2025, Chapter 2
What’s New is Old Again
Stjarna was the wrench in our works of phono preamp development.
You see, until Stjarna, we were on a path.
Mani 2 measured great and sounded good, really near the state of the art for an inexpensive phono preamp. Evan, our resident phono aficionado, liked it.
Then we did Skoll. Skoll, in many ways, measured even better than Mani 2, and it had a lot more options, and more gain to play with. When Evan got Skoll, he declared, “Mani is dead to me.”
And then we did a completely bonkers solid state prototype, one that checked all the boxes of what everyone thinks they want in an all-out phono preamp these days. Fully discrete, differential Nexus, transimpedance or conventional input, hundreds of parts, insane measurements (almost -110dB SNR at 40dB gain!!!), tons of settings, dual mono, woof woof. When Evan got that, he sneered at Skoll. “Superseded!”
But while we were playing with these phono preamps, I also decided to finish up an idea that Mike and Dave were playing with, like, a decade ago. As in, an all-tube phono preamp with enough gain for moving coil cartridges, while still having a fully passive RIAA network. So I put something together. And gave it to Evan.
“So this measures even better?” he asked.
“No, it measures quite a bit worse,” I told him. “It’s tube-based, so it will be noiser, and will have higher distortion.”
“Noisier?” Evan looked concerned.
“Still probably fine in most systems,” I told him.
Still, he looked doubtful. Somewhat reluctantly, he packed up the prototype and took it home.
Some days went by. I didn’t hear anything from him, so I figured the prototype was at least still working. But the days went by, and I didn’t hear anything. This usually bodes well, as in, our staff are hoping we forgot they brought it home, so they could keep listening to it.
When I finally asked him, he grinned and said, “Oh yeah all those other phono pres? Put them in the trash. Right in the trash. This kills em all.”
I laughed. “Despite the measurements?”
Evan shook his head. “Oh your measurements can bite me.”
The Realities of Phono Preamps
Here’s the thing: phono preamps are weird. They have to do much, much more than any other component in the analog realm. More like what?
- More gain. Phono preamps need to do from 40-60dB of gain. That’s 100-1000 times the input signal. And the more gain, the more noise and distortion. So when a low-output moving coil cartridge running 1000x gain, times 4x thru the preamp, times 20x through the amp (you know, 80,000x gain…eighty thousand times the input…read that again…Eighty. Thousand. Times.) has a bit of hiss and hum at the speaker, yeah, like, well duh.
- More equalization. Phono sources have a baked-in equalization curve, aka the RIAA curve, which boosts bass and suppresses highs. This is due to the physics of cutting a record, where you don’t want a ton of bass causing big excursions in the groove, and you want to boost the highs so you can filter out some of that pesky hiss when you’re playing a record. Suffice to say, phono preamps are not “flat,” nor are they intended to be. And, even with the tolerances of modern parts, it’s not unusual to have 0.5dB differences—which, yes, are audible.
- More flexibility. Different gains (at least two settings, one for moving magnet, at about 100x, and one for moving coil, at about 1000x, are common), different loadings, maybe a mono switch for mono records (yes, they exist), or even different equalization settings for non-RIAA recordings (yes, they also exist, because record companies didn’t standardize to start).
Still, if you use a modern op-amp, say like an AD4898 or OPA891 or something like that, plus 0.1% resistors and 2% capacitors (1% if you can get them), you can get pretty spectacular results from a phono stage. Yes, even if you use passive RIAA between 2 gain stages, which makes things a bit more complex.
If the above sounds familiar, that describes Mani 2. Great op-amps, modern high-tolerance parts. That gets you nearly state of the art performance (90+dB SNR at 40dB ref 2V, 0.1-0.2dB RIAA tolerance).
Of course, you can do more.
Parallel a bunch of low-noise JFETs, and you can do better, like Skoll.
Parallel even more low-noise JFETs in a unique Nexus™ architecture, and make the stages beefy so you can reduce the RIAA impedance, and you can do even better, like the unreleased Hati, which really pushes the state of the art. (110dB SNR at 40dB balanced.)
And then you have tubes.
Tubes suck for phono preamps. At least on paper.
Why?
Low noise? Nope. Physics says no. Even low-noise tubes like 6N1Ps, or even the fabled WE417, absolutely suck when compared to modern JFETs. And paralleling them? Yeah, ah, well, sure, but with the heater current requirements, as well as overall cost, you need to think long and hard about paralleling tubes. Oh yeah, and we’re still talking about the inherent noise of the tube, not noise from imperfect plate supplies or heaters. Bottom line: no tube phono preamp will ever be as quiet as solid state. Period.
Equalization accuracy? Yeah, problems here as well. If you use the tube to drive a passive RIAA network, the equivalent plate resistance of the tube will influence the RIAA curve. You’ll need to take that into account. Or specify only one type of tube, and hope they don’t age out too bad.
Gain? Argh, no free lunch here either. Tubes are inherently low gain devices. JFETs and BJT transistors can have single-stage gains in the thousands or tens of thousands. Tubes are 20-100.
All this adds up to: it’s a lot more difficult to do a tube phono preamp.
Especially when you are shooting for 60dB of gain, like we were with Stjarna. I mean, look around. How many 4-tube phono preamps do you see with 60dB of gain? How many without a pre-preamp section based on JFETs? How many without an internal step-up transformer? Ones that you can roll different kinds of tubes in?
Yeah. The number gets very, very small, doesn’t it?
Bottom line, Stjarna is special.
“But if tubes suck, why do it at all?” someone asks.
Great question.
Because It’s There
No, seriously. And no, I’m not talking in some Victorian code-speak about climbing tall mountains. I mean, because we have a metric ****ton of 6N1Ps, and it would be nice to do something with them.
“Oh gawd, you’ve gone full corporate asshole,” someone says. “You did this because you have a lot of parts?”
Weeeeeeeeelllllll…not entirely.
But it didn’t hurt.
Here’s the thing: 6N1Ps are really nice tubes. And one of their key attributes are that they are fairly low-noise for a tube. So a phono preamp isn’t beyond the pale.
It also didn’t hurt that when Mike and Dave were messing around with a tube phono preamp, they were using 6DJ8 tubes. Which are kinda-sorta like 6N1Ps. 6N1Ps are more rugged and lower noise, though…
…and so when I dug up Mike and Dave’s old prototype for a show, and saw it did 58dB of gain, I realized: we have a potentially really neat phono preamp here!
“So this is a Mike thing?” someone asks.
In short, not really. I don’t want to mislead you. Enough time had passed between Mike and Dave’s Unrealized Prototype and my idea that:
- Dave forgot what he did with it
- Mike and Dave didn’t have schematics
- I am lazy, so I didn’t trace out exactly what they did
So this is a Jason thing, inspired by Mike and Dave’s old ideas.
It’s the same in that it’s a 4-tube design with a lot of gain. It’s different in that it has a whole lot of flexibility and technology packed into it. Switchable gain and loading was never in the cards for their prototype; it was part of Stjarna from the start on mine. It’s the same in that it has a dual mono power supply. It’s totally different in its discrete regulation, DC heaters, and custom transformers.
But I’m still underselling Stjarna. It isn’t just a way to use 6N1Ps, or a design direction resurrection.
I think, more importantly, it’s an
exploration.
As in, I was curious about how a tube phono preamp would sound.
After all, I’d started my phono preamp journey exactly where an objectivist would: with Mani 2, an op-amp based, high-measuring phono preamp. And Mani 2 taught me some interesting lessons, including:
- Sound was strongly influenced by capacitor quality
- The gain stage—the op-amps themselves—didn’t matter as much(!)
Yeah. I know. But that’s what I heard.
All that continued until Skoll, where I decided to go discrete…and no-feedback. Now, the gain stage mattered. Even with the same RIAA network, Skoll sounded better than Mani 2.
And Hati was even better.
So how would tubes do? Would they be better? Worse? Was there some crazy synergy between technologies of similar provenance?
I had to know…
…and that’s pretty much how you got Stjarna.
Evan’s reaction, and the opinions of other early listeners, seem to confirm there is something about a tube phono preamp, something fairly magic.
“But it can’t possibly be accurate,” someone says. “Just look at the measurements of this thing compared to Mani 2 or Skoll. It must be coloring the output!”
Except, well,
maybe not.
Here’s why: you don’t know the total fidelity of the recording chain, do you? Especially with old recordings, and especially with ones pressed onto vinyl, with equalization baked in, from a master made by a mechanical cutter that has its own limitations.
I mean:
- What is the inherent THD of the microphones?
- What’s the studio noise floor?
- What medium was used for the original recording? Analog tape? Digital? What are the limitations of old analog tape or early 14-16 bit digital?
- How was the recording mixed? On more tape? With what THD? With what noise floor? What’s the THD of the compressors or limiters or effects applied? How’s the noise with those devices?
- What kind of cutter was used to master the record, and what’s it’s inherent THD? I mean, you’re talking about a device that cuts grooves in metal; it needs a ton of power, it’s a mechanical, magnetically-driven device. You really think that’s running .00003% THD?
- How good was the master after being cut? Was the engineer having a good or bad day?
- How about the pressing itself? Was the master new? Fresh? Hot?
- And the vinyl material—how good is it? How perfect? Thick?
With all these variables, it’s amazing that we get any semblance of fidelity at all. But it works. And when it works well, it really can be magic.
But let’s dispense with this idea that a tube phono preamp is gonna muck up what’s on the record. Records are realistically probably running about the same level of THD as the phono pre, when you get right down to it.
And, to take it a bit further, records and tubes are probably running better THD than your transducers—that is, your speakers or headphones. Really good speakers and headphones can manage 0.1-1% distortion. That’s it.
“But your phono preamp is getting near there at high gain,” someone says.
Yes, at high gain
with full output.
As in, at 2V RMS.
That’s a spicy, spicy record!
In reality, you don’t run full output from any record all the time. Music is dynamically changing. Sure, you could hit a 2V peak. But most of the time it’ll be 10 or 20dB lower. And then distortion will drop down to levels a hundred times lower than a good transducer.
Aside: this helps illustrate why spectacular THD numbers aren’t highly correlated with sonic perfection, and why we can do things like blind tests between Magni and Vali and come out confused: we don’t typically use full output; levels change all the time; much of what we listen to is below where THD is rated; dynamic effects aren’t even disussed.
So if you like what you hear from Stjarna, it’s not necessarily coloration.
And here’s the thing: if you want better measurements, it’s dead-easy to get them. Mani 2 does it for $149.
From Old to New
Stjarna, from the start, was intended to be a tech-forward phono preamp. Development started concurrently with Skoll, so the overall idea of a microprocessor-controlled device with relay ladder loading and relay-switched gain was always where we were shooting.
Of course, we gave you more loading options, leading to a comically immense array of LEDs on the front. This was still pre-Forkbeard, so I shrugged and considered it acceptable, but breathed a big sigh of relief when we went to Forkbeard later.
So how’d development go?
Not too bad! The first prototypes pretty much worked. The discrete regulation for the HV supplies had some problems and needed tweaking, and the gain changing relays worked in reverse, but the basic idea was solid.
Aside: I hate products like this. This usually means that something else that should be dead-simple is gonna be a huge fight.
The second round of prototypes cleaned up the discrete regulation problems and fixed the relays, and, after adding standoffs to the board (the early ones I’d literally forgotten to put anything on the board at all), we were ready to start thinking about production.
Except:
“Hey Alex, I need to get some more Valhalla 2 transformers for the protos, can you have some sent out?”
Alex looked at me blankly. “Valhalla 2?”
“Yeah. We’re gonna use the Valhalla 2 transformers in Stjarna.”
“Ahhhhhhh…” Alex said. “I wish someone had told me.”
“Why?
“Because we scrapped them.”
“All of them?” I asked.
Alex nodded.
Well, that sucked. I’d planned to use Valhalla 2 transformers left over from the end of Valhalla 2 production. We literally had thousands of them, due to over-ordering before we got our ERP system up and running.
“We can order more,” Alex said.
I shook my head. The Valhalla transformers only kinda-sorta fit in the chassis, requiring a milled pocket for the taller one. “Let me try something.”
What I tried was the new Valhalla 3 transformer, which was, at the time, a bit too hot-running for Valhalla 3. But running two of them for dual mono in Stjarna was no big deal at all.
I did another prototype with the Valhalla 3 transformers, and, by that time, a Forkbeard port. That one pretty much just worked.
Of course, nothing is ever easy, so the Schiit crew at head-fi convinced me to add a mono switch (another relay, another LED, reprogramming, etc). And I decided to add a standby mode that completely turns off the tubes—no heater, no HV, nothing. And there was a bit of tweaking on the RIAA network, which is really really good…it’s the only place where we can use 1% capacitors and 0.1% MELF resistors.
And then, the final bit of hilarity:
“What about the remote?” Tyler asked.
“What remote?”
“The Stjarna remote. Does it use the same one as Skoll?”
Ah crap. I got a horrible sinking feeling. “Ah, maybe. Let me look.”
And then I remembered the Skoll remote was plastic.
Oh yeah, and the functions Stjarna needed meant it was too big for our standard aluminum remote. It would need something big and stupid, like the Loki Max remote.
Aside: yeah. Remotes suck.
Arrrrghhhhhhh.
“We have Forkbeard,” I told Tyler. “Maybe this one doesn’t come with a remote.”
Tyler looked uneasy.
“Ah come on, Mr. We-Should-Get-Rid-Of-Manuals-Entirely,” I prodded him. Tyler has long opined we shouldn’t have manuals at all, just a QR code to point people at something online.
“Do you think it’s too early, though?” he said.
“Too early to get rid of remotes?” I asked. “To, specifically, get rid of something we still need to draw up, have prototyped, integrate with the product, and buy a short run of, because this is a limited-run kind of thing?”
Tyler nodded. “I know, but this thing is also going to be bought by people who may, ah, not want to go app-only.”
Sigh. He was right. I drew up the remote.
And that’s how Stjarna kept its remote.
“What about Forkbeard,” someone asks. “Bluetooth and tubes sounds like a horrible combo. Isn’t there interference?”
Surprisingly, no.
Even at 60dB of gain, you can’t measure the impact that Forkbeard has. Thank BLE. It’s very efficient. Heck, the module only draws about 0.5mA while operating (yes, 0.5mA, that is not a typo.)
So you get your choice—IR remote or Forkbeard.
And, perhaps even more importantly, after some transformer and chassis revisions, you get a Valhalla 3 as well!
The Hazy Future
“So I see you’re calling Stjarna a limited run,” someone says. “Do you mean limited like Folkvangr, or limited like Mjolnir 3?”
Fair question.
Because Folkvangr really was limited. It was limited by the amount of 6N6P power tubes we had. It was also limited by demand. It had a nice short run. I enjoy the heck out of mine, and I know a bunch of people enjoy theirs. And we can continue to supply tubes, service, and support for that kind of short-run product indefinitely.
Mjolnir 3 was limited because I didn’t think there would be many people who wanted a big, heavy, hot desktop amplifier. Demand was much higher than we expected. So we kept making it. I believe it may have run its course, but maybe I’m wrong. I expect you’ll be able to buy a Mjolnir 3 until the middle of 2025 or so. So still somewhat limited.
Stjarna is limited like Mjolnir 3, as in I’m not sure how many people really want a big crazy tube phono preamp that is nearly $2K once you get into tax and shipping. Maybe I’m wrong, but I think a couple of limited runs (one black, one silver) is it.
If demand exceeds what I expect on Stjarna, would we keep it around? Probably.
But then there’s Hati, the as-yet-unfinished all-out solid state headphone amp. It still needs a bit of tweaking to ensure the cascaded Nexus stages don’t go wacky with extreme low-frequency input (think warped records), and we need to experiment with transimpedance input a bit more, but that’s something we could do, too. It wouldn’t be massively less expensive than Stjarna, though.
So what does this all mean?
It means, at least for a time, you’ll have the option to step up to Stjarna, one of the most all-out phono preamps on the planet. I mean, 100% tube gain to 60dB…and Forkbeard? Ah come on!
And, if that doesn’t thrill you, there’s still Skoll and Mani 2, both really excellent phono preamps in their own right.
(Oh yeah and now Skoll F has Forkbeard. Har.)
No matter what you choose, or even if you choose no phono preamp at all, I hope you enjoy!
