2023, Chapter 10:
Solidification
Freya S has been replaced by Kara, and it’s now a new world.
Hyperbole? Not really. Not when Kara is sooooooo much more insane than Freya S. I mean, it will swing nearly 100V on the balanced outputs.
“Whoa. Wait a sec. Why would you even want that?” someone asks.
Hold on. Let me keep going for a bit. It’s not hype to call it a whole new world when Kara is an entirely new preamp, with insane output capabilities, running on 64V rails, 2x the transformer size, dedicated single-ended stages for higher SE performance, measurements that rival IC designs with our discrete low-feedback Nexus™ gain stage, great sound, and a headphone jack.
“Wait wait wait!” you cry. “A headphone jack?”
Yes. And it’s not connected to some lame op-amp headphone amp, it uses the same discrete single-ended stage as Kara’s SE outputs, running on +/-32V rails, with a whole bunch of protection wrapped around it to make sure those insane voltages are totally safe.
In many ways, Kara is a return to the great solid-state preamps of the past. In many other ways, it’s a product of today.
But I’m getting ahead of myself. Let’s take a look back at preamps past.
The Golden Age of Solid State
A few decades ago, we’d reached a plateau of solid state performance. Engineers understood enough about solid state to avoid designs that sounded bad, devices designed for audio were readily available, and there were plenty of options for both amps and preamps for people who didn’t want to mess with tubes.
How good were these preamp designs? Some stunningly good. Check this out:
That’s a Sumo Athena from 1990. Just pulled off a shelf and tested, not refurbed or recapped or whatever. I’m sure the electrolytics are pretty dry, which is probably why the hum isn’t exactly state of the art. But you get the idea.
Now, for you are saying, “Scoff! That’s not state of the art!” I invite you to look at the gain.
As in, that preamp has
20dB of gain.
And it’s -110dB THD+N.
WITH 20dB GAIN.
From 1990.
And it was fully discrete.
Oh yeah and it was $699, if I remember right.
Now, the Athena from that time isn’t a modern product in a lot of ways. It was geared towards the systems of the time, so it had a built-in MM and MC phono preamp (switchable by removing the top cover!), a “CD Bypass” gain-of-1 function (with discrete buffers that measured worse than the line stage), and a focus on recording, with prominent knobs for both tape loop and input selection.
Sharp-eyed readers will see we cheated a bit on this measurement, since the noise floor ain’t so great on this 30+ year old product (dry caps, probably)—it’s A weighted, while the measurements for Kara that follow are unweighted.
But it did perform very well. It did this for several reasons, but most notably it had a good discrete topology, and it had +/-35V voltage rails.
Aside: Apply +/-35V rails to the popular high-measuring op-amps of today, and you get one result: boom. They can’t take it. Rail voltages above 15-18V are the province of discrete.
Why are such high rails important? Because they allow you a broader linear operating window. 2V RMS output, or +/-2.8V, is 9% of a +/-15V design’s rails, but only 4% of a +/-32V design. And the less the design swings relative to its rails, the more linear it can be. And the more linear it can be, the less need to rely on crutches like huge amounts of feedback. Now, we can go to low-loop-gain designs, and still have exceptional performance.
The new Kara is an homage to these golden-age analog designs, using modern devices, our exclusive Nexus differential, discrete, current-feedback, low-loop-gain topology…together with advanced microprocessor oversight and a relay stepped attenuator that even Athena’s Noble potentiometer can’t touch.
And what are the results? Like this:
Oh wait you want high gain? How about this:
Yeah. Pretty much state of the art. Without an op-amp (integrated or discrete) in sight.
Oh yeah, and Kara is $298 in 1990 bucks. ($699 today). With a relay attenuator and remote control that Athena could never hope to have. Oh yeah and balanced I/O that Athena didn't have. That's
real progress.
Freya S was a great-measuring preamp. Kara kills it. And, despite its measurements, it sounds very, very good. I’ve always used Freya S. Kara is now my preamp of choice.
“Wait a sec, you said you don’t talk about sound,” someone says. “Why are you talking about sound here?”
Simple: because there’s a perception that high-measuring devices can’t sound good.
That’s untrue. They can.
So we’re saying Kara is a high-measuring, good-sounding product. That’s all. We’re not claiming it sounds better than the Arglebargle Deglampulator III or the Terahaut Formaldehyde or the SAS RX-4040axII. Nor will we start speculating as to whether it’ll sound best in your system, or trying to sell you on its virtues. As always, it’s up to you. You choose.
The Path to Kara
For those who’re confused about this whole Kara/Freya S thing, let’s talk about how we got here. In short:
- Freya S was our first balanced solid-state preamp
- Kara is the replacement for Freya S; it’s still a balanced solid-state preamp
Freya S was introduced in 2019 as part of our first product “Thunderdome.” A “Thunderdome” is where we introduce multiple variations of a product to see what people actually want to buy. The idea is that we make more of what people like, and stop making the ones they don’t like. That kinda-sorta works; for a while we stopped making two of the Thunderdome products, but then people started asking for them in droves, so we started making them again. And the response on Magni and Heresy were such that they both stayed in the line. So maybe not a true Thunderdome. But hey, nothing’s perfect, and we ended up learning a lot.
The idea behind Freya S was simple: give people a solid-state preamp option in the same Freya form factor, while using as many common parts as we could.
It was also the first product that had our Nexus topology, a little earlier than Ragnarok 2. Not exactly to plan—Ragnarok 2 was supposed to be first. But Freya S was absolutely a highlight for Nexus. It sounded insanely good, and worked very well.
But, in sticking with common parts, we used one of the standard Freya+ transformers. This meant we were going to have +/-17V rails.
At the time, I didn’t think it was a big deal. The measured performance was very good, and it was a balanced preamp, so it could still swing crazy numbers at its output—like almost 50V. Plenty fine.
We also kept the same front panel layout, which was, well, OK, and the same bottom chassis, which really didn’t matter, and the same relay ladder, which was a great choice. The commonality paid off, too, because it meant Freya S could come in insanely cheap—$599 for a preamp of its capabilities is a screaming bargain.
But still, I thought of those old preamps of yore, with huge voltage rails. And I remembered that when I was working out the issues with Loki Max, which also used Nexus, and which actually needed larger rails to deal with the gain an EQ could put into the system.
And that’s where I finally did a design with +/-32V rails–and Nexus. And oh boy did it work well. And oh boy we finally had a transformer that could do what I needed to do. And oh boy that’s the day I started thinking of doing a “Freya S2,” using that transformer.
But I got into other projects, and, fortunately, learned a few more things. Like how to improve Nexus. Like how to better manage SE output. Like how much I hated the button placement on Freya (looks pretty, but is inconsistent on which buttons affect what LEDs…I use the preamp everyday and STILL get them wrong).
So I put together the first prototype of Freya S2.
At the time, it was still called “Freya S2.” It had the improved Nexus stage, high-volt rails, dedicated SE stage, and a bunch of other small tweaks, including button placement.
And it worked really well!
It worked so well that Cameron begged the first prototype off of me. Cameron has always used a Freya S as well, so this boded well for the design. A little tweaking, a couple of board fixes, and we’d be off to the races.
But of course I had to have a stupid idea.
You Ain’t Got Jack
Here’s the thing: you have idle thoughts. Or at least I do. The problem is, sometimes I act on them. And that’s how Freya S2 gained a headphone jack and became Kara.
Up to Freya S2, no Schiit preamp ever had a headphone jack. The reason, for Freya and Freya+, is simple: the tube stage of those preamps has exactly zero chance of powering a low-impedance headphone effectively. It’s simply not gonna happen. The reason, for Freya S, was also simple: its line-level-optimized Nexus stage would simply go boom if you plugged headphones into it. It didn’t have big enough output devices to survive.
But with Freya S2, I’d gone to new, much more advanced output devices, the same ones we use in Jotunheim. In addition to being more linear than the old outputs, they were also capable of sourcing significant current. Its gain stage was, in effect, closer to a headphone amp.
And a discrete headphone amp, running on +/-32V rails, should be super, super good.
(And, to be really honest, I wouldn’t have to stack a headphone amp on top of my preamp just for late-night listening, which is probably the real driver for this change.)
So, to test out this idea, I tacked a headphone jack onto the single-ended stage of Freya S2, and plugged in a headphone to test.
Sizzle.
Smoke came out. I’d killed the preamp.
Ah crap, I realized. The prototype had been playing at the time, and the jack had shorted the channels to ground when I plugged in the headphone. Even with fairly beefy output devices, the high rails meant it could get into real trouble real fast.
That almost stopped the project right there. Because, after all, we didn’t need a headphone jack. And it was only going to be an SE jack, just a ¼”, no XLR balanced output at all. The choice to forego balanced was deliberate: most people who aren’t big into headphones won’t have balanced headphones. This was a headphone jack on a preamp. If people wanted a bigger/better/balanced headphone amp, we made tons of them.
But instead of killing the headphone jack, I decided to take it as a challenge. And what came out of it is a fairly amazing bit of safety engineering.
Let’s discuss the problems with the TRS jack:
- It has a point, when plugging in, that both channels are shorted together. This isn’t too bad, because most music will be similar on left and right channels.
- But it also has a point, when plugging in, where a channel is shorted to ground. This is bad. Boom smoke bad.
- Unplugging results in the same problems, starting with the short to ground.
These shorting problems are why lots of pros are careful to turn down everything before plugging in their headphones, and why lots of gear recommend the same. But not everyone is going to be careful, and all it takes is a loose plug to screw everything up.
So best to protect against these kinds of problems.
As with everything, there are easy solutions. All sound like ass.
- Current limiting of the gain stage. Dead easy. Used in tons of speaker amps. Clips nasty and sounds like butt. No thanks. We don’t do that.
- Big ol output resistors. Dead easy. Makes for crap output impedance. Also no thanks.
- Active jack sensing and configuration. Test and see what’s plugged in, reconfigure and connect in real time. Easy if you use a headphone driver chip or codec with that built in. Pretty horrible in terms of sound—you’re reconfiguring grounds with MOSFET switches and stuff like that.
Now, we have our own transparent protection scheme, used in Magni+ and up, which measures current from the power supply and shuts things down when things get out of hand. Works so good, we only recently got our first smoked Magni+ back (after selling them since November).
So I started by adding that to the Freya S2 proto. Unfortunately, +/-32V rails are wayyy different than +/-17V rails, so it could still have smokey failures.
And that almost ended the project.
But I realized:
we can also sense when the headphone jack is being plugged in. The TRS jacks we use have three contacts, and three fingers on the opposite side. We could use the ground contact as a sense, and have the microprocessor open up everything until it was safe to connect.
That was neat. And that worked!
Problem is, you could still pull the jack out, and terrible things would happen when it shorted.
Argh.
But then I realized:
as you pull the jack out, the other fingers make contact with their opposite sides. Which meant you’d get audio on those contacts. In normal operation, no audio. If it was shorted, you’d get audio. And if you sensed whether audio was present or not, you’d have a failsafe for the jack in toto.
So I did another board with that sensing in place. Cranked the music, jammed in a headphone…no problem. Pulled it out…no problem. Left it shorted, cranked the music…click, everything safe, no problem.
What’s really cool is this protection system is completely out of the gain stage, out of the loop, out of the audio entirely. So it doesn’t affect the performance.
Pretty cool, right?
Yeah, and also the longest part of Kara’s development. Ah well, it was worth it.
What’s In A Name?
Sometime in the process of working out the headphone jack issue, Freya S became Kara. This was a totally natural thing to do, I thought. It elevates Kara to a preamp of its own, not always in the shadow of the looming Freya+.
And it deserves to come out of the shadows. I like Freya+, sure, but I’ve always used Freya S, because it always works, there’s no tube nervosa, and (to me) it sounds better.
Kara is another step up. And it’s very different.
- Different feature set. Kara is now our only preamp with a headphone jack. It’s also the only preamp with significant protection and management for headphones, and the only one that has both Nexus differential and single-ended discrete stages.
- Different topology. Freya+ is the quintessential tube preamp, updated for the modern era, with some smart features like tube shutdown. Kara is a complex, fully discrete, fully differential Nexus, current-feedback design.
- Different measurements. Freya+ is very quiet and measures very well for a tube preamp. Kara is near the state of the art.
- Different controls. Kara has a rational control layout—each button affects the display of LEDs to its immediate right. Easy to understand. Freya+ has the old layout, which looks pretty, but has buttons affecting LEDs to the left and right.
- Different chassis. In addition to the rationalized control layout, Kara now has a random slotted top chassis to improve cooling.
So, yeah. New name. New capabilities. Hearkening back to the golden age of solid state preamps. In Norse mythology, Kara was a Valkyrie, the “wild, stormy one.” I don’t know about wild and stormy, but I do know that Kara is going to make choosing between itself and Freya+ very, very difficult.
