Chapter 31:
Name Me One Non-Standard Format That’s Succeeded, Ever, Or, A Trickster Cometh
The DSD surge started in 2013, shortly after the announcement of the DOP (DSD over PCM) USB protocol.
It started pretty innocuously. Starting in early 2013, we started receiving a few emails asking if we were planning to add DSD decoding to our products. It was a literal handful to start, but as we got into spring, the inquiries started coming faster, as a number of companies introduced DSD-compatible DACs.
As the company’s marketer, I wondered if this surge in inquiries would become a movement, so I asked Mike about the possibility of adding DSD to the existing Bifrost and Gungnir.
Mike groaned. “DSD. Argh. No.” I waited for him to explain, but he didn’t go any further.
“Why not?” I asked. I actually knew some of the technical reasons, but I wanted to hear it from Mike.
“DSD requires completely different filtering,” Mike said. “It’s essentially wideband noise. You want that going to your amps?”
I shook my head. Running ultrasonic noise into an amp is a good way to test it to destruction.
“So, we need way more aggressive filters to get the noise out,” Mike said.
“But, technically, the AKM DACs do DSD, right?”
Mike shook his head. “Technically, yeah. But
doing DSD, versus
doing it right, are two different things. If we put in the DSD-appropriate filtering, we’d be compromising our analog stage performance for PCM. And it’s not as simple as switching it in and out, because that would require more space on the analog board, and I don’t even know if we have the hooks on the analog board input, anyway.”
“So no DSD,” I said.
“Not without a lot of changes. For Bifrost, we’d need a new USB input board, a new main board, and a new analog board. Technically, yeah, that’s just upgrades—“
“—but it’s essentially a whole new product.” I finished for him.
Mike nodded.
“So what if DSD gets big enough to matter?”
Mike laughed and waved a hand. “Remember HDCD?”
I nodded. HDCD was a technology of the early 90s that was supposed to be the One True Savior of digital, allowing more dynamic range to be encoded on special disks that could only be decoded by a specific digital filter.
“HDCD almost took down Theta,” Mike said. “We got in screaming arguments about it. My marketing guy said the same thing you did: ‘What if it gets big? Everyone else is doing it. We’re going to lose sales if we don’t have it.’”
“I didn’t say those last two things,” I told Mike.
“Yeah, but just asking about DSD implied it,” Mike said. “You’re worried that we’ll lose sales, or we’ll miss out.”
I shook my head. Though Mike was right, in a sense. If DSD became big, we’d be vulnerable to other products that offered DSD playback.
“Stop worrying,” Mike said. “Where did reel to reel go? Nowhere. Where did quadraphonic go? Nowhere. Where did Elcassette go? Nowhere. Where did DAT go? Nowhere. Where did minidisk go? Nowhere. Where did HDCD go? Nowhere. Where did SACD go? Pretty much nowhere. I expect DSD will pretty much go the same exact place.”
“But what if it doesn’t?”
Mike groaned. “These special formats all end up the same place, because there’s no software for them. When there are more DSD downloads available than SACDs, let me know. Then I’ll start worrying.”
But the Inquiries Kept Coming
In fact, they intensified. As the press flogged the new shiny thing known as DSD, we began to get several inquiries a day—on slow days.
“Mike, we should do something about DSD,” I told him, finally.
“Ignore it,” he said. “It’ll go away. It’s just the press. They’re so monumentally bored, they’ll talk about anything, including a non-starter like DSD.
“But what about, just, you know, as a CYA.”
Mike sighed, and was silent for a long time. He knew we were getting inquiries. He knew some people really wanted DSD. And here he was, between his partner’s paranoia and his experience with dozens of nonstandard formats that have come and gone.
“You want me and Dave to divert time from Yggdrasil to work on this?”
I crossed my arms. That was the ultimate threat—taking time away from a product that was literally the antithesis of DSD, and which we believed would help redefine the digital market in toto, to work on something that could be a passing fad.
Mike laughed. “You do.”
“I just think it would be safer—“
“To do what everyone else is doing,” Mike finished. “To jump off the cliff, just because everyone else is doing it.”
“Can we really not afford to take a look at it?”
Mike looked thoughtful. “Okay. Fine. I’ll think about it. That’s all I’ll say right now.”
Mike’s Thoughts
Time went on. DSD inquiries continued. I watched our sales cautiously, but they kept increasing for all the DACs—definitely not an indication that DSD was a gotta-have thing.
But the press kept flogging it, the articles kept coming out, and rumblings of lower-cost DSD DACs started to surface (prior to this, DSD DACs were pretty eye-wateringly expensive.
Eventually, Mike came back to me, grinning like a fool.
“Okay. Here’s what we do. We make the least-expensive DSD DAC on the market.”
I blinked. “What?”
“If they want this format to succeed, they need wide adoption. And you ain’t gonna get wide adoption for a grand and a half.”
“So it replaces Modi?”
Mike shook his head. “No. It’s a standalone DSD-only DAC. That’s why, even though it’s gonna be the cheapest DSD DAC out there, it’s still going to sound insanely good. We’ll do the filtering right. Just for DSD, and only for DSD.”
“But what if you want to play both PCM and DSD?”
“If you’re so into DSD, convert it on the fly,” Mike sniped.
“Seriously.”
Mike looked thoughtful. “Put a switch on it. Then you can run the output of your current PCM DAC through it.”
I sat straight up. “So you can use it to add DSD capability to any DAC!” I cried.
Mike nodded, looking very pleased with himself. “Exactly.”
I nodded. That was a perfect fit with Schiit’s ethos.
Keep your existing DAC, add DSD, see if you like it, then go from there if you do. Instead of throwing your existing DAC away to get a DSD-compatible one.
“How cheap are we talking?” I asked Mike.
Mike grinned. “Not much more than Modi.”
Okay. Now this was getting good.
Tech Challenges
There were just a few problems with this plan—starting with the fact that we didn’t have any DSD-capable USB receivers. The CM6631A we were using didn’t accept DSD streaming or DSD over PCM. C-Media was planning a CM6632 for later in the year, which would be DSD-compatible, but late in the year was too late for our plans.
Enter Dave.
Note: when I say, “It’s in Dave’s hands now,” that means it’s somewhere in complicated software/firmware land, from which it will hopefully emerge with working software/firmware at some future time.
Dave’s plan was simple, but somewhat insane: use a 32-bit Microchip microcontroller to do our own unpacking of the DSD-Over-PCM standard, and then send that along to the DAC. Yeah, quite a programming feat. But he did it, and soon we had a prototype that could play native DSD, using a Crystal Semiconductor DAC.
There was only one problem: it sounded like crap. Dynamically compressed, soft, boring, and lifeless. Yeah, I know, it measured fine, so it should sound fine, right? Not in this case.
“Why’d you do Crystal?” I asked Mike, one day when Mike, Dave, and I were together at the Schiithole.
“Crystal will do 2X DSD,” Mike said.
“But the Microchip controller would have to be faster to unpack it,” Dave said.
“So we can’t do 2X DSD right now?”
Dave nodded. “But with a faster processor, we could.”
“But not now,” I confirmed.
“Right.”
I frowned. “Why don’t we just use AKM, then? We know they sound good.”
Mike shook his head. “We have no idea what they sound like when they’re fed DSD.”
“Isn’t it worth a shot?”
Mike and Dave looked at each other. Dave shrugged. Mike sighed.
And a few weeks later, we had another prototype—this one with an AKM DAC. And it sounded worlds better. It still measured pretty much the same, but it had a lot more life and energy. It had dynamics and pace. Both Mike and Dave smiled when they heard it.
“But AKM doesn’t do 2x?” I asked.
Dave shrugged. “It might do it undocumented, but--”
“—but we don’t know,” Mike finished for him.
I sat silent. Should we wait to add 2X capability, for the literally 20-30 recordings there were out there done in 2X? It would mean another prototype cycle, and maybe different code, and maybe some unforeseen problems.
“And it would take different filtering,” Mike said. “This is as good as it gets for 1X DSD. Throw 2X in there and we start having to make some different decisions.”
Still, I sighed.
“Let me propose a solution,” Mike said, as he usually does when I’m hesitant about something. “Let’s bring this to market, see how it does, and if DSD really takes off, we can work on a 2X solution, or whatever we need.”
I nodded. That made sense.
“Good.” Mike said. “Though I doubt if we’ll ever have to do any more work…”
Peak DSD
Mike seemed confident that DSD was a non-starter, but his comment seemed to be out of sync with the public at the next TheShow Newport, which we attended a couple of months before introducing Loki. At TheShow Newport 2013, literally every other question from passerby was, “When will you support DSD?”
Mike still didn’t look too worried.
And, although I didn’t know it at the time, that was the absolute peak of DSD.
Loki Cometh
We introduced Loki under the banner of “Add DSD to any DAC for $149.” At the time, the least-expensive DSD-capable DAC was $849, so this was quite a coup.
Or so we thought. It turned out that the idea of a DSD-only DAC and switching system was a little more challenging than we thought. Some people thought we were converting DSD to PCM and running it to the main PCM DAC (why, when you can simply do it in software?). Some people thought we were taking the analog output of their DAC and converting it to DSD (fat chance on that one.) Some people really, genuinely wanted to throw out their old DAC, rather than run DSD through it.
And lots and lots of people didn’t like having to run two USB cables (one to their main DAC, and one to Loki), and switch between the two on their playback software. For a main DAC fed by SPDIF, the switchover was easier (and seamless if they were using a different player, like a CD player, for their PCM content), but it still wasn’t something that most people wanted to do.
That, combined with the appearance of new, inexpensive DSD/PCM DACs, quickly cooled Loki’s sales. Mike will still argue that doing PCM and DSD in the same DAC is a compromise, and the math (and measurements) are on his side, but convenience usually wins out over sonics when you’re playing at the lower end of the market.
Still, these wouldn’t be insurmountable problems if we wanted to do, say, a Loki 2 with automatic interface switching. It could then interface seamlessly with a PCM DAC. But it would be significantly more expensive, especially if we added DSD 2x (or 4X, or 1000X, or whatever the latest unicorn format is today.)
And I suspect that’s the way we’d end up going if we were to continue pursuing DSD—not adding it to our current DACs, but making a seamless, dedicated DSD DAC with interface switching.
But as of the time of this writing, I don’t think it’ll happen.
DSD Today
Today, Mike has crossed his arms and declared, “No more DSD development, unless something really big happens.”
Why?
Because, from our point of view, it looks like we’re past the peak. Despite dire pronouncements from other manufacturers saying, “You can’t move a non-DSD DAC with a boxcar full of Ex-Lax,” we haven’t seen it. Cases in point:
- Sales of our DACs continue to increase—and to accelerate
- Inquiries at TheShow 2014 included literally two (that is, 2) half-hearted questions about whether or not we were going to support DSD, in stark contrast to the literally 200 questions the year before
- Sony’s presence at TheShow (featuring DSD prominently) was a ghost town
- Email inquiries have fallen from a dozen a day to maybe one or two per week
- The predicted “opening of the vaults of DSD” hasn’t happened—there are still only a few hundred recordings available, many of which have questionable provenance (more on that later—and even SuperHires’s announcement about Warner probably won’t answer even a tiny fraction of the questions on provenance, and prices remain TBD)
Sure, there are plenty of DSD-capable DACs out there, including some that do 4X and 8X DSD…but where’s the software?
Let’s face it:
- The bulk of the industry remains compressed streaming—and that ain’t DSD
- Pono is another question mark in high-res—and it ain’t DSD
- Apple could come out of sleep-mode on high-res if Pono or other products prove there’s a demand—and if they do, they define the market—and they won’t be using a Sony format, I bet
And, the elephant in the room:
- The most important part of a recording is the master—paying attention on that side will reap benefits beyond any format
So What About the Future
Okay. Let’s say the next Sony reorganization (they ain’t exactly healthy these days) doesn’t kill DSD, but results in them releasing 20,000 DSD recordings of popular artists, all with DSD-guaranteed-from-the-start provenance, for, say, $5.99 an album.
Would this result in a whole lot of DSD out there? You bet.
Would it be a game-changer? Absolutely.
Would it have us dusting off plans for a Loki 2, or working on ways to include DSD decoding in our DACs without compromise? Yeppers.
But I think that scenario is about as likely as the disembodied head of Steve Jobs giving the next Apple Keynote.
What’s more likely is this:
- DSD recordings will continue to be a small part of the market
- DSD recordings that actually start as DSD, or were converted direct to DSD from master tapes, will be an even tinier part of the market
- High-rate DSD with the same provenance will be even smaller
- DSD recordings will continue to be very expensive
- Some people will continue to really like DSD, and will flip us off as they pass us at shows
- More people won’t care, as long as the music sounds good and doesn’t cost a fortune
- Even more people won’t care if it’s DSD, PCM, or compressed, as long as it’s available to download at a good price
- And the vast majority of people will never have any idea what the hell us crazy audiophiles are talking about, as they happily stream compressed music for a small monthly fee (or free)
- About 10 years from now, a new quantum-based encoding format will come out so that everyone can buy their music again…
How about we deal with the elephant in the room, before worrying about formats, hmm?
.
To be equally pedantic, the term "bit perfect" is simply concerned with the digital signal transmission process. It means that the original sample is the output and that it has not been changed in any way. No filters, no re-sampling, no signal processing. It has nothing to do with the D-A process, and the necessary filtering and smoothing that requires. It simply means that the entire original sample is used in that process and it is not altered in any way. A DAC is much more than a bit-perfect transmission device, but at the same time no delta-sigma device is genuinely bit-perfect due to dithering.
