2023, Chapter 15
At Last, Unity
Let’s cut to the chase: This is the most insane Magni to date.
After hearing the actual production product, in the actual production metal, I sent an email to the guys in Corpus saying “Holy ****balls, this doesn’t have the impact of Mjolnir 3, but other than that, this thing is absolutely endgame.”
And this is me listening to a Magni
with an internal DAC card.
That costs $189—
with the DAC. Or $119 without.
Welcome to Magni Unity.
The Start of a Crazy Consumerist Nightmare?
No matter how good it is, no matter how excited I am about this new Magni, some of you are already yelling at me.
“One year! Only one year and a whole new Magni? What the heck! Are you gonna start changing stuff every year whether you need to or not? Who do you think you are, Apple?”
Yeah. I know. One year between Magni iterations is unexpected.
Going into 2023, I didn’t think we’d be upending Magni+ so soon. It was a 2-year product. Maybe 3. It was really, really good. And I can say that this one-year product cycle thing ain’t gonna become a thing, because, well, it’s a lot of work. And we only want to change things when it really, really matters.
But, with Magni Unity, some things fell into place unexpectedly. Which was good. It was time to change. Heck, you could say it was necessary to change. Because the world is different now.
And this isn’t change for the sake of change. Oh no. This is the most change in Magni since Magni was Magni.
Let me explain. But before, let’s take a quick trip to the past.
The Bifurcation and Unification of Magni
The introduction of Magni 5, or “Magni Unity” marks the first time since 2017 that we’re introducing a single Magni.
In 2019, we split Magni 3 into Magni 3+ and Magni Heresy, a discrete current-feedback design and an op-amp based design respectively.
The reason for the bifurcation? Measurements.
At the time, there was an idea that Magni wasn’t a spectacular-measuring amp (not true—see
The Search For Unity chapter) and that we weren’t competent engineers because we never stuck a high-measuring op-amp in a box. So we did. And it worked very well. It got near the state of the art, with very little effort. And the idea that we couldn’t do measurement-centric products began to fade.
When we updated the line in 2022 with moderate electronic changes in an all-new chassis, we stuck with the bifurcation with Magni+ and Magni Heretic.
The main reason? Despite literally a dozen prototypes, I couldn’t improve Magni+’s performance without ruining the way it sounded. The Magni+’s current-feedback, low-loop-gain discrete topology did have a particular sound, and was chosen in our internal and external blind listening panels over op-amp and current feedback designs—but it never got much better than -105/106dB THD+N.
So we kept Heretic…
…and I kept searching for ways to improve Magni+’s performance.
And Then There Was Midgard
Remember the Midgard chapter, when I said I was stuck on the Magni+ design? And remember that I was lamenting that if I had a bigger chassis, a bigger budget, and could bring in a few tricks, I could really make an insane better-than-Magni amp that was still current feedback and low loop gain—and that’s what became Midgard?
Well, it turns out the new gain stage—what we’re calling our super-linear, high precision current-feedback topology—also performs very, very well when it doesn’t have the advantage of stacked rails!
As in, in the standard Magni.
As in, it was what I was looking for back in the Magni+ development days—and had missed.
Blame me trying to save parts. Blame me simply being stupid. But the fact was, for a few more parts, plus some carefully selected gain devices, I got a lot more performance out of a new Magni prototype. Much more than I ever expected.
How good? Like -116dB THD+N good. That’s better than any discrete Magni—10dB better than the outgoing Magni+, and even better than the voltage-feedback, high-loop-gain stage of Magni 2 Uber, the previous winner.
Aside: beating a voltage feedback discrete stage with a current feedback stage in THD+N is no small feat! Sorry, I know, engineering nerd feat, nobody cares. But I do.
Aside: Magni Unity also looks very different. With 4 current sources and 4 LED voltage references, there’s a whole lotta glowing going on under the hood of Magni Unity.
But measurements aren’t the whole story. Here’s the thing: this new glowy gain stage achieves this performance without increasing loop gain. It does it by increasing the inherent open-loop linearity of the stage. With this new stage, we’re within a couple of dB of op-amp stages using 130dB of feedback—while using only 30dB! That’s nuts.
And engineering-ese isn’t the whole story, either. This is also the Magni that many of our early listeners responded with, “Wait, this is a Magni?” shaking their heads in disbelief.
And there’s also the market…
Above: Magni Unity analog, typical performance.
It Continues with Economics
Here’s the thing: it’s harder and harder to keep to that $99 price point that Magni started at. Heck, we lost it last year when we went to Magni+ and Heretic, because, despite advancements in chassis design (our new recurve chassis) and easier assembly, literally everything else got worse.
The chassis changes we made, which were designed to get us to a lower overall cost, were completely eaten with higher materials and powder coating costs. So a planned decrease in costs became a holding action.
At the same time, parts costs went up 10-50% across the board, PCB costs went up about the same, and assembly charges went so bonkers that we seriously considered bringing our own SMD line in-house.
This really shouldn’t be surprising, because Magni was originally priced at $99 in 2012—and that was for an amp with no gain switch, no preamp outputs, a much less advanced gain stage, only rudimentary protection, and much more crude cosmetics. Just keeping pace with inflation means Magni should be $134 today. Adding all the other features means it should probably cost a lot more.
The result was, last year, a Magni and Modi stack hit $238, or $40 more than when we started shipping the $99 stack.
Argh.
Talk all you want about inflation and value, but there’s a powerful psychological thing about hitting a sub-$200 price point.
The problem was that if I de-contented Magni and Modi to hit those old $99 price points, they’d be profoundly different products. Magni would have to have fewer parts and probably lower power output, and lose at least one gain level. Modi would have to lose Unison and its dual-mode, zero-mA power supply, and probably its coaxial input as well.
So how do you re-take that price point, without significantly compromising the product?
By going to a single chassis.
And now more of you are yelling at me. “Ah hell, there are tons of one-chassis products, and they all suck!”
Exactly.
The trick is in going to a single chassis with zero compromise. Or, in other words, keeping the exotic discrete gain stage, the three gain levels, the advanced protection systems…aaaaaaand a real Unison USB DAC with excellent I/V conversion and great performance.
Aside: much of my “holy heck!” reaction at the sound of the finished Magni Unity is because this is a single box.
Best of all, if the DAC was an optional card, now Magni could be modular and upgradable. A mini-Jotunheim. Something completely unique at that price point. Which is amazing and crazy and very much in keeping with everything we do.
What’s funny is, I almost missed the trick completely.
What makes it even more funny is, I did it years ago and had mostly forgotten about it.
Out of the Memory Hole
Here’s the thing: when the Magni Piety was debuting, I came across some early prototypes—specifically of three different Magnis. One was what became Magni Piety, one was what became Magni 3+, and one was a third path that was more a typical Continuity architecture.
And all were built on a Magni layout that incorporated an optional DAC card!
So why didn’t I go forward with that idea wayyy back? Because it was, I thought, too confusing. It was easier to price Magni 3 and Modi 3 each at $99 and have a killer $198 stack. We didn’t need to get into modularity and variations on our least expensive amp. It was such a non-starter that I’d never even completed a DAC card for that variation of Magni. It had never been finished.
And yeah, given the economics of 2016/2017, that mades sense.
Now….oh boy, give me the economic advantage of losing one whole chassis! Especially since Unison USB DACs could be made very, very small. And also since the DAC could share a really nice, linear power supply with Magni itself.
I mean hey, this had a chance to jump under that $199 price point—and deliver amazing performance..
So I set about bringing the old design out of the memory hole. And it was a reasonable bit of work. Because everything had changed, from the chassis to the USB connector. And also because I had a bit more time under my belt in designing modular products.
As far as I’m concerned, any modular product needs to follow a certain set of rules.
- The cost impact on the base product should be minimal. As in, there shouldn’t be a ton of money going into fancy connectors and electronics on the main board. The module should bare the vast majority of the cost burden, to make it fair for people who don’t want or need the module. We’re pretty good about this in Jotunheim, but it does add relays, brackets, and connectors that wouldn’t be there.
- The space impact on the base product should be minimal. Early Jotunheims used vertical cards that ate space. Later ones use horizontal cards which allow more layout freedom.
- There should be a minimum amount of modules. Jotunheim actually has 3 optional modules. This makes sense for a fairly high-end product, which has a minimum take-rate on the phono module as well. Magni would have to have much less. Like only one.
And all of that is reflected in Magni Unity: the extra cost is a single switch and a 5-pin header on the base product, the DAC card uses no space on the main board, and there is only one DAC option: a Unison USB ES9018 card.
Aside: and there is no chance of a True Multibit option, sorry. It takes up too much space and power, sorry. We need smaller DACs. And more efficient DSPs. Maybe someday. But don’t hold your breath.
Best of all, it’s easy to build Magni Unity with or without a DAC card, or to add the DAC card in the future. And, it’s easy to swap the card in the case of a better DAC in the future. Which makes Magni Unity the only product anywhere near its price to offer both the flexibility of a modular design and the promise of upgradability, so it doesn’t have to be dumpster-fodder.
And, with the amazing new analog gain stage, the amp side will stay relevant for the foreseeable future.
The DAC? Let’s talk about that a bit.
DACs Without Compromise
When it came time to do the DAC for Magni Unity, I briefly considered using the C-Media CM6635 USB receiver we use in Hel and Fulla. But I quickly realized that if we were shooting at a real, no-kidding, no-compromise DAC, we’d better be using our own Unison USB.
Unison USB puts a stick in the ground. It says, “Yeah, we may only have a USB input on this DAC, but it’s the best USB input out there, the only one built from the ground up to be the best thing for USB audio and nothing else—no wacky formats-of-the-week, no weirdo drivers, just a real proper UAC2 thing, plug and play and never a problem.”
From there, on the whopping 1.8 square inches of DAC board, we managed to stuff on an ES9018 DAC running with OPA1656-based I-V converter stages, +/-17V linear supplies, plus separate regulated 5V and dual 3.3V supplies with isolation between the USB supply (which runs the 32-bit Microchip microcontroller only) and the DAC/analog supplies.
Now, getting there was interesting. Because while there were only a couple of prototypes of the Magni Unity main board, there were a half-dozen for the DAC.
At first, I thought we could run the ES9018 as a voltage-output device. This is a common way to run ESS DACs when looking for lower parts count and a simplified output stage. It’s how we do it in Fulla, for example. There’s nothing inherently wrong with this, but it never really performed where we wanted it to.
So, after tweaking that for a couple of versions, we went to a design with a proper I/V stage. This worked pretty well! But it did take a few revisions to get it where we wanted it, with additional filtering, isolation, parts swaps, and other small tweaks.
Aside: the I/V stage is one reason Magni Unity runs warm when you’re using it with the DAC card. High-speed op-amps like the OPA1656 run pretty hot. On top of a serious Class AB discrete output stage for the amp itself, that’s a reasonable bit of heat.
Aside to the aside: but nothing to worry about. We make stuff that runs wayyyyyyyy hotter—for far past the warranty period.
The result of all this? A tiny card that snaps into a small header in Magni Unity. Run it with the DAC for a single-chassis wonder. Run it without the DAC so you can try out Modi Multibit 2 or one of our crazy DACs. One thing’s for sure: Magni Unity’s new gain stage is fully capable of keeping up with the DAC of your choice.
Above: Magni Unity DAC card, output from Magni, mid gain.
The End of Op-Amps
Someone has probably noticed that this is the end of op-amps in our analog products. (Well, with the exception of Mani 2, which uses some really good parts that are hard to beat with a discrete stage).
In short: yes, this is deliberate.
There’s nothing wrong with op-amps and products that use them, but you all gotta admit that reading about yet another widget that uses an OP1234 and a BF123 and a A5535 is, well, kinda boring. And isn’t there an image in the back of your mind of a whole bunch of audio engineers sitting like a dog at the back door, waiting for the next high-measuring chip?
We decided not to wait. Our current-feedback discrete stages are absolutely exotic and industry-leading; stick ‘em in a solid milled-from-titanium billet and they wouldn’t bat an eye in the car-price space. We’re weird and sell them for op-amp prices. Or sometimes below op-amp prices.
And so yes, we’re out of the op-amp game. We’ll continue improving our own discrete designs…and not waiting.
Reclaiming Silver
Others have noticed—holy crap proper brushed aluminum is back!
Yes. Because we think it looks cool.
Unfortunately, yeah, it comes at a price premium. Blame inflation. Blame more costly post-processing. Blame what you want. Save a bit and get black. It looks great.
But I like aluminum.
So we have aluminum once again.
Longer Warranty
This is a real simple section: we decided it was time to increase the warranty on Magni. Now 3 years. Because they’re super, super reliable.
Is This The End?
Every time we bring out a new Magni, I say something like, “Well, I really can’t do much better than this.”
And Alex reminds me, “You’ve been saying that since Magni 2 Uber.”
Sigh.
So is there something better than Magni Unity on the horizon? Nothing I can envision. Certainly not next year. Definitely not without significant changes in parts or budget. I mean, yeah, I can make a better Magni. It’s called Midgard. But it’s also $100 more.
Here’s what might change things:
- If new paired or quad matched transistors show up. This might allow me to simplify the amp by reducing the number of packages on the board, which would reduce cost. But it wouldn’t necessarily be a great idea. More transistors are usually part of an increasing-loop-gain strategy, which isn’t us.
- If any of our current devices go away. Then we may be forced into a redesign. Currently everything looks good, but who knows what’ll happen in 3-5 years?
- If shocking new inexpensive/fantastic regulators/capacitors show up. Yeah, don’t hold your breath.
- If economics change significantly—as in, if producing the boards and chassis goes up or down in price a lot—then things change. If we can reduce cost, we’ll reduce the retail price. We’ve done this a bunch of times. We don’t subscribe to the endless-cost-increase wackiness that some audio companies love. But if costs go up, we may end up having to make some hard choices—including, maybe, putting more money into the product to make it even better, and move it to another tier.
- A flood of new op-amps that make us re-evaluate our position. Who knows? It could happen. But as with regulators, I wouldn’t be waiting, arms-crossed, for it.
In short, this is by far the best Magni I could ever imagine. And it’s getting harder and harder to imagine better.
And yes, “best” means “with the internal DAC.”
I hope you enjoy this all-new, one-box, best-ever Magni Unity!
