Chapter 4:
“You Always Say You Have Schiit to Do, Why Don’t You Just Call It That?”
It’s a long road from thinking to doing, though, especially when you’ve been out of the game for so long.
I was rusty, incredibly rusty. I’d forgotten a lot of what I knew about engineering, simply because I hadn’t used it in so long. And a lot of stuff had changed.
Change. As a single example, let’s consider a conversation I had at Semicon, the semiconductor manufacturer’s trade show, about 2001 or so. I was talking to an engineer about a new product, and mentioned that I’d been an engineer and had worked with VLSI gate arrays back at Magnavox—cutting-edge prototypes that cost $100,000 each.
“How many gates?” the other engineer asked me.
“About 100k,” I told him.
He laughed. “Not the price, the number of gates.”
“Right, a hundred thousand.”
The other engineer laughed even harder and waved a hand, as if dismissing a servant. “100k? We put stuff like that in toasters today.”
Yep. 12 years took $100k prototypes to $1 commodities. And now, 12 years later, we’re all carrying smartphones. In another 12 years, we’ll be wondering how we got along without Google Ambient and pervasive intelligent packaging.
But back to Schiit. There were a billion questions before we got started:
• Was everything SMD (surface mount electronics, very hard to work with)?
• Where had component costs gone?
• What kind of components were people using for audio these days—was discrete design even feasible any more?
• Where should you get them? In the old days, the reps would come visit us and give us data books and quotes (yes, as in ink printed on paper), but how did it work now?
• Same for PC boards—what were the costs now?
• What PC board layout software did people use these days?
• How about schematic capture?
• How about CAD for drawing up chassis—in the old days, we just used Illustrator?
• What did it cost to make a chassis?
• Where would you get it done?
• Who’d do the transformers?
• What were transformer costs like?
• How would we assemble the products? Sumo had done in-house assembly of pc boards and products, Theta had gone to board houses (that is, people who put parts in PC boards) and only done final assembly, and lots of people were going to complete contract assembly (typically in China) and only marketing and supporting the products.
• What safety approvals did we need?
• What did a modern R&D setup look like? We were familiar with Audio Precision analyzers, but we’d never used one, and didn’t know what they made and what they cost. Mike still had his ancient Stanford spectrum analyzer, which is what we used to use in “the old days”, but was that good enough? I had an old HP 339 analyzer—same questions applied there.
• Where would we do it all?
But you can obsess over all those questions and turn them into a giant wall that keeps you from ever doing anything (see Chapter 2 on big-company paralysis by analysis), or you can do some digging.
I dug. And quickly found out that a lot of the information I needed was online. Component availability and cost, PCB cost, transformer cost—that was all there. Most surprisingly, all of them were much cheaper than we’d paid twenty years ago. Thank globalization, or the downturn, or whatever, but even parts that were made in the USA were far, far less expensive than we expected. And, despite the bleating of the apocalyptic crowd about how "there ain't no more good audio parts out there," there were actually plenty—and even more new options when you got on the surface-mount side.
But then there was software. Mike used (and still uses) Altium for doing schematics and PCBs. Altium is 100% old-skool big-ticket software. Hint: if they don’t list the price, you don’t want to ask. Altium starts at about $7200.
Not a good price, especially since I was still in the “foolin around” phase. The company had no name. Centric was deep in the tank. I didn’t want to pay for anything I didn’t really need. In fact, as I thought about it, I decided on a goal: Can we start this for $10K, including the costs of the first run of products?
$10K we could gamble with. If it went nowhere, well, hell, as Mike said, Christmas presents. We’d lost more money at Centric on bankrupt companies and deadbeat buyers. $10K was doable. It wouldn’t thrill me to flush it down the toilet, but it wouldn’t kill me, either.
But $7200 suddenly meant that I would be looking at something more like $20K. I wasn’t ready to commit to that. So I looked a bit more, and discovered that software, like everything else, had changed. In addition to Altium, there were other, lower-cost alternatives like Eagle. And there was a funky little open-source program called KiCad, built by some crazy student as a senior project. It was free.
Yes, free.
Interlude: Let’s pause for a moment to salute open-source software. Who would have ever expected that Microsoft would have been routed by open source? Who would have predicted that open-source is what powers most of the internet? Who would have known that there would be opportunity created all over the internet by software like Linux, Wordpress, Drupal, Joomla, and a hundred others?
Just consider Google. Google gives away things like Android, Gmail, Google Apps for Business, Google Drive, and plays with crazy stuff like self-driving cars and longevity enhancement. But Google is really an ad company. Yep. That’s how they make their money. They sell ads. Their ads show up everywhere—unobtrusive text ads in search and on websites, or banners served up on millions of sites across the desktop and mobile world. And those ads drive the free software that billions of people use. Pretty amazing business model.
But, back to the free software? Sure, why not give it a try.
Most free stuff isn’t very well worked out, and crashes a lot, but Kicad was pretty full-featured and stable. It was missing some very standard features (like cut and paste! And undo!) Still, I’d worked with inconvenient software before, some of it very expensive (I’m looking at you, Photoshop 2.5.) I decided that Kicad was good enough to use for the first layouts.
And the gamble paid off. Not only did Kicad work well enough for those first layouts, I still use it today. Now, Kicad is much more full-featured, and has a solid roadmap of updates and an active developer community. It even has undo (wow.)
CAD? Yeah, we looked at a number of open-source options, and eventually ended with Alibre (now Geomagic.) How’d we arrive at that decision?
1. I remembered their name from the days when we were doing marketing work for MSC Software.
2. They had a $99 special.
Yeah. We’re cheap. What can I say?
But the CAD story doesn’t end there. In fact, only today (2013) are we really getting serious about 3D CAD. The learning curve for parametric modeling was just too steep. To start, we went back to the old days—2D drafting in Illustrator.
The 01 Cometh
Okay, so how did we end up with Schiit’s simple, minimalistically-elegant chassis? It’s almost entirely a story of economics. To make something that would compete with Chinese prices, we needed a cheap box. Period.
That immediately threw out a couple of things:
• Lots of little pieces and fancy cosmetics, like Chinese tube amps
• Anything machined out of a solid piece of aluminum—you don’t want to know what that costs when you aren’t making Apple-esque quantities in China
In the past, we’d do a steel clamshell and a thin aluminum front panel, like we did on Cobalt. Three pieces. Two steel. That’s good, because steel is far less expensive than aluminum—and if it’s damaged in handling, it’s easy to send it back to be refinished.
But we had another problem: getting rid of heat. Amps need heatsinking, especially Class-A amps like we were thinking about making. Traditionally, you use board-mounted heatsinks or hang a slice of heatsink extrusion out the back of the amp. The problem with those approaches was that they were pricey for the amount of heat we had to get rid of. In the case of the heatsink extrusion, it also meant another cosmetic part, and another chassis component, to deal with.
That’s why we soon decided to use the chassis itself as a heatsink—economics. The problem with that was that it killed the old “steel clamshell and front panel” design. Steel doesn’t work very well as a heatsink. The chassis would have to be aluminum. And it would have to be fairly thick aluminum, too, so it could effectively spread the heat.
At first, I thought about extrusions.
Note: extrusions are where you take metal and squeeze it through a form, like toothpaste. Except way hotter.
In fact, originally, the 01 (our extremely imaginative name for our first product from our as-yet-unnamed company) was supposed to be a 7 x 7 x 2” sharp-cornered square, with the volume pot set exactly in the middle and slot vents on the top. The 7 x 7” square would be a custom extrusion, and the top and bottoms would be flat aluminum panels.
One catch: what do you extrude?
• The outside perimeter, like we originally planned? Wow, that’s a BIG extrusion. Sure, you can do it. Bring $30,000 or so. No.
• An L-shape for top and front? Nope, not enough surface area to get rid of the heat. And it’s still a big extrusion. No deal.
• A U-shape for front, top, and bottom? Now you have the area you need, but now you have to deal with extrusion tolerances. The open end of the “U” might be too far open—or too closed. Notgonnahappen.com.
But a U-shape…that could be bent from sheet aluminum. And it could be grained before bending. And you could precisely control the tolerances. And, combined with a steel inner “sled,” you had a simple two-piece chassis.
Of course, the first drawings were still 7 x 7” square. In fact, the first prototype of what would become an Asgard was designed for a 7 x 7” square. At least until reality intruded, in the form of the transformer.
The transformer. It was great, because it was small, cheap, and efficient, and mounted right on the PC board. It was a flaming hunk of crap because it would peg an EMF meter across the room. This meant that any PC board traces running close to it would automatically pick up hum from its magnetic field. In a square chassis with the transformer in the center, that meant nearly every trace. It hummed so bad it was unlistenable.
And nothing would kill it completely—none of the off-the-shelf transformers I had, nor a custom one I had made at MCI Transformer, with fancy interleaved windings and a copper ring to cut down the field.
But if I simply moved the fancy MCI transformer away from the board, the hum disappeared—whisper quiet.
So that design went in the trash, and I drew up a new chassis. That’s how the 01 got its 6 x 9” form factor with an offset volume pot—to move the transformer away from the input traces and circuitry. Seems really simple, in retrospect. Like I said, I was rusty.
But that was really just the first step—figuring out what the chassis would be. Next was the big question: Would it be inexpensive enough to make us competitive? For that, we needed quotes.
Let’s pause here and talk about manufacturing. When you need something made, you have two choices:
1. Buy the machinery to make it yourself.
a. In the case of aluminum and steel chassis, this means $100K CNC mills, punch presses, laser cutters, a precision brake, a timesaver, an anodizing tank, powder-coating equipment and an oven, silkscreening gear or laser engraving.
b. Plus people to run all this equipment.
c. Plus stuff you probably just can’t do—try to get a new anodizing shop approved in California. Have fun with that.
2. Contract with someone who can supply finished parts.
a. Very simple!
b. Until they screw up.
c. More on (b) later. As Mike says, “There is nothing more certain than death or taxes than your metal supplier will screw up eventually.”
Okay, so let’s say you’re sane and go with (2). Now you need to find a manufacturer who understands:
a. What “consumer level” finishing is
b. That your deadlines actually mean something
c. You expect them to hold close to the pricing they quoted after the first run
d. You are not an aerospace company or government contractor (translation: you are not made out of money)
This really isn’t as bad as it sounds. Clear communication with any outside supplier is absolutely key. Most metal suppliers in the USA are not doing consumer products. They’re making instrument panels for submarines, or screws for aircraft, or heavy frames for industrial equipment, or precision-machined stuff for scientific gear.
This means that if you expect to get consumer-level products (that is, nicely finished with a very low rate of cosmetic imperfections), you need to go in and show them. Clearly explain what has to be perfect, and what isn’t cosmetic. And pay for a “first article,”—this means, sample—to see how close they can get. If they can’t get it in 1, run.
Anyway, back to MFG.com and local suppliers.
In Southern California, we’re lucky to have the remnants of a manufacturing base. It’s largely left over from the aerospace heyday, but the ones who survived have learned how to do consumer and industrial products. So we had a few to choose from.
But I also wanted to get a bigger perspective, so I also looked at MFG.com. If you’re looking to have something made, it’s a great sanity check. You can select from manufacturers around the world, or limit it to the USA, and you can have virtually anything quoted, from machining to plastic to metal injection molding. If you limit the search to the USA, don’t expect a flood of quotes, but you will definitely end up with some options.
Now, if all of the above sounds like a lot of work, it was. Between doing schematics and laying out boards and researching suppliers and screwing up the first design and having to do it all over again—while at the same time buying a new scope and other assorted test equipment, getting prototype parts, doing research online, etc, it always seemed like I was running out to the garage (where the workbench was).
“I’ve got schiit to do,” I’d tell Lisa, and disappear.
She’s endlessly patient, but one day, she’d finally had enough. “Why don’t you just call it Schiit?” she shot back, crossing her arms.
“Call what schiit?”
“The new company. You’re always saying you’ve got schiit to do. Why not just call it Schiit?”
At first, I laughed. A company called Schiit? No sane company would do that. If we proposed that name to any Centric client, I imagined what they’d say. Way too out there. Can’t believe you’d propose that. Piss off too many people. What a crazy idea. Then they’d fire us.
But I’d had 15 years of marketing playing it safe, second-guessing everything we did, and watering down every great idea until it was meaningless. Maybe you can blame my decision on that history. Maybe it was nothing more than that.
And this company wasn’t about playing it safe. Hell, we were trying to reach Chinese prices here in the USA. And do it without a million-dollar investment. That was about as crazy as it got.
“Nobody would ever forget it,” I replied, finally.
“It would cut down your marketing costs,” Lisa agreed.
“And we could say we make some really good Schiit.”
Lisa laughed. “Why not? Go ape Schiit.”
“And Schiit happens,” I agreed.
“If you don’t have our stuff, you’re up Schiit creek,” Lisa added.
I nodded and sat back. Suddenly it didn’t seem so crazy. Hell, the word was meaningless for, what, 80% of the world that didn’t speak English? And if you spelled it funny, it could sound vaguely German.
Hell, the Teutonic connection opened up all sorts of stuff, including all the old Norse mythology. They named, like everything, from gods to spears to crows and forks. An endless source of non-alphanumeric names.
“Norse and German aren’t the same,” you say? Well, it doesn’t matter. We have comic books and movies to learn from. We’re dumb Americans. We mix stuff up. What’s more, we’re from California, where Thai-Mexican fusion food sounds like a good idea (and really is.)
And a name like Schiit would be unforgettable. Nobody could ignore it.
And, fact is, great marketing polarizes. Some people hate it. Some love it. An ad that hits the middle ground of “nice” is pure crap. Which is what most companies shoot for. Might as well cash out the whole marketing budget, roll logs of $100 bills, and have a big bonfire.
But we weren’t here to hit a nice middle ground. We were here to be unforgettable. (And hey, we didn’t have that many hundred dollar bills, either.)
And in that moment, everything gelled. We would be Schiit.
