2016, Chapter 13:
Into the 2-Channel World: The Saga of Saga
(and, um, Freya too)
Saga and Freya? Yes. As in, both were developed at the same time, so I think it's best to talk about them together.
But I'm really getting ahead of myself. We shouldn't be talking details at all yet. Even product types are more detail than we need. In fact, what we need first is a primer on what "2-channel" is, how it differs from "desktop audio" or "personal audio" or "multichannel" and "home theater."
So let's start with that.
Backing that Schiit Up: A Brief Disambiguation of Types Of Audio, with Bonus Historical Perspective
Okay, let's start at the start. Once people had figured out that audio was, in fact, comprised of pressure variations in air, they soon discovered that they could record it and play it back. The first commercially viable system was, of course, Edison's wax cylinders. They were created by cutting a recording directly onto the cylinder, and played back using a purely mechanical amplification device—a needle, diaphragm, and horn.
If this sounds less than ideal in, well, all terms—from the slow cutting of audio cylinders, to handling and mounting the cylinders themselves, to the just-a-few-minutes play time, to the inevitable squawks and resonances of the mechanical horn amplifier, to the fact that the playback devices didn't have a volume control (it's mechanical, remember? Want less volume, get a smaller horn. Or stuff a sweater in it. Literally.), it sounds less than ideal because, well, it sucked major balls. But when it's the only thing ya got, well, it's pretty miraculous stuff.
Why do I go so far back? Because, at this time, nobody thought in terms of more than one channel of sound. Everything was mono. Even when advances came in playback technology—starting with flat records that could be pressed in molds, rather than painstakingly cut on a cylinder—we were talking about mono sound.
Fun fact: 78 RPM records had a play time of only 3 minutes per side. Yes, three. As in, three minutes. Sit back and think about that for a while. And, as an added bonus, the records were full of abrasive. Yes, go back and read the previous line again. Instead of smooth, quiet vinyl, they purposely loaded the record with grit that would cut the needle to fit the groove. No, you can't make this Schiit up. But if you ever wondered why there's so much background noise on a 78 RPM record, well, the fact that it's full of abrasive scratching against a crappy dull needle is the reason.
(And, a corollary to this fun fact: some audiophiles still consider 78s to be the peak of recording fidelity. Why? In many cases, there were no electronics in the signal path—it was a purely mechanical recording. Yeah. Perfect sound forever...well, except for the mechanical resonances of the recording chain, the extremely limited bandwidth of the transducers and the medium itself, and the oh-my-gawd, what-the-hell-is-that-stuck-toilet NOISE. Note to the Luddites out there: sometimes there is real progress in audio. Honestly. Electronic recording and the LP were absolutely two of them.)
For a while, everything in audio was mono. Home playback was mono. Sound reinforcement was mono. Movies were mono. Performances were recorded in mono. Radio went out in mono (and, if we're talking AM, extremely band limited—like 4kHz top end there.) You get the picture. Mono mono everywhere, and not a "stereo" or "2-channel" to be seen.
This started changing shortly after the introduction of the LP (yes, there were MANY mono LPs). The recording industry, in an attempt to siphon as much money out of the pockets of their victims as possible....er, wait, never mind, let's say "in an effort to increase the overall realism of the recorded musical experience, and to provide a new dimension in sound," introduced the idea of "stereo" recordings—that is, recordings with 2 channels.
Stereo was a fundamental shift in the audio universe—and an absolute boon to the audio industry.
Let's start by talking about the fundamental shift first. Stereo was a huge shift in many ways, but the biggest thing it brought to audio was the ability to recreate a virtual "image" of the performance—the illusion of breadth and depth in the playback. Coupled with the very low-noise LP technology (and the insane luxury of having 22 minutes per side before you had to flip the record), stereo was a revelation.
Of course, many early stereo recordings didn't show off this virtual "image" to its maximum potential, instead relying on gimmicky "ping pong" stereo effects with instruments or performers hard-panned to one side or the other. But the potential for a much more realistic recorded music experience was there.
It just required a complete re-thinking of everyone's audio system.
And that's why it was such a boon to the audio industry. Instead of one channel of amplification, audio enthusiasts needed two. Instead of one speaker, two. Instead of a mono cartridge and mono preamplifier, they needed stereo sources.
So what happened? People went out and bought.
They bought big. They bought all-new stereo consoles that provided an all-in-one way to enjoy this new two-channel phenomenon, or they went and bought an extra amp, another speaker, and new sources. Sales swelled, and everyone was happy.
Aside: the notion of high-end cables wasn't a thing back then, so I'm not counting them as part of the overall economic boom, but yeah, people needed more cables, too—assuming they weren't using an all-in-one console.
Of course, everyone wasn't happy forever. Eventually, stereo became the norm. New companies had sprung up to provide affordable, highly integrated solutions for stereo listening—stereo receivers with AM/FM tuners and phono preamps built-in. Speakers were sold in pairs. Stereo consoles still did a booming biz. Stereo reel-to-reel tape began to creep in, both for recording and playback.
But sales were no longer exploding.
So what to do? The recording industry, remembering the huge success they had with stereo, tried first to increase the number of channels again—this time to 4. This, they dubbed "quadraphonic." The promise was an even deeper immersion in the recorded music. Which, in its way, made sense. Quadraphonic was kinda the first foray into surround, long before Dolby Pro-Logic came on the scene.
The problem was, quadraphonic records were difficult to encode and finicky to play back, and—the death knell—was that there just weren't that many of them. Audiophiles, by and large, decided to leave quadraphonic—and its associated sales of 2x the speakers, 2x the amplifiers, new sources, new recordings, etc—on the sales floor. It tanked.
From there, the recording industry maintained a healthy distance from formats with more than two channels. The digital audio revolution was entirely a stereo revolution, for example.
But that would change with the coming of home theater. With the advent of Dolby Pro-Logic (and, later, Dolby Digital and other digital surround formats), suddenly, 5 channels plus a subwoofer started to come into its own.
Aside: I think Odeon was the first company to sell speakers in 5.1-packs—5 satellites and a sub—but, as they say, that and $4 will get you a fancy coffee. And, added bonus, Mike's Angstrom company was the first to offer an upgradable surround-sound processor. So, yeah, we know a thing or two about this. And no, we won't be doing surround ever again, thank you. I've covered the reasons why before.
Why 5 channels, and not 4? And what's this with the .1?
It starts with the unique requirements of movies. 5-channel surround was designed primarily for movies, not music. There are two front channels, two side (or surround) channels, and one channel—the center channel—to help keep dialogue and center-of-screen sound, well, centered. Unlike traditional stereo, 5-channel surround is designed so that people can more realistically sit off-axis. The center channel is the biggest part of this.
The .1 channel? That's technically the "LFE", or low frequency enhancement channel—band limited below 100 Hz, it's really just for feeding to a subwoofer to create the thumps, bumps, and t-Rex footsteps that are a staple of movies today. So, in total, 5.1 channels.
For a while, home theater was going to be the Savior Of All That Was Right and Good and Profitable in the audio biz. And, on the surface of it, it made total sense. You needed 5 speakers (or more), plus a subwoofer. Plus a multichannel amp. Or more amps. Plus a surround processor, A/V preamp, or AV receiver. And, since we have now passed through the Gates of Neurosis known as "high end cables," it usually meant lots and lots of cables as well—dozens of yards of fancy wire to connect all the speakers, plus 11 or so RCAs, plus source cabling, plus video cables, plus plus. Great news all around! Everybody makes a ton of money.
Except...well, one little thing. Space.
As in, most people don't have a dedicated space for a home theater setup. And the idea of 6, 8, 13, or more speakers in the living room isn't usually the best idea for domestic bliss. (And, today, with more and more people living in close quarters, especially people just starting out in their careers, well, the downstairs and next-door neighbors may have the ultimate veto power on the idea of 1000+ watts and 15-inch subwoofers pounding away late at night.)
So, home theater was not exactly the home run everyone hoped it would be. Most people seem content to listen to their latest flatscreen, or (at best) add a sound bar and sub and call it a day. Home theater is a dedicated niche. It may be a relatively stable niche, and it might be a niche where aficionados spend lots on their dedicated systems, but it's a relatively small niche.
Of course, now there are schemes to do 7.1 or 7.2 or 9.2 or 11.4 or 16 or 32 channels (not kidding). Together with 5.1, these schemes have come to be known in the industry as "multichannel." Concurrently with this, stereo has come to be known (at least in the US) as "2-channel."
So, multichannel and 2-channel. Does that make more sense now? Maybe not. But let's stick to them, since they've been defined.
And let's add two more terms: desktop audio and personal audio. These are where Schiit debuted, and they're the industry categories we played in exclusively—at least for a short time. We actually have had a pretty good presence in 2-channel audio since the first Bifrost, and we expanded the reach even further with the Ragnarok and the Multibit DACs.
But we've never really had products dedicated to the 2-channel market.
Until now.
What is a “2-Channel” Component?
"So what separates dedicated 2-channel products from desktop products?" Someone might ask. "A lot of your products already have preamp outputs, Ragnarok can also drive speakers, and you already said that plenty of your DACs are used in 2-channel systems."
Yep. True. But, to address each in order:
Preamp outputs are not full preamp functionality. Yes, many of our headphone amps also have preamp outputs. This makes sense, since a lot of people use them to run powered monitors on their desktop, or to inject some tube flavor into their solid-state speaker-dedicated systems. But these aren't true preamps. They're missing a bunch of things, most notably:
- Multiple input handling. Most of our headphone amps are single-input devices. Preamps usually accommodate multiple inputs. And by "multiple," we mean "more than the 2 of Mjolnir 2." Yeah, Ragnarok qualifies, but that's really not a preamp—that's an integrated amp. More on that below.
- A focus on super low noise and other typical preamp concerns. Although most of our amps are quiet enough to be preamps, they weren't designed from the start for this duty. Some come with significant ease of use restrictions, such as Valhalla 2—which requires a specific start-up/shut-down sequence when used as a preamp, and the large-value output caps (necessary for headphone amp use) may cause DC-sensing circuitry on the input of some amplifiers to trigger, even though there's no actual DC on the line.
- Remote control. And this is the killer. None of our headphone amps have remote control, because, well, they're supposed to be sitting on your desk, or near at hand (headphone cables are, in fact, not infinite in length). And Ragnarok doesn't have a remote, because it was conceptualized and designed too early in our thought process regarding 2-channel gear. 2-channel stuff (where you're sitting back in your easy chair, looking at speakers across the room) really benefits from a remote control.
Ragnarok is hardly a desktop amplifier. If you're using it on your desk, you have a huge desk—or you have much greater patience than I do about how much crap you'll tolerate on your desk. It's really an old-school integrated, designed to be put in a rack. Yes, it's a headphone amplifier. And yes, it's a speaker amplifier. And yes, it really should have a remote. We'll get to that. Eventually.
DACs are just the beginning—lots of people like one-manufacturer stacks. Sure, we sell lots of DACs into 2-channel systems, especially Multibit DACs. This makes total sense, since the superior imaging of the "megacomboburrito" filter is much more apparent on speakers.
Aside: haven't tried your Multibit DAC on speakers? Have speakers? Do it. Trust me.
But, back to the subhead. DACs really are just the beginning. There are plenty of people who want to create a system from a single manufacturer's products. The rationale is, "Well, they designed them together, they probably go together better than a cobbled-together system." Of course, an equal number of people want to pick the best from each manufacturer and assemble a unique system—so different strokes, and all that. But the reality is: if we did more than DACs, we would sell a lot more than we do now.
There's a giant missing component: speaker power amplifiers. This is our elephant in the room. We're not going to go very far in 2-channel if we don't have power amps that can run speakers. Yes, we have Ragnarok, but a 60WPC amp is not exactly going to light the world on fire. We need three-digit-into-8-ohms-per-channel kinda amps.
But that's a subject for another chapter. We're here to talk about preamps.
And the more I looked at the preamp market alone, the more I realized that there was at least as much room for disruption as on the desktop—maybe even more...
Early Thoughts On Saga and Freya
Preamps are pretty open-ended. They can be all-singing, all-dancing—with tons of inputs, built-in phono preamp sections, hell, with digital inputs and outputs even. They could transform all analog inputs to digital. They could handle all analog and digital separately. They could do just digital. They could do just analog. They could have a line preamp stage with gain. They could have just a buffer with no gain. They could be completely passive. They could be solid-state. They could be tube. They could be balanced. They could be single-ended. They could be remote controlled. Or they could eschew remote (dumb, but yeah, well, they could.)
So, the logical question: where the hell do you start?
I guess a more logical and methodical person would start by looking at the market, doing a survey of what's out there, compiling a list of features and price points, and using that to create a matrix of threats and opportunities.
Of course, we didn't do that.
In fact, what I more than likely did (I honestly don't remember, which lends credence to this guess) is start drinking and thinking about what would make a good preamp.
Now, I wasn't working entirely in a vacuum. In fact, one member here on head-fi made a very powerful case that doing a passive remote preamp would be a great step for us, based on the very limited options available in the market. Another data point was the very strong sales of the SYS, a product I did mainly on a lark.
So, "passive" was on my mind from the start.
Aside: "passive" refers to a preamp without a gain stage, or (to be blunt), a preamp without an amp. Yes, I know, the terminology makes no sense. It would be more accurate to call it a variable attenuator. And some companies do. And they probably confuse the hell out of many prospective customers. Hence, we'll keep calling it a "passive preamp."
Also, remote control was at the top of the list. I've had many 2-channel systems, and the ones with remotes are by far better than the ones without. Call me lazy, but jumping up and down a half-dozen times to adjust volume—then doing it again on a different recording with a different average level—is not my idea of a good time.
So, passive, remote. What else?
Well, I was able to strike a couple of ideas off the "big list" right away—digital (as in DAC, ADC, or both functions), and phono preamp.
"Whaaaa?" Some of you cry. "But I'd love to have a product that would handle both analog and digital, and man oh man, there's a lot of turntables out there these days."
Yeah. And we have DACs for digital reproduction. And adding both ADC and DAC functionality to a preamp means it isn't really a preamp anymore—it's a mixed-signal processor. Or at least that's how I see it. If you transform everything over to the digital domain for processing, it's fundamentally changing the analog inputs. If you handle both separately, wow, you're talking a big, complex product that's entirely full of digital noise, even if you do it right.
So digital? No. Sorry. Not even as a card like in Jotunheim. Jotunheim is designed to be a compact, do-all product. The preamps I saw more as thoroughbred, end-game stuff.
And phono? Yes, there are tons of turntables out there. And we already have the Mani phono preamp. And if we wanted to do an even better phono preamp, it's better to devote an entire, much larger box to that functionality. One box, one function gives you a better chance that that single function is well-served. Or at least that's how we see it.
But even with those two big decisions made, we had lots of choices to make:
- Passive or active? Passive works for a whole lot of systems, especially ones with short cable runs and high-gain amplifiers. But there are other systems that need gain...especially if you're an aficionado of music that's recorded without excessive compression, like many classical titles. And if your cables are long (like, say, to monoblock amplifiers), you may need an active gain stage to drive them. It's really best to have both, if you can—but that's a really bizarre design decision.
- Balanced or single-ended. Argh. Balanced means a lot more parts, more expensive connectors, a larger chassis (the connectors are large.) But balanced works well with our products that have balanced output (which are usually larger products anyway). Single-ended can be simpler, smaller, and a lot less expensive. Again, it's really best to have both, if you can.
- Tube or solid-state? Again, tube is great for most systems. But some people like the certainty of solid-state. And both can sound very good. This is another hard call, and one that is much more a judgement call than passive or active.
- Motorized pot or relay attenuator? The most important function of a preamp is volume control. Remote volume control can be done in three ways: with a motorized potentiometer—that is, a motor that actually physically spins a volume pot, or with a relay-switched stepped attenuator, like we did in Ragnarok, or with a volume control chip. I won't even discuss the third option, though it's been showing up in many previously "high end" brands. Sorry, that's simply lazy. And when it comes down between a motorized pot—and all of their associated vagaries, like poor tracking at low volumes and eventual wear—and the relay-switched stepped attenuator, which has near-perfect channel tracking at all volume levels, relays rated for 15 million cycles, and nothing more than a couple of resistors in the signal path, well, it was really no contest. At least one decision was easy.
Since you're reading this chapter after the introduction of Saga and Freya, you already know the punchline. Instead of making the three hard decisions above, we decided, mainly, to do it all:
- We did passive and active, in both products. Because, as explained above, different approaches work better in different systems. I expect many, many people will be perfectly happy with the passive approach, but if you need an active stage, you need an active stage.
- We did balanced and single-ended, in two different products. The hyper-affordable Saga, and the still-way-cheap Freya. Of course, Freya does both balanced and single-ended—and, with tube gain, it converts everything to balanced output.
- We did tube and solid-state, in both products. In Saga, you can switch between passive and a hybrid tube buffer stage. In Freya, you can choose passive, a JFET buffer, or a balanced tube gain stage.
Yeah, I know. Sounds obvious, right?
But it really wasn't so obvious in the beginning. Hell, I had a completely solid-state Saga laid out before scrapping it and deciding to commit to a tube stage. And even that went through several revisions before I found a tube hybrid output stage that I felt good enough about to make into a prototype.
Freya took a bit of agonizing, too—I thought very seriously about making that a tube hybrid as well, so we could use just two tubes (instead of 4). The all-tube stage won in the end, mainly because Freya already has JFET solid-state outputs.
"Why tubes?" Some of you are asking. "I want solid-state!"
Yeah, and that's why there's the JFET buffer on Freya. Or wait a bit, and I’m sure we’ll have octal LISST available eventually.
"But seriously, two tube preamps?"
Well, I just explained how you can run them solid-state, and passive, but here's a longer (and more mercenary) answer: because there are a helluva lot of good affordable solid-state amps out there, but not so many affordable tube preamps. And when you start saying, "affordable, remote-control tube preamps," and "affordable, remote tube preamps from a credible US manufacturer," and, dare we say, "an insanely low-priced preamp, period, even if it was just a remote attenuator," well, the playing field gets down to a field of two very, very quickly.
No. Seriously. Think about all the solid-state amps out there from companies like Emotiva and Outlaw. Think about even more that sell for a pricing tier just higher than those two companies. Then ask yourself how many tube preamps these guys make. Like, zero. I think you'll quickly come to the conclusion that I did...that the market to sell a tube preamp to these possibly "tube-curious" audiences is much, much higher than a solid-state preamp.
Yeah. There you go. Call me a businessman. That's OK.
"Okay, then, why these tubes? Why not use the same ones as in Lyr 2, Mjolnir 2, and Vali 2?"
Aha, now that's a better question, and one that deserves a serious answer. It's because we hate you and want you to buy a whole new series of tubes, because we recently acquired all manufacturers of 6SN7-type tubes.
Kidding, of course. Dang, you're so serious.
Here's the real reason: because they're better tubes. 6SN7-style tubes are by far the most linear tubes I've ever worked with. At typical line-level output, Saga measures almost like an op-amp based design, with triple-zero distortion figures (as in, 0.000X%). That is 100% insane.
Aside: and yeah, we'll have the usual nonsense about how "this ain't really a tube amp, get a tube amp with tons of distortion so you gets the reeeel tube sounds." That's fine. You want a soft, fuzzy, blurry tube preamp, there are plenty of them out there that provide loads of second-harmonic distortion. The reality is that Saga and Freya, while the distortion is small, are true tube preamps, with typical tube harmonic structure (mostly 2nd harmonic, decreasing rapidly into the noise on higher-order distortion). They're tube preamps—they're just very clean tube preamps. And Saga is even cleaner than Freya.
So yes, we went with these tubes simply because they perform better. Plus the fact that we can treat them right—with 300V on them in Freya, and 200V in Saga—these are real tube preamps, not plate-starved designs.
In short, we used these tubes because they were right for no-compromises preamp design. Please accept my apologies if this kicks off a new tube acquisition spree for you. The good news is that the rolling options are rather more limited (6SN7 and 6SL7 types only, including the Russian 6N8C and 6N9C). Best to stick to 6SN7 only for Saga, unless you want more distortion—the hybrid part of its tube output stage is transconductance-matched to the 6SN7. 6SL7s will work fine, but they'll have measurably more distortion.
The Final Decisions, and First Prototypes
Okay, this chapter is getting ridiculously long, but let's talk about the tech decisions a bit more, the first prototypes, remotes, light pipes, and where we went from there.
For both Saga and Freya:
I mentioned that we were set on using relay-switched stepped attenuation from the start. I didn't mention that we decided from the start to do it just like Ragnarok—that is, with a microprocessor sensing a DC value on a real potentiometer, so the volume knob would operate exactly like a "real" one, with stops at the top and bottom of the range. This eliminates the need for LED light bars, light circles, screens, or other devices to indicate the volume level. It also impacts the remote. More on this later. We also decided to give Freya 128 volume steps, double that of Ragnarok and Saga—and 0.5dB steps, for finer control.
- The remote...we played with the idea of using pretty much every remote technology out there, before coming back to one of the oldest and most boring:
- Ultrasonic. Yes, just like grandma's old TV. Huge advantage: it's completely nondirectional. Huge disadvantage: it's pretty much a from-scratch design, including the remote handset. Big pain in the rear end, that—and costly. Also unknown reliability. We didn't go far with that.
- Bluetooth. As in, "hey, cool, you can use your phone and an app to control your Schiit!" On more research, also as in, "Hell, I don't want to find my damn phone to use the remote, give me a separate handheld instead." Yes. Even amongst our youngest customers—the vast majority hated the idea. Also kinda costly, and it would require us to produce and maintain apps. We went pretty far with this.
- Proprietary RF. Like ultrasonic and Bluetooth, this is also a nondirectional technology. However, we've had experience with very flaky proprietary remotes in the past, and, well, remotes most importantly have to WORK.
- Infrared. Yep, the old standby, same thing you use for your flatscreen and receiver and such. That's what we did. It's cheap, it's easy, and it's pretty much painless.
- The remote, part 2. Yeah, it's a cheap plastic credit card remote. Consider how much the products cost. Would you like to add $100 to the cost of a Saga for a machined aluminum remote? Seriously, if enough of you want this, I'm sure we can make it happen. But it simply wasn't a priority. I've seen machined remotes that probably cost more to make than Gungnir. These aren't those kind of products.
- The remote, part 3. I've mentioned before that the Ragnarok didn't get a remote because we were unsure about how to manage its behavior. When you use the remote, the volume knob won't turn. Which means your system can end up with a different volume level than the knob indicates. This could possibly lead to some problems if you were careless about it. The way we addressed this on Saga and Freya is to have a clear indication when the remote is in control—and software to force you to turn down the volume below the remote level before the knob takes control again.
- The chassis. Yes, these are the new-style, Jotunheim-esque chassis. There's no need for a lot of heat dissipation (well, aside from the tubes, which hang outside the chassis), so there's no need to use the chassis as a heat sink. This saves our butt in assembly. See next point.
- The LEDs. Even though Saga and Freya are minimalistic devices, they have a lot of LEDs on the front of them. So many that we thought about doing custom light pipes (molded plastic tubes to channel the light from surface-mount LEDs to the front panel). In the end, we didn't, mainly because light pipes have their own vagaries:
- They can't be in groups of 5 or more—they have to be singles. This means they are a pain in the rear end to put into the board, and result in very little or no labor savings.
- This means that they're never going to align perfectly with the front panel, even if we heat-stake them to the board, even if the front panel is machined to have "guiding cones" to get them in the right place.
- The per-piece price was still relatively high, even at 100000 pcs.
- The Jotunheim-style chassis allows us to align the individual LEDs from the underside of the product. This is the main reason. If it was a U-bend, like the Asgard 2/Valhalla 2/Lyr 2, we'd be toast—there would be no good way to align the LEDs that are placed near the center of the chassis.
Cool. So how about Saga alone?
I mentioned Saga's hybrid tube buffer stage, but I didn't go into any real technical particulars. Suffice to say, it's a weird choice. It stacks the tube on top of a PNP transistor. The main reason I did it was to reduce output impedance even further...resulting in Saga's about 170 ohm output impedance. It also reduces distortion. And, in listening tests, I think it sounds better than a simple cathode follower. But a lot of people have told me I'm crazy, too.
And Freya?
I mentioned Freya's JFET buffer, but I also didn't give any real details on that. Bottom line: it's pretty much the same as the buffer we use in Yggdrasil. It's interesting to switch it in and out of the circuit in real time, while watching the display on the Stanford. Interesting in that it's essentially impossible to tell when it's switched in. Except for a 1-2dB increase in noise floor, the measurements are exactly the same at line level. The JFET buffer far, far outperforms the Stanford's (very) low-distortion signal source, in terms of THD. The only peaks in the measurement are from the Stanford.
I mentioned Freya's differential tube stage, but again, I didn't give any real details. Let me correct that. First, by "differential," we mean "differential." This is a real, fully balanced, differential tube stage with tube buffered output. No solid-state in sight. No balanced-to-single-ended conversion. The relay attenuator is fully balanced as well. With a rail voltage of 300V, this differential stage has insane headroom for a line-level (2-4V RMS) device.
And in case I missed it, fully balanced and differential all the way through. Run a balanced input into Freya, and it's never converted from balanced and back again. Run a single-ended input into Freya, and it's automatically converted to balanced by the inherent properties of the differential tube gain stage. Use the JFET buffer or passive options, though, and it remains single-ended.
All of these decisions were in place when we went to do the first prototypes, late in 2015. Unlike many of our products, Saga and Freya stayed pretty much exactly the same throughout development.
As usual, when the prototypes showed up, I put them together, fired them up and made sure they worked DC-wise...and that's where the problems started.
The problems were in the transformers—all three of them. Saga's transformer was monumentally screwed up—producing only 18V on the high-voltage rail, when it should be more than 200V before regulation. I looked back at the specs, thinking I'd sent the transformer manufacturer the specs for the solid-state version of Saga that never was. Nope. It seemed right. There was no reason for it. I contacted the manufacturer and they agreed to do another prototype.
Freya? One transformer was screwy, producing only 1/2 the output voltage it should (this one was for the JFET buffer stage—it was running like +/-10V rather than +/-20.) Again, no explanation on this. Maybe the transformer guys were just having an off week.
The other Freya transformer was my fault. I'd under-spec'd the tube heater rail current (by about 1/2), which was pulling down both the heater and the high-voltage rail. It's kinda hard to regulate to 300V when you're starting with 260V, you know.
I also screwed up the transformer pin outs—a classic Jason screwup. Sometimes it's because I do the board before I get the transformers, and sometimes it's because they end up changing the transformers, and sometimes I'm just an idiot. Just like I get power switches backwards from time to time (that's just lazy, not checking my work.)
In any case, the Freya was limping, at least, (and it had a 5V rail that worked fine), so I handed it off to Dave to do firmware. While he was still working on it, I got a replacement transformer for the Saga that appeared to work fine, dropped it in, and gave him that one to work on, too.
A couple weeks later, Dave came back with some hilariously copper-taped and big-flying-component versions of the preamps. I took one look and laughed.
"I screwed up the power dissipation, huh?" I asked him.
"Just a little bit," Dave agreed.
The problem turned out to be mainly in the heater supply. I'd planned on having regulated DC heaters. That's great, it's what we do in Lyr 2 and Mjolnir 2. But the plan starts falling apart when you're talking 2.4 amps of current in the Freya. Even the 600mA in Saga was tough on the transformer.
"Go to AC heaters?" I asked, doubtfully. AC heaters are fine, but, if not laid out well, could hum.
"I never had a problem with hum on AC heaters with 6SN7s," Mike said.
"But these are preamps," I protested.
"Line preamps," Mike sniffed. "Not phono."
"Build it, see if it works," Dave said, ever the pragmatist.
"Okay," I said. "Then I'll get new transformers, and you can get the software done."
Dave blinked, surprised. "The firmware is done."
"What?" I asked. The boards looked so rough it didn't seem they could possibly be done.
"They work," Dave said.
"They work? As in work?"
"Right. The transformers just get realllllllly hot."
Dave was right. The high-voltage transformer in Freya would literally double as a hot plate after a few minutes. But it worked. The inputs switched. The volume volumed.
"All I need is the remote," Dave said. I hadn't received the sample remotes for approval yet.
Aside: we handed off the production of the remote to a turn-key remote manufacturer—we just specified what type of remote, what codes, and what buttons. If we'd done it ourselves, we'd have risked getting tied up in the design and production of it. This way, it was a really, really easy handoff.
(And, like I said, if you want some fancy milled machined aluminum remote, tell us. The more who want it, the higher probability we'll do it. But we weren't going to do it to start.)
"Do you think it'll just work?" I asked Dave, miming the remote. I was still kinda floored by having two working (well, limping) preamps, just a couple weeks after handing them off for firmware.
Dave shrugged. "IR is easy. It pretty much always works."
"If you say so."
Maybe I've been through too many painful product development cycles, but a little voice kept whispering: It can't be this easy.
Well, surprise surprise: it pretty much was that easy.
I tried Saga and Freya with AC heaters, and they worked just fine—the hum measured no differently than the DC versions, and it was well below what it needed to be for a line preamp. Just as Mike had predicted.
Board layout is critical with AC heaters, though, so I did another set of prototypes—now with new transformers designed for AC heaters. They measured the same, just fine. Though the voltages were still a bit low, argh. So we went back for another round of transformers.
Aside: why so many problems on these transformers? No clue. The manufacturer seems to have internalized (finally) our directive to "make sure the damn things don't hum," but now it seems like they may be sacrificing performance to make sure there's no hum. Ah well, as long as it works in the end.
And—here's the capper—when the remote control samples came in, I handed one to Dave...and he handed me back two preamps a week later, both happily working with the remote.
Like. Friggin. Magic.
Yeah. Almost like we're getting good at this (but as soon as we think this, well, that way lies ruin. Best to expect more problems than you have. There is no power of positive thinking in this one.)
A Seismic Shake-Up in Value?
So what are we left with here? Two preamps, both remote-controlled, both with relay-switched stepped attenuators, both of which can be run passive, both of which ship with tube or tubes, both of which can be run completely solid-state.
And (drum roll)...one that costs $349, and one that costs $699.
If this doesn't sound like anything else out there—if this sounds insanely, crazily, stoopidly cheep, well, that's because it is.
And it's that way because that's what we do. We design something, we figure out how much it costs is to make, and then we apply our standard margin to it. We don't price things "at what the market will bear," or "higher so people will take it seriously," or "higher because there's literally nothing like it, anywhere near the price."
Call us old.
Call us crazy.
Call us reactionary.
But that's how we think it should be.
We think these are absolute, barking mad deals. Even more now that we've emerged from the design phase and spent some time looking around at the market. Hell, there are passive stepped attenuators (only, not with tube stages or remotes) that cost a lot more! When you start talking 6SN7-based preamps from credible companies, you start adding zeros really, really fast. And when you start talking about relay-switched stepped attenuators, you are talking really, really rarefied air. That's completely beyond-the-pale cost-no-object high-end design. Hell, it's beyond a lot of "high end" companies that have given up and simply use volume control chips now (barfarolaomaticdeluxe, is how I feel about that, but I'm biased.)
So, do I sound proud of our new preamps? It's because I am. I'm thrilled to be bringing some real value back into high-end 2-channel audio. I'm pleased that the price is so shockingly low (I had to run the numbers twice myself to believe it.)
But in the end, it's up to you.
