Separate names with a comma.
Love it Jason! I hope I get a passing mention. :atsmile:
Looking forward to following this! I have a Bifrost/Uber in my bedroom system along with a Modi/Magni for headphone listening and a Asgard 2 in my office. I love their Schiit!
Definitely looking forward to reading this, especially considering that a Schiit amp and dac are on my shopping list for this year
Here we go!
To the people who make Vali, (one of the best TOTL sounding amp for budget a price) and Uber Bifrost (best performance to price ratio dac). I'm in.
The Line is Down. Here’s an Undocumented Test Rig. Fix It.
My first day at Sumo in 1989 was maybe the most bizarre first-day-on-the-job ever.
I literally just walked in the door to find Ed Miller pacing the hallway, eyes darting from left to right. He’d clearly been waiting for me to show up.
“Oh good, you’re here,” he said, motioning for me to follow him.
I trailed him past a series of nondescript offices, out onto the factory floor, and into a cluttered, messy little room with a giant glass window that looked out onto the PCB assembly line.
“We have a problem,” Ed said. “I hope you can fix it. I took a look, but I don’t have the notebooks. But if you can’t fix it, maybe you can whip something up.”
I wasn’t really listening to Ed. I was staring at the battered blue-and-white engineering workbench. Or trying to. Because the entire top surface of the workbench was littered with about fifty pounds of discarded resistors, capacitors, output transistors, screws, wires, solder blobs, candy wrappers, scrawled post-its, and other unidentifiable electronic and non-electronic trash. And this was what was on top of a hunk of low-nap industrial carpet. Below it was about another inch of the same kind of junk. Below that, there was another piece of carpet, considerably more frayed, and another inch or two of junk.
Yes, this was what the previous engineer did. Instead of cleaning up, he just threw another rug over it and piled more junk on top.
“So, want to have a look?” Ed said, pointing at a battered aluminum Bud box in the middle of the carpet-desk. It had ten red LEDs on it, a socket for a power transistor, and a pushbutton.
“What is this?” I asked, trying to come back to reality.
“It’s a MOSFET matcher. But it’s not working. And the line is down,” Ed said, pointing out the window to the PCB assembly team. Outside, ten women looked back at me, arms crossed, clearly waiting for something. “They can’t get back to work until the MOSFET matcher is fixed.”
Ah, crap. It finally sank in. They were waiting for this test rig. Everything was at a dead stop without it.
And the job of fixing it had just been dumped on me.
I said nothing, suddenly realizing just how much I was in over my head. I was seven months out of college. I didn’t even know MOSFETs needed matching. I didn’t know what they matched. And I certainly didn’t expect to see some homebuilt device at my great new engineering job. Hell, I’d just come out of working on spread-spectrum communications at Magnavox APS’s “black hole” lab, where $100K gate array prototypes were all over the place.
“Do you have a schematic?” I asked, feeling a little ill.
“No. That’s the problem.”
Great. Just great.
At that moment, I thought about asking for my old job at Magnavox back. They told me I had an open door any time. And it was easy. Real easy. Hell, they got pissed when I worked too fast, because they couldn’t bill out their entire project to the government. They’d never throw me into something like this. Hell, they’d given me two weeks just to play with the layout software when I started at Magnavox.
Maybe I didn’t want to be in audio after all.
But no. I wanted to be in audio. I loved audio. I had a speaker company on the side, one I’d started in college. My senior project in engineering had been a switched-capacitor adaptive noise reduction system. I had years in audio. I always dreamed about going into audio. I’d almost fallen off the chair when I’d seen the Help Wanted ad in the paper, and its enticing offer to “join the Sumo engineering team, and advance the future of audio.” I’d been beside myself when I was asked to come in and interview.
And that was when I made the decision to stay. I couldn’t let it all go now. I’d figure it out. Somehow.
“I’ll take a look,” I told Ed, and shooed him off.
And then I sat there, crapping my pants. The women were still staring at me. Nobody was working. Everything was in my hands.
No pressure, no pressure at all.
Business lesson 1: say you can do it. Then deliver—at all costs.
I can make up a story about how I brilliantly fixed the tester, but I really just got lucky. The problem was just a bad connection. I reattached it, and it worked. I was a hero.
But I still knew absolutely nothing about what I’d gotten into. Looking at the schematics of the Sumo products was like looking at a Olde English codex—massively confusing and completely incomprehensible. Why did they use so many parts? What did they all do?
Don’t laugh. I was a green engineer. Very green. Sumo was only the second “real” job I’d had since graduating. And it was leagues away from the regimented, spotless, cutting-edge environment of Magnavox APS, where I’d mainly done software and PCB layout. It was messy, old-school, a union shop, and part of a larger company (Califone) that owned it, at least for part of the time I was there.
I didn’t believe companies like Sumo existed in the brave new world of nearly-1990. When I interviewed, Ed showed me their layout room, where they taped up all the PC board artwork, and their blueprint machine, where they ran copies of hand-drawn schematics. Those things had been done with software and pen plotters at Magnavox.
“You don’t actually use those things, do you?” I asked him, looking around for the CAD workstations. But there were none in sight.
Ed just laughed, a little nervously, and shooed me off to the rest of the interview. At the time, I thought his titter meant, “Of course not.” But it turned out it meant, “How else would we do it?”
And yes, the first boards I did for Sumo were taped up. It was what had to be done. And I learned a lot from doing it. In 6 months, I understood what made a good amplifier (or so I thought) and was designing new Sumo products as Chief Engineer. In two years, I was Vice-President, Engineering. And by the time I left, I’d gotten us into new markets with new products, and made the company much, much more efficient in terms of production and parts commonality.
But it wasn’t all rosy.
This second anecdote is should be a poster for What Not To Do, Ever, at any company.
My second job at Sumo after the MOSFET tester heroics was debugging their shiny new Andromeda II amplifier. It was selling briskly, because it was one of the few amps that could drive the insane load of the Infinity Kappa 9, a popular speaker at the time. The problem was, Andromeda IIs were also coming back, blown up, in scary numbers.
I dove in, went back to the books, and quickly noticed a big problem with the amp: the P-channel MOSFETS were only rated for 75% of the current as the N-channel MOSFETS, which meant they were much weaker than Sumo’s previous engineer might have expected. Combined with a slow breaker-based protection system, that could end with blown-up parts in a hurry.
I found an alternate part and dropped them in the amp, then had to deal with some instability problems. Eventually, it worked and didn’t oscillate.
But the revised amps, in my testing, still smoked. And by smoked, I mean, parts would literally catch on fire, and flames would come out of the vent holes. What’s more, it did it after the protection system did its job and blew a breaker!
It was a mystery. What made it even more mysterious was the fact that the amp didn’t always smoke. After several rounds of testing, an epiphany hit: think like a customer. What would they do if their amp stopped making music when the breaker blew? They’d probably turn up the volume on the preamp to see what it would do.
And, you know what? Every time I turned the volume up after the breaker blew, the amp smoked.
Digging deeper, I found that the new power supply didn’t completely shut down when the breakers opened. Which meant the front end, and the drivers, could completely drive themselves to death (fire) when the breakers blew. I rigged up a circuit to shut it down, and voila—the revised Andromeda II behaved as it should. No blown MOSFETs. No smoke. Just a relay click and nothing.
So, I did what any idealistic young engineer would do—I told everyone in the management meeting that we had to stop shipping Andromeda 2s immediately, revise the boards for the new outputs and the more advanced protection system, and then ship new, perfect, safe amplifiers.
So what did they say? Come on, you know what’s coming, don’t you?
They said, “Are you crazy? We have to make numbers this month. Ship them anyway.”
And what did they do? They shipped them, of course. Lots and lots of them. I think about 70% of them eventually came back, and contributed to an extreme service load that never went away in my 5-year tenure.
Business Lesson 2: Don’t ship stuff that blows up. Ever.
Yeah, I know, it sounds like common sense. But it’s amazing how common sense can go by the wayside in companies that live and die by receivables financing. When you hear “we have to make numbers for the end of the month,” be scared. Very scared.
Sumo was, by and large, the company that taught me what not to do.
Exhibits of What Not To Do:
1. The example above—shipping stuff that you know will break. Come on, this isn’t rocket science.
2. Don’t ever tell anyone, “I don’t care that you say it can’t be done at that price. I’m VP of Marketing, I make four times what you do, and I’ve already sold a bunch to our dealers. That’s the price, make it work.”
3. Re the above: never sell anything you haven’t made yet. Period. Ever. Nor tell someone about stuff that’s coming up, or show products that aren’t yet products. If they aren’t on shelves ready to ship, they don’t exist. (It took us a while to remember this one when we started Schiit--oops.)
4. Don’t lose customer returns. Or use them to fix other customer returns. Yes, I know, more common-sense stuff.
5. Don’t try to go too broad. In Sumo’s case, this meant getting into speakers. The Sumo Aria was an amazing planar speaker. Also, it was an amazing pain, because most of them broke, and they were made on a contracting basis for us by an outside company.
6. Re the above: speakers seem to be easy. Anyone can put a driver in a box and consider themself a speaker designer. That’s why there are so many speakers out there. What you really want, when you’re creating your niche company, is something with high barriers to entry. Electronics are a lot tougher than speakers, with lots of components and safety standards and FCC. There will be less competition, which is a good thing.
7. Don’t be cheap, especially to the point of having the checks be late. Especially paychecks.
But Sumo had one thing going for it. Its heart was in the right place. We were trying to make inexpensive components that could compete with the “best of the best.” We didn’t do insanely overwrought chassis for megabuck amps. And this taught me to be efficient, and work with what we had. It’s one of the reasons we’re good at production engineering now.
Business lesson 3: don’t dwell on the negatives—learn from them.
And Sumo was where I was converted from a hardcore objectivist to subjective-objectivism.
“Ah hell,” some of you are saying now. “I don’t want to hear this hoo-ha about how all amps sound different. Properly engineered amps run within their limits all sound the same, anything else is placebo/misdirection/insanity!”
Amen, brother. Or that’s what I would have said in 1989. By 1990, I wasn’t so sure.
When I started at Sumo, I already had better amps than they made. At least in my mind. Two Carver M-1.5T monoblocks, 350 watts per channel, sporting Bob Carver’s latest and greatest Magnetic Field Power Supply, or something like that. They were lightweight and ran cool. Light-years ahead of Sumo’s giant, hot, heavy Andromeda II, which was only rated at 200 watts per channel.
And I knew those amps well. I had a small company manufacturing speakers, called Odeon. (See the note on speakers above.) We used the Carvers for testing, development, and demos. We had them up to wall-shaking levels. They were pretty damn good.
So, when Sumo’s president said, “Hey, take home an Andromeda 2, and let me know what you think,” I wasn’t too excited. I put it off. And I put it off again. I didn’t want to tell him the Carvers were better, or lie and say the Sumo was better.
Eventually, I took one home. I told the other guys in the speaker company, “Hey, look at this dinosaur, there’s no way it will beat the Carvers.” It was all a joke.
Until we turned it on.
Holy crap. Not only did Sumo’s “underpowered” amp wipe the ground with the Carvers in terms of higher output, it also sounded better. Way better. The rest of the team came out of the other room and just stood there, dumbfounded.
I think Eddie spoke first. “So, when do we get one of these to keep?” he asked.
And that’s what started me down the road to subjective-objectivism. Not pure subjectivism, of course—measurements are still very important. But it led me to dig into the reasons why the Sumo amp sounded different than the Carvers. It started with simple things, like current capability and rated power.
It continued into gain stage structure and out of band performance, and led to me dramatically changing Sumo's amplifiers--to the point of doing a zero-feedback, single-gain-stage preamp shortly before I left (Artemis—very, very rare) and several no-overall-feedback amplifiers (The Ten, The Five, Andromeda III) that measured as well as the full-loop-feedback amps that preceded them, but sounded much better.
And yeah, yeah, I know: You're crazy. All amps sound the same!
Business lesson 4: don’t discount personal experience.
There are a thousand other stories about Sumo, but let’s cut to the part where I met Mike Moffat of Theta Digital.
At the time, Sumo just happened to be in the same business park as Theta. I knew who the company was, of course. Everyone in audio knew Theta. They were really tearing it up in the DAC market. And Sumo, like everyone else at the time, was working on our own DAC. In many ways it presaged what we’re doing today—it was modular and upgradable, and the DAC card itself could be added to a Sumo preamp.
But I didn’t want to meet Mike. No. It was too intimidating. Plus, they made expensive stuff, and we made cheap stuff. Plus, he’d probably be a golf-playing blowhard who was too full of himself.
But Sumo and Theta shared a components sales rep, and she kept insisting I meet with them for drinks. After a while, I relented. And I forgot all my rationalizations for avoiding the meeting.
Mike was, and is, a character. Instead of being uptight and high-and-mighty, he was very casual and approachable. Hell, after he got to know us at Sumo, he’d sneak out of Theta to come over to use our bathrooms to change into a suit when he was going to the opera. He didn’t want his employees to see him in a suit, but the LA opera was the only place he could hear unamplified music.
At Theta, I found the company I wanted Sumo to be.
Mike’s Theta ran on incentives. Employees were paid bonuses based on the number of units shipped, and on their individual performance, and on stepping up to do tedious things, like upgrades. Some of his techs made several times their salary in bonuses. The office was casual to the point of not even having part numbers for their parts—instead of a 05-1225, a 1K 1/2W 5% resistor was called a 1K 1/2W 5% resistor.
Yes, I know, not very exciting. But consider the results: in ½ the space of Sumo, Theta was selling 10x the dollar volume of products. Their net profit was easily 8x that of Sumo.
I told the president of Sumo this. His response: “That’s stupid, paying people bonuses. Then you have variable salary cost, you can’t predict it.”
I wanted to say, “Well, it seems a whole lot more stupid to run an inefficient business like this,” but for once, I said nothing.
It was becoming very clear that nothing I could do would change the way the company ran.
Business lesson 5: be open to meeting new people, and transformative ideas.
And that’s why I started moonlighting for Theta. Mike Moffat was really intrigued by the idea of doing an inexpensive DAC. After more dinners and more drinks, we finally hatched the idea of Cobalt. The Cobalt 307 was my design, with input from Theta—a true hybrid of Sumo’s ideas and Theta’s ideas.
Cobalt blew up the DAC market, selling 1000 per month for some time—which happened to be about 2.5x the total market size according to one industry pundit. The combination of solid name and inexpensive price really set the high-end world on fire.
And—I think it’s important to note here—“inexpensive” would seem pretty pricey today. The Cobalt 307 was $599. In 1993. That’s about $970 today. If we’d been able to sell Cobalt direct to the customer, like we do Schiit, it would have been $349.
Yes, that’s how much the dealer takes. More on that later.
But in 1993, selling direct wasn’t feasible. We would have had to take out full-page ads in all the magazines to the tune of $20,000 or so a month, and we would have had to have multiple employees in a full-time call center to take orders. There was no Amazon Marketplace. No Shopify. No pay-per-click advertising. Hell, there was no viable internet. It was a different world.
At Theta, I also designed the discrete, current-feedback output stage of the top-end Theta Gen V, mainly on a bet. Mike and Dave—Mike’s lead engineer at the time—were convinced that op-amps were the way to go, but I’d learned enough about discrete design to know they were wrong.
“I can design a stage that will work better than any op-amp,” I told them.
“Even on measurements?” Dave asked.
“Even on measurements.”
They took the bet, and I came up with a design that was an exercise in insanity. 260 parts on a 4 x 6” Teflon circuit board, with two PCM63 DACs in balanced configuration.
But it beat the op-amp stage, both in measurements and in the listening room. And that’s how the Theta Gen 5 was the first discrete output DAC that Theta made.
Business lesson 6: take a chance, do crazy things…a lot of times, it’s worth it.
Even as Theta was kicking ass, audio was getting, well, weird. Theta stuff cost a lot to make, so it was priced very high. And we had the dealer vig to pay, of course.
But Theta’s products weren’t priced high because they were lookers. It was all about the technology inside.
Theta’s competitors took a different tact: make it pretty and even more expensive. Theta’s balanced Gen V was $5,500. Mark Levinson “outdid” Theta with a $16,000 DAC. Krell upped the ante with $32,000 amplifiers. The magazines ate it up. The race towards “gold-plated Bentley” audiophilia was on.
Mike and I didn’t get it. He didn’t want to put a $250 board in a $2500 chassis. He wanted to make game-changing stuff. But it seemed the magazines were only interested in the megadollar price tags. Eventually, that led Mike to start Angstrom and get into the field of surround sound.
Me? I went evil. I went into marketing...
Hey! This is head-fi, when does the audiobook version get released???
I have yet to buy a Schiit product but I do look forward to it.
Thanks for putting the story out there! I do get a kick out of every new announcement or product you guys make.
I still have a Cobalt 307 with the TLC power supply in one of my systems. Nice sounding little DAC. I have a Sony ES CD changer and Sony ES Mini Disc player hooked to it. Does not get much use these days. Very interesting read......
Love my Lyr (brings out the best in all my headphones), and I'm very interested in reading more. Mr. Stoddard, thanks so much for sharing this!
I have a Cobalt 307, also. Cool.
Good schiit, Jason.