah, it’s one thing to have a working PC board, and it’s another thing to be shipping. Even getting to a fully working board can be challenging. But, to get to shipping, you need to:
- Have that fully working PC board—in our case, we will go through 2-5 revisions of the initial proto to make it right
- Have the firmware fully functional, tested, and busted—and, in some cases, this means multiple firmwares, as with Modi 2 Uber, which has both a microprocessor and a USB input that need to be programmed. And by “busted,” we mean, “hand it to some people and have them do everything crazy they can think of, including pushing all the buttons simultaneously, pushing the buttons on power-up or down, etc.
- Have a working, fitting, cosmetically perfect chassis in large quantities, ready to go—again, we usually go through a couple of revisions on this, though some of our chassis are at “Revision J” by the time they’re released to production…but then you’re waiting for production to arrive, and crossing your fingers that it’s perfect.
- Have any knobs/buttons in large quantities—see above, though it’s usually just one rev and we’re done
- Have all the fasteners for that chassis in large quantities—surprisingly nontrivial, especially in the case of complex products like Ragnarok and custom stuff like we use with Fulla…Ragnarok uses literally two dozen different screws/nuts/standoffs/washers/spacers/etc.
- Have any custom insulators/etc for that chassis in large quantities—see the two products above, which use either custom Sorbothane parts, or custom electrical insulation parts
- Have all the parts for the PC boards (and we can be talking hundreds to thousands of parts, each and every one of which might make finishing the product impossible—talk to Alex about this…)
- Have complete documentation necessary for the PC board assembly house, including any special test procedures—or what you get back from them might not be, well, what you expect
- Have your team ready for any new assembly challenges—especially with complex products like Ragnarok
- Have a shipping box that will protect the product tested, busted, and ready to go—sounds simple, but a cheap, sturdy box is not as trivial as you think…put a prototype in it and ship it across the country a few times to check
- Have a manual for the product fully written and ready to go in the box—no manual, no time? Then you’d better at least have online instructions. But really, a manual. Seriously.
- Have the technicians and final sound-check guys ready to test the product—if they don’t know what to expect, again, brace yourself for the unexpected
- Have at least one staff member ready to answer questions about the product, based on actually having used it—again, sounds simple, but this really isn’t trivial
And this is in addition to all the marketing stuff:
- Product description—what is it, what does it do, why is it special, how is it different? Yes, I know lots of people don’t read, but you do need this.
- Product photography—and I’m not talking photos with your iPhone. Even if it is an iPhone 6. Get some professional work done.
- Specifications. Testing, retesting, confirmation, and more testing, decisions on claims (even if your products measure 0.0004% THD, are you comfortable claiming this, in all systems, on all analyzers?)
- Press release—an email release in PR-speak that tells the press you are serious, and covers all the basics about the product, including price and availability
- Community announcements—like on head-fi.org
- Facebook, if you’re crazy enough to do that
- Product FAQ—it really does help. You should do one. And sit back the first day and watch the online feedback so you can add to it.
- Excerpts and links to any early reviews—if you have them
- Anything else—stunts, videos, laser-etching your logo on the moon, skydivers, rockets, soda dispensers in bathrooms, whatever, yeah, marketing is stupid, but cover the first stuff before “anything else”
This giant-ass list above applies to updated products as well. Yes, it does. Don’t tell yourself it doesn’t. Updates, real updates, give you a shortcut in only one place: chassis. You may be able to use something very similar, and cut down on the rev cycle. But that’s about it. A revised product is a new product, unless you are so stunningly cynical you’re gonna reprogram it a bit and slap a “2” on it.
The line may now be too large. When you look at it, we have a hell of a lot of options. Which is really cool, because not everyone likes the same thing. But we are getting into “confusion by profusion,” so this is something we’ll have to look at. Don’t panic—we’re not thinking of killing off anything, but we will be trying to make it more clear how the different options stack up…and, of course, we’re going to weigh any additions to the product line very carefully in the future. Again, don’t panic—I think we have a pretty good grasp of where the shortcomings are…and this does not preclude moving into different markets (more “traditional” preamps and power amps…though our evolving ideas are very much non-traditional…)
Harsh? No, realistic. Again, want to start a business? You’re gonna have to be realistic. You need to step back, look at what you’ve done, and evaluate it as Spock-like and dispassionately as possible.
Am I down on Schiit? No, not at all. It’s the most exciting thing I’ve done, ever…and the stuff we do have coming is, well, pretty darn wowie. Looking at it dispassionately, even.
Would I have done things differently, introduced less products this year? Maybe. Not sure. Get back to me on that one later.
The Ideal Scenario
“Okay, so what would be an ideal number of products for the year?” you ask.
Cool. Let me preface this by the fact that I’m not a trained, professional product manager, and I never spent any time at a big company that introduces dozens, or hundreds of products a year. So this applies to niche businesses…maybe only niche audio businesses…or maybe only Schiit.
I think it’s about 4 products per year.
Why?
Engineering-wise, it’s a comfortable rate. At least for us. Our engineering cycles usually start with “what ifs” more than a year before the product is a reality. But once we’ve decided we have a product to do, 3 months between releases allows engineering to concentrate fully on the product, rather than pinballing between several different ones.
Production-wise, it fits very well with typical cycles. Metal has a cycle of 6-8 weeks, and finding all the other parts can take another few weeks, and dealing with out-of-stock parts, first articles, and production ramp-up is another few weeks. Releasing every three months allows you to work through the issues on one product at a time, which is much more sanity-inducing than several.
Marketing-wise, it’s ideal. Think about it. You introduce a new product, and everyone’s talking about it for a month. Another month, and you get the inevitable “it ain’t that great” people coming out of the woodwork. Another month, and it’s settled down into the line well, and the press who were excited about it have moved on to different products…so when you introduce a new product, they’re ready to hear about it again.
“So you’re not going to introduce more than 4 products per year from now on?” you ask. “If you go over that, the world will implode?”
No. Not at all. I’m just saying that from a technical and marketing standpoint, it seems like 4 per year is ideal. The number might even be lower. Or it could be higher.
In short: 2015, we’ll see. We have 4 “definitelys” and a couple of “maybes” on the board now…which is a heck of a lot better than 16.
And…if you think we could precisely space 4 products at 3 month intervals, you have a lot more faith in us than we do. Let’s hope all four don’t ship December 20. 2014.*
*Don’t worry, they won’t. And Yggy will be first.
So, About These Second-Gen Magni and Modis
Why did we feel we had to update the Magni and Modi, after only two years in production?
Well, 2 years is actually a pretty long run on the low-cost side of things. And the market changes over time. And, we’d had plenty of feedback about how a gain switch would be a good addition to the Magni. And, we’d also had plenty of feedback about how a Modi with multiple inputs would be a good thing.
So, all in all, it was time to take a look at them, beginning in early 2014.
Yep, the dev time was that long. Not because it was problematic, but because we intentionally held the products back in order to introduce them later in the year. A little later than we would have liked, yeah, but hey, sometimes things don’t go precisely to plan.
It started with me playing with Magni. Or, more specifically, Magni. The gain switch was a must, of course. But if I was doing that, I figured I might as well take a look at the whole thing and see if it could be improved upon.
And, like many engineering challenges, the question was, “Where do you stop?”
Staying within the original Magni budget meant fairly limited changes, but it still did allow for some pretty significant work. The gain stage had originally been designed as a minimal-parts design. With the addition of a handful of parts, though, I could turn it into a constant-feedback design that promised to sound significantly better than the original.
Aside: so what’s all this constant-feedback blather? It’s constant feedback across the audio band. Now, Cordell disproved the old Otala TIM hypothesis, but in my work, I’ve always noted sonic benefits to having an open-loop stage with bandwidth larger than the audio range. So, I extended the open-loop bandwidth to greater than 20kHz with some parts additions and other tweaks, changed the operating point of the front end (it now runs quite a bit more current, and has smaller resistor values for even lower noise), and, of course, put in the gain switch.
Aside to the aside: yes, this is heavy geek-speak. If you know Cordell and Otala and concepts like thermal noise, this makes sense. Though you might not believe what I’m saying about the subjective side of things. That’s cool.
Aside to the aside to the aside: bottom line, Magni 2 sounds better. And it’s not subtle.
How much better? The Magni 2 was originally called the Magni Uber.
But, while I was doing this cost-constrained version, I wondered, “What would we get if we threw some more money at this?” So, I built another super-over-the-top Magni Uber Squared, which had a whole host of improvements, starting with preamp outputs, better, adjustable voltage regulators and higher rails, a complementary-input VAS to cancel even more noise and distortion from the front end (and yeah, I know Self sez you don’t need no complementary VAS, that’s cool, maybe we hear differently), much bigger power supply capacitors and a giant new wall-wart, and a couple of other little tweaks. This super-over-the-top version I figured we might never build, but I had to know how it performed.
How’d it do? So well that we decided to build both of them. One of our listeners said, “Just stop building everything else and do these,” when hearing the Magni Uber Squared prototype for the first time.
Aside: don’t worry, we’re not going to do that. He’s insanely cheap, and that explains a lot of his comment. It is really good, though.
So, with that decision, we knew that we’d have two new amps. There was some production budget left over, so I drew up a small solid-aluminum knob and added an aluminum top to the Uber to make it look a bit fancier.
Beyond that, prototyping to production was pretty uneventful. Magni 2 and Magni 2 Uber are two of those “boring” products that you really like to have.
Modi? Modi took some more turns.
Why? Partly because the original Modi was a very nice DAC, for a $99 DAC. Mike pretty much thought of every trick he could apply to it. So, that made it difficult to make better.
One obvious thing we could do was to open up higher sampling rates—extend it up to 24/192. Of course, this would require drivers for Windows.
Aside: really, Microsoft? No USB Class 2 support in Windows 10? Reeeaalllly? Android phones do this now, you know. Do you realize how stupid you look to the audio community? Yeah, I know, you don’t care. But you still suck.
To get around the “hey I just got my first DAC and I can’t install the drivers” problem, we added a switch. In Standard mode, Modi 2 is locked down to USB Audio Class 1, and 24/96 max. But you can pretty much plug it into a toaster* and it will work.
*I am kidding. Unless the toaster is also a computer. Which is possible around here.
Put Modi 2 in Expert mode, and you can run 24/192, but then you have to install drivers. But at least there’s one mode that works, and we’re setting expectations that, well, Expert mode is for Experts.
Plus, we tweaked the parts and layout a bit, and there you have it: Modi 2.
“So why U not change the DA chip and do 32/768 and DSD 8X and stuffs?” some people are wondering. Well, it’s simple:
- The AKM4396 is still one of the highest performing DACs, ever—check its specs against some of the flavors du jour
- It is also one of the best sounding DACs, ever—and yeah, we know all about the new ones
- It’s not even 32 bits internally, so we can’t claim 32 bits
- 24/352 and 24/384 content is comically scarce, and let me know when you find some DSD 8X content
Fun fact: I did a custom version of the Modi firmware that enabled up to 24/384 for laughs. I really wanted to introduce this as a 24/384 DAC, but then be the only manufacturer to say, “And we did it only for the numbers, unlike everyone else, we’ll tell you that this really doesn’t matter.”
But Mike talked me out of it. So, you could have had a 24/384 Modi 2. Which would mean nothing. But it would have a big number on it. Maybe screen it on the front and put racing stripes on it, like they used to put the “5 Speed” badges on the back of Hondas in the 1980s.
Aside: don’t bother asking me for the firmware. Seriously.
In the case of Modi 2 Uber, though, things were very different. Mike and I had talked about doing a Modi with USB and optical input for a while. I’d even drawn up some chassis for it. (This will come in later.) But when we got down to actually talking about a step-up from Modi, Mike opined:
“We should just do the whole thing,” he said. “Optical, coax, USB.”
“Just like a Bifrost,” I said.
“Yep!” Mike grinned.
“Like a mini Bifrost,” I repeated.
“Right!” Mike said, still not getting it.
“Don’t you think that might cut into Bifrost sales?”
Mike laughed. “Who cares? Plus, if they don’t get the whole upgradable thing, or don’t want it, then you have a heck of a stack for $300.”
Hey, I’m not one to argue. So that’s why Modi 2 Uber is pretty much a mini Bifrost. Mike also lobbied for the Bifrost pushbutton on the front, too, so you can thank him for that.
But Modi 2 Uber is a very different product than a Modi 2. Because it has optical and coaxial inputs, it needs a power supply. It can’t rely on the USB input alone for power.
Which gives it a big leg up on Modi 2 right there—a linear, low-noise, multiply regulated power supply. With a wall-wart, yeah, but a linear supply will beat USB power anyday.
It’s also different in another way: we have to switch, and manage, multiple inputs. Like all of our other DACs, Modi 2 Uber doesn’t use asynchronous sample rate conversion to make all the inputs a convenient single bit depth and sampling rate—which means we have to detect the bit depth and sampling rate, and change clock multiples on the fly for multiple inputs.
This means that Modi 2 Uber has a microprocessor. Which has to have its own firmware. So, when we’re producing Modi 2 Uber, we install 2 firmwares, one for the USB input, and one for the microprocessor clock management.
So, yeah—just like a mini Bifrost.
Metal Hilarity
Well, it’s funny in retrospect. While it was happening, not so much. Modi 2 Uber was a bit of a pain on the metalwork side—more so than ever before.
And no, not because of fit and finish. Because of revisions.
Remember, I said that I’d done drawings of a 2-input Modi 2 Uber? Yeah. Those got sent out to quote. Then, when I changed it over to 3 inputs with an aluminum top and button, I sent those out to quote, too.
Then, thinking we might be able to show the products at Can-Jam, I asked for a finished first article of Modi 2 Uber, complete with the rev level of the 3-input product: Rev D.
Then, when the metal guys asked for the silkscreen, the problems started.
“Hey, this screen doesn’t match the metal,” they told me.
Weird, I thought, and checked the dimensions. But it was fine, it should fit. So that’s what I told them.
“It doesn’t fit, like, at all,” they told me.
“What?”
“It’s like it’s for a different product.”
Ah, crap. What had I done? But I checked the PO and the drawing, and everything matched.
A few emails later, and the mystery was solved: they’d built the older “Rev B” drawing by mistake. At least they didn’t build a few thousand of them. But it still set us back.
Then, when we got the PC boards in, again we had problems—they didn’t fit. And again, the metal guys were the culprit. They’d built the Rev D drawings, rather than the Rev E I gave them after we got the (corrected) first articles.
With a mistake of this magnitude, you have two choices:
- Stand on ceremony, return the metal, and wait 6-8 weeks.
- Change the board to match the older revision and throw away the boards we had. This was fairly simple, and probably something we should have done in the first place.
6-8 weeks, as of mid-November, was a no-go…and we were rapidly running out of Magnis. New boards, on the other hand, could be had in 5 days.
So that’s what we did—went with the older metal rev, and changed the board. Now you know why the Modi 2 Ubers are a little late.
What Will We See Next Year?
Let’s get this off the table first: Gen 3 products? Nope, you won’t see any in 2015.
As I mentioned before, upgrading Asgard 2 to Asgard 3 would be a monumentally different thing than going from 1-2. We’ve addressed pretty much all of the first generation’s limitations within the product budget. So, don’t plan for an A3 next year.
So what will we expect to see? Less products, but more significant introductions. In fact, I think I can safely say that all of our planned intros will be “one and only” products that will really surprise you.
It’ll start with Yggdrasil, of course. Q1 is the target. With any luck, it’ll be mid-Q1. But we’ll see. I don’t think we need a beta test in this case, either, so when they’re available, they’re available.
And that leaves three mysteries.
Mystery 1. Expect to wrap your mind around a couple of new concepts with this one. It’s really hard to say anything more about this…but consider that one component of it will be compatible with some of our other products.
Mystery 2. I’ve said I’ve been working on a new balanced topology, and that will be incorporated into one of the planned mystery products. It does not replace anything we’re currently selling, but it does take us in some surprising new directions. And that’s all I’ll say about that.
Mystery 3. And expect something that nobody is expecting from us. I can’t say more.
And, of course, there are a couple of wild cards, too…we’ll see what happens with those as we get further along in development.
And with that, we are fully up to date, and this book is complete…but I hope you put up with me posting new stories from time to time…about every other week.
kongmw
100+ Head-Fier
- Joined
- Jun 2, 2008
- Posts
- 146
- Likes
- 11
Christmas comes (less than a day) early!
Thanks for the chapter Jason. It's always been fun reading about the behind the scene stories.
Thanks for the chapter Jason. It's always been fun reading about the behind the scene stories.
rmullins08
100+ Head-Fier
- Joined
- May 12, 2014
- Posts
- 469
- Likes
- 98
Another great chapter. Perfect way to kill time on a slow day at the office until quitting time.
I got half of the new products this year (well, will have half when the Uber combo arrives Friday).
Looking forward to what else in store next year.
I got half of the new products this year (well, will have half when the Uber combo arrives Friday).
Looking forward to what else in store next year.
- Joined
- Nov 6, 2008
- Posts
- 1,195
- Likes
- 371
No Gen 3 products but what about more Gen 2 products? Maybe the wild cards?
Whitetriton
100+ Head-Fier
Thank you very much Jason. I've read and enjoyed every chapter. I definitely hope you continue to add posts in the future.
Happy Holidays to the entire Schiit team.
Happy Holidays to the entire Schiit team.
