2017, Chapter 13
Sailing Off the Edge of the World
- When a Leading Manufacturer said, “flat is pure,” they willed the entire high end audio world to be one of flatness.
- This word cannot be understood as referring to the purity of sound, that is, listening and satisfaction, as reproduced by an audio system.
- Yet it does not mean solely purity; such flatness is worthless unless it produces various outward titillation of the eardrums.
- The penalty of flatness remains as long as there is lack of control of tone (that is, equalization), namely till our acceptance of tone controls.
- The Audio Press neither desires nor is able to remit any control of tone except those imposed by the Audio Engineer or the Artist.
- The Audio Press cannot remit any control of tone, except by declaring and showing that it has been remitted by the Audio System; or, to be sure, by remitting guilt in cases reserved to its judgment. If its right to grant remission in these cases were disregarded, the guilt would certainly remain unforgiven.
- The Master Reviewer remits control of tone to no one unless at the same time he humbles him in all things and makes him submissive to his Audio System.
- The penitence is imposed only on those lacking control of tone, and nothing should be imposed on the users of equalization.
- Therefore the Master Reviewer through the Audio Press is kind to us insofar as the Audio Press in his decrees always makes exception of audio impurity.
- Those Fanboys act ignorantly and wickedly who, in the case of the impure, reserve penalties of withholding tone control.
Okay. For those of you who think I’ve gone totally barking mad, look up Martin Luther’s 95 Theses, and compare the first 10 to my hack-job.
So, you think I’m still completely bonkers. Cool. Skip this chapter. Because we’re sailing right off the edge of the world here.
Until now, you may have lived in a safe and familiar place, where The Original Recording was perfect and sacrosanct. You have been taught to judge your system on how little it degraded the Sacred and Holy Original Recording. You may have spent thousands of hours and thousands of dollars trying to recreate the Original, rather than sully its purity.
And, to be fair, we lived in the same world. None of our products allowed even the most basic of signal-altering functions—no bass, no treble, no crossfeed, no DSP, no room correction, nothing to change the Sacred Original. Because that’s just Not What You Did in the Hallowed Halls of High End.
Well, at least not since the early 1980s, when manufacturers decided that tone controls were an affront to The Sacred Original Recording’s purity, and removed them from their gear.
The press gleefully reported this new development, again and again, without any push-back from the wild world of online forums and social media, because, hey, this was a simpler time and things like that didn’t exist. The 17 audiophiles on Compuserve might have sighed resignedly over the loss of tone controls, but there weren’t enough of them to make a stink, and they simply weren’t driving the media bus.
So, over time, more manufacturers fell into line (gleefully rubbing their hands together, thinking, “leaving these damn tone controls out will save a lot of money and time, heh heh”), and the lack of tone controls became dogma.
And that dogma was such that, when people asked me if we were going to have tone controls on Saga and Freya, I responded, indignantly, “Of course not!”
There was no way we would sully the Sacred Original Recording.
No way.
It just wasn’t done.
Everybody knew that.
Except, yeah, tone control was nice when the Sacred Original wasn’t, er, ahem,
perfect. Because, ah, yeah, there are plenty of imperfect recordings out there.
And, except, yeah, tone control could be nice if it had enough bands (or well-chosen bands) so it could compensate for an imperfect transducer. Because, ahem, again, embarrassingly, not all transducers were perfect, either.
And, well, when you got down to it, tone control could also help with room problems. Because even with a great recording and great transducers, rooms sometimes got in the way of reproducing the Sacred and Original Recording. Especially when those rooms collided with real life—real lives that included screens and comfy furniture and shelves with important mementos and family knick-nacks and stuff that you couldn’t just declare, “Dear, I’m gonna throw alla this stuff out, okay, and add some bass traps and reflection damping.”
Yeah.
And, most of all, when something becomes dogma, it
needs to be questioned.
And when that dogma creates a monoculture, it begs to be blown apart.
That’s my long-winded way of saying, “Welcome to Loki Mini, our first equalizer.” And by “equalizer,” we mean “tone control.” As in “knobs you turn to change the sound.” As in, yes, we are
most definitely sullying the Sacred and Holy Original Recording.
Call us heretics. It’s OK.
Why Equalizers Went Away, the Full Story
I’m oversimplifying when I say that EQ was killed by lazy manufacturers and a conspiratorial press. Sure, that was part of it, but the reality is that EQ sucked ass in many ways, and, in many ways, deserved to die.
If you are of a certain age, you remember the graphic equalizer. A veritable sea of cheap open-frame sliders, begging to be tweaked into a rictus grin. No matter how many channels—5, 10, 31—they always seemed to be set into the same basic “smile.” Boost the bass—get those Cerwin-Vegas pumpin—and then raise the highs. Midrange? Who needs that? Boom and tizz is where it’s at.
Boom and tizz…and noise.
If your equalizer had a real bypass switch, you may remember the rush of hiss and hum that hit you whenever you engaged the EQ. Shhhhhhhhhhhhhhhhh…but when the music started, who cared? Hoom! Hoom! Hoom! Sizzle crackle blert hoom!
Yeah. Running the Sacred Original Recording through 5, 10, or 31 LM1458 op-amps will do stuff like that. And no, I’m not exaggerating. Yes, they were absolutely using LM741-quality op-amps in lots and lots and lots and lots of the equalizers out there. These are about as far away from a modern audio-quality op-amp as you can imagine. The LM741/1458 are the McDonald’s hamburger of op-amps. Lowest common denominator. Crappiest quality. Noisy, slow, and obsolete, they shouldn’t ever be allowed near audio—not even as a servo.
Add to the op-amps crappy components like cheap ceramic and electrolytic capacitors, lowest-common-denominator open-frame slider potentiometers (all the better for collecting dust and becoming a scratchy mess over time), and products cost-engineered to maximize the number of bands, rather than quality, and you can begin to get a better idea why equalizers disappeared from the audio scene.
Now, every once in a while, there were exceptions. Heck, in my car audio days, I absolutely loved a Spectron three-band EQ that was based on the (pricey, in that era) TL072 op-amp. It had enough bands, and enough range, to make a meaningful difference in sound, but wasn’t super-noisy (and it used sealed pots for use in a harsh car environment.) Of course, it wasn’t cheap—I think about $299 in 1982, or almost $800 today.
And, of course, there was the Cello Audio Palette. This heroic 6-band equalizer was 100% discrete, and used custom-built stepped switches to precisely control each band. Of course, it was also $25,000 in 1989—or, a bit over $50,000 today (!!!).
Still, you get the picture. When most of the EQ out there is literally destroying your music with noise and distortion, it might be best not to have an EQ.
That (and the manufacturers, and the press) is why EQ died.
EQ Done Right
When we were introducing Saga and Freya, after I’d stopped reflexively answering questions about tone control with “of course we ain’t gonna do that!”, I began asking sitting and wondering, thinking and drinking, and, eventually, asking myself heretical questions.
Questions like:
- Is it time for EQ to come back?
- If so, can it be done well?
- If we do it, does it invalidate everything we’ve done to date?
Let’s tackle each of these in turn.
Is it time for EQ to come back?
Some would say, “Hell yes! No high-end manufacturer is offering good, solid, well-engineered analog EQ on their gear. This is a monoculture. Monocultures are begging to be disrupted. The first in the disruption pool gets all the toys! Splash!”
But others would say, “Nah, no way. iTunes gives you a built-in digital EQ. I can download free digital EQ that allows me to tune out the specific frequency of a gnat fart in a 20,000 cubic-foot room. Hell, for a few bucks, I can get a calibrated microphone and do room correction. Analog EQ is for old farts with one foot in the grave.”
And still others would squint and frown and rub their chin, and say, “Hmm, the interfaces for digital EQ really suck balls, and room correction sometimes causes more problem than it fixes, and you know, just having a few knobs at hand is really, really convenient…maybe if it’s good, and doesn’t cost a fortune, analog EQ might be worth it.”
So yeah, we considered all these viewpoints carefully for 50 milliseconds, before we skipped to making prototypes designed to answer the second question:
Can EQ be done well?
Sure it can. The Cello Audio Palette, from contemporary reviews, was great. You can almost always achieve great by throwing money at a problem. A better question would be, “Can EQ be done well, at reasonable cost…and what form would it end up being?”
The answer isn’t simple. There are many forms of equalization. Let’s go through just a few:
- Passive LC EQ. Yes, Virginia, you can do passive EQ using just inductors (L) and capacitors (C). Sounds pretty audiophile-y, right? But, passive EQ means insertion loss. “Insertion loss” is a fancy engineering term that means “the signal is quieter coming out than when it went in.” This means you need to have an amplifier to bring it back up to the original level. This means noise. If you lose 12dB due to insertion loss, then amplify back up to the original level, it’s a good assumption that you’ve just cut your signal-to-noise ratio by 12dB. (And for the armchair engineers out there, yeah, I know this isn’t a given, but I’m assuming typical use.)
- Nonconstant-Q Active EQ. This one is more interesting. Like the passive EQ example above, you can use LC filtering and a single gain stage for multiple bands (well, to be more technically correct, a single gain stage and buffer) to create an effective equalizer. The single gain stage and buffer can be literally any topology imaginable—op-amp based, fully discrete, current-feedback, tube, Class A, whatever. Sounds pretty high-end-ish, doesn’t it? It is, and if you do a good job on the gain stage, it can have exceptional noise performance, unlike the passive example above. The one oddity (or feature, depending on your POV) is that each band of equalization won’t have a constant “Q”. This is a fancy engineering term that means the band starts out broad when you first adjust it, then becomes more narrow as you continue adjusting. We like this approach, because it allows for small, broad adjustments (think tonal tilt) at low levels, then transitions to address peaks. We also like it, because it needs only a single gain stage and buffer, not racks of gain, so that means a discrete design is feasible. If the bands are chosen well, this can be a great EQ. That’s why we chose this topology for Loki Mini.
- Constant Q Active EQ. You can also design the equalizer to have a constant Q, no matter how much adjustment you’re doing. This is usually how the 5-band, 10-band, and (LOL, no Schiit) 31-band EQs of the 1980s were done. This requires a gain stage for each band. So, if you have a 5-band EQ, you need 5 gain stages. For a product like Loki Mini, this would mean op-amps, because there ain’t no way we’d fit 5 discrete gain stages—per channel—on it. The constant Q feature, we felt, was of marginal utility.
- Parametric EQ. This is the full-boat crazy EQ. Parametric EQ allows you to select the center frequency, and (usually) the Q, or the “peakiness” of the adjustment. If you want to adjust just a tiny sliver of frequencies, or a broad selection, parametric EQ gives you this kind of control. This can be immensely helpful to address specific peaks or valleys in frequency response. Does the transducer have a peak at 8.7kHz? No problem, adjust the EQ to that point, then vary the Q until it precisely addresses the problem. Sounds great…until you’ve used one, and shook your head at the fact that things still sound a little off, even when they should be perfect. Oh, and parametric EQ? Think LOTS of gain stages. So discrete is out, unless you want to end up with something that costs like an Audio Palette. Digital parametric EQ definitely has the advantage over analog parametric EQ in this respect.
Aside: my deepest engineering education is in analog control systems. I’m very comfortable in the s- and jw-domain, and I know what’s up with Bessel, Linkwitz-Riley, Sallen-Key, and more. Maybe it was inevitable I’d end up doing an EQ.
So, yeah. EQ can be done well. It just takes a nonstandard approach, and you can easily go with inductor-capacitor (LC) filtering, a single discrete buffer, and a single current-feedback gain stage, and get amazing results.*
*Well, within reason. LC filtering for the bass ranges means big, big inductors—as in, 1 Henry or even larger. Engineers are cringing right now. For the rest of you, that means “a big hunk of metal and lots and lots of turns of wire—stuff that won’t fit into a Modi-sized chassis.” You’re looking at—at minimum—a 1.5 x 1.5 x 1.5” square, with 100 ohms or so of DC resistance…and if you want it to be linear, to keep your THD down, you’re also looking at an exotic core, like an 80% nickel version. That’s why Loki Mini uses a gyrator (a simulated inductor) for the lowest-frequency band—an actual inductor would be too large to fit into the chassis, and would significantly impact cost, even if it did fit in the chassis.
But more on this later. Let’s talk through the last question.
If we do EQ, does it invalidate everything we’ve done?
Although I want to flippantly say, “Well, if you don’t like it, don’t buy it, problem solved!” this question deserves a serious answer. And here it is:
If you don’t like the idea of sullying the Sacred Original Recording, don’t get a Loki. Problem solved.
Okay. Let’s be serious. Because we have devoted quite a few words to the preservation of the original signal. Our multibit DACs are based on trying to deliver the best recreation of the original samples. All of our gear is flat out to far beyond the ranges of human hearing. And our preamps don’t have tone controls.
And now, we’re saying it’s OK to change things?
No. We’re saying
it always has been.
But only if you
want to. And that’s what this is all about. This is about choice. You can
choose to go on as you have, and ignore the fact that Loki exists. You can preserve the original samples, faithfully play back your music on systems with no tone controls, and search for the most perfect transducer (or, most perfect transducer and room) without any, ahem, “artificial aids.”
Aside: well, maybe besides those crazy cables and colored tube amp and that old CD player with rolled-off highs you love so much. The fact is, many audiophiles have been pursuing ways of altering their system’s sound, ways that just don’t have electronics inside and knobs on the outside.
Or you can
choose to change things. You can use devices like Loki to bring your overall system more in-line with the Original Recording. Or you can tune it more to your individual preference. There’s nothing wrong with any of these approaches—leaving it pure, trying to make it more pure, or pursuing your own goals.
And—this is important—none of these approaches invalidate our pursuit of originality in the DACs or flat response in any other product. If you want to change things, you need a solid foundation to build from, or else you may end up in a state of constant change, churning and buying and swapping and tweaking to try to find an elusive Perfect Recreation of the Original—which, ironically, might not be so Perfect and Original, but just Perfect for You.
Accurate DACs and uncolored associated electronics provide a solid foundation for you to start from, if you’re seeking perfect reproduction. If that’s your end, that’s great. We’re not going to stop making the rest of the gear because we now offer Loki.
Aside: nor will we stop making the rest of the gear when The Gadget appears—a whole new class of audio adjustment. Yes, we’re going into full heresy mode, here.
But…again, I’ll repeat: it’s your
choice. We’re not here to try to take over an entire recording industry here. We’re not trying to make it so everything has to be processed through The Gadget or heard through a Loki. We don’t have DRM features like selective degradation, phone-homes, and device serialization/deactivation in Loki or The Gadget. Use them if you want, or ignore that they exist. The market will decide if they live or die.
We may be heretics, but we aren’t monsters.
Back to the Basics
“Okay, fine, you mentioned you did some prototypes,” the more technical-minded of you are saying. “But how did development go? Was it a nightmare? Loki looks more complicated than stuff you usually do.”
Actually, it was a really interesting development process, one that took us down many new avenues, and one that will pay off well beyond the Loki Mini. And I don’t mean “interesting” in the “wow, it sucked,” way. It was interesting because:
- We learned a whole lot about the real-world properties of inductors, a passive electronic component we don’t typically use many of.
- We also learned a lot about buffer design, and their interaction with complex loads.
- We were able to validate another use for our single-ended current-feedback topology, as used in Vidar and Magni 3.
The main thing we knew, when approaching the design of Loki, was that it was wayyyy different than the stuff we usually made. The goal of an amplifier or preamplifier is to be as accurate as possible—flat frequency response, low noise, low distortion. Loki had the same low noise and low distortion goals, but it was also designed specifically to alter the frequency response.
To do this, it needed a buffer that was linear enough to drive a reactive network that included potentiometers, resistors, capacitors, and inductors, in-between the buffer and the inverting and non-inverting terminals of the gain stage.
Lesson one: when you start connecting simple 2-FET buffers to these kind of networks, things get interesting in a real hurry.
And, in fact, many of our lessons started long before any PCB was ever made for Loki. Early prototypes used potentiometers standing in the air, wired in-between a small buffer board and a Magni 3 prototype. This was just so we could see how the distortion changed in real time, and what the effects of multiple bands were on the overall response, and on the accuracy. Unlike many of our projects, Loki started at the most basic level—the “will it work, and, if it does, what works best.”
In this early stage, we tried a whole lot of different things:
- Different gain stages (like ones based on the old Magni, ones based on op-amps, ones based on single transistors, before we finally settled on Magni 3’s basic topology as the best-performing combo.
- Different buffers. Hell, I went through a dozen different buffer designs. Dave threw some in the mix as well. Some were completely insane—like the 7-transistor “final boss” load-invariant buffer that was squirreley as hell, and therefore didn’t make it to the end of the road, but performed insanely well when it worked. Think -120dB on 2nd and 3rd harmonic—unloaded, or when loaded with 600 ohms. This is better than the signal generator on our Stanford analyzer. What we ended up with for Loki Mini was an optimized, stacked FET buffer that gives great performance, but is much less insane, and much less complex.
- Different inductors. The performance of many small inductors is (ahem) less than ideal. Start driving them and watch the distortion go nuts, even at the 6th, 7th, and 8th harmonics. Scary. We went through a lot of different options before selecting the ones used for Loki Mini.
- Different gyrators. Like I mentioned before, big inductors were off the table for Loki Mini—they simply wouldn’t fit. Because of this, we used a simulated inductor (using an op-amp and capacitors) for the lowest frequency band. But even that caused problems with distortion. Switching to a gyrator topology reduced the loading and eliminated the distortion problems. But again, it was something we had to learn.
- Different options. Three bands or four? Five? Six? What potentiometers? What chassis? Before we did anything to fit in a chassis, we did a proof of concept board about the size of a Jotunheim with 6 bands, and options for inductors or gyrators for the two lowest bands. It had no protection, no niceties—just something for us to tweak on. We destroyed a couple of these boards working through all the options.
And depending on the options we chose, we had very different products. Lose a couple of bands, do the lowest band with a gyrator, and go to smaller Alps pots, and we had something that would work just fine in a Modi-sized box. This could be a neat, inexpensive way to get a taste of EQ, without spending a ton of money—easily under $200.
Or, go pure LC and put it in a bigger box. Problem with that was that the cost ballooned. The early inductor prototypes were a mess. They ended up having to be re-wound on 80% nickel cores, and they cost more than some of our transformers. You also had to do extreme shielding on the transformer in the bigger box, to keep its field from being picked up by the inductors. That would be a lot more expensive product—maybe on the order of $500, even if we lost one of the bands to eliminate two very expensive inductors.
Or, you could go completely crazy and do a relay-switched stepped potentiometer version, with a true center and precise, repeatable control. You could even do balanced I/O with a Pivot Point style topology. You could do remote control. Heck, you could use motorized potentiometers just to sense the position, and have them all turn when you hit a preset. Of course, something like that would be big (at least the size of Freya) and costly (over $1,000.) Yes, at that point, it might end up being one of the best equalizers ever made, but that was a big ask.
And then, suddenly, I realized,
We don’t just have one product. We have a family of products.
That is, if we wanted to go that way. But if introducing one EQ was heresy, introducing three would be lunacy. It would be best to start small, and see if anyone was interested in the product. Hence, Loki Mini.
Aside: will there be other, bigger Lokis? Depends on how the market reacts. Judging from early adopters’ comments on the Loki Mini, I think it will do well. I am much more skeptical about a crazy remote-control, all-LC, balanced in/out device that costs more than $1K.
And, once we decided on starting small, development went relatively quickly. It’s a complicated board—there are lots of parts in a Loki Mini—but other than that, and other than the need for a potentiometer “girdle” that improves pot alignment with the chassis, it went relatively smoothly. Which you might have expected, given all the playing and tweaking we did up-front. Two revs on the PCB and we were ready to place an order. All in all, about a year of dev work. Pretty efficient, when you consider that Jotunheim and Ragnarok took multiple years.
Aside: the name. Yep. I’ll own it. We recycled the name. Quite on purpose, in fact. I always liked the Loki name, and I was sad when it died on the DSD add-on DAC. And Loki the trickster is perhaps the most perfect name, ever, for such a heretical product.
Loki: King?
Of course, there are a ton of you out there, crossing your arms and saying, “Yeah, yeah, yeah, typical Schiit hype. How can something that costs only $149 be good enough to use in a real high-end system?”
Well, at least to my ears, it’s
perfectly good enough. We included a real bypass switch on the front panel so you can A-B it instantly. When Loki’s out of the signal path, it’s as out as it can be—as in, the Precious Original only passes through a relay and a resistor. When it’s in and the controls are set flat, it’s also pretty damn flat—the custom Alps pots with a very precise center help with that.
I’ve used Loki to improve the response of headphones I don’t particularly like (sorry, no names here, but let’s just say that some people have been surprised when they’ve walked into my office to see me wearing headphones they thought I’d never tolerate.
I’ve also used Loki in speaker systems, most notably in our old listening area, which was, to put it mildly, quite suboptimal. It made the overall system much more listenable—and more accurate.
Aside: Loki Mini is great for relatively small adjustments. You need to turn the pot almost 30 degrees for the first 1dB of change, and, as noted before, it’s a broad change to start, then peaking at the band center. The degree of change accelerates as you go more off-center of the pot, topping at +/-12dB for the 20 and 8K bands, and +/-6dB for the 400 and 2K bands.
And, even more interestingly, almost everyone who’s tried a Loki Mini has ended up keeping it. A couple of our more “audiophile types” here even said to me, “Wow, it doesn’t screw up the music, I thought it would.”
Yes, I know, anecdotes are not data. But I do think that Loki Mini is a cool little device, and it offers insanely high performance, even well above its cost. I mean, take a look around for other all-discrete, LC-filtered equalizers out there. You probably won’t find many. Or any.
But in the end, it’s up to you. Do you want to stay in the land of audiophile dogma, or is it time to commit your own heresy?
