Discussion in 'Headphone Amplifiers and Combo (DAC/Amp) Units' started by Vtory, Aug 11, 2020.
Would you say this wouldn't be a big improvement over a Liquid Spark?
Magnius makes a ton of sense to me as a product. The landscape of balanced amps was simply not affordable enough. When it first dropped the THX 789 was offered for $400, the same price as Schiit's own Jotunheim. The new measurement darling A90 retails for $499. For the people (mostly college kids I'm assuming) who these products are marketed toward, $400--$500 is a lot of money, especially if you need to buy a DAC and headphones to fill out the balance of your system. Now, you could argue (successfully I think), that these buyers shouldn't be futzing around with balanced audio at all and would be better off buying something like a Magni/Modi stack. But that's beside the point. There are outfits like Drop, Monoprice, and Topping that are offering these products, and I'm sure a lot of newbies look at them and think, "Oh, it's balanced, so it must be better than what I have." So Schiit completely blew up the value proposition in this end of the market with the Modius/Magnius stack. For the price of a Jotunheim (amp only), these upstart audiophiles can now buy a fully balanced DAC + amp stack that passes all of the ASR tests with flying colors. That's the point. Magnius is not trying to compete with $2k RME interfaces, or preamp setups based around $600 stepped attenuators, or even the amp performance of Jotunheim. It's not hard to understand why Schiit synchronized the Magnius release with an official review on ASR. It's intended for their audience (and redditors).
We can send a x2 loaners out. @ChaChaRealSmooth: Can you arrange. I also have the IFi DAC signature (very nice) for you and @Vtory.
My position is that chip designs will never surpass good discrete designs in terms of subjective sound quality. Nothing has changed.
None of the discrete designs are perfect and each have their own flavor. The zero point zero zero zero designs also have their own flavor; but this is not the point of these designs. If the idea that zero point zero zero zero appeals to your particular belief in science, that one number measured at 4Vrms (or 2Vrms), which will explode your eardrums and likely your headphones too after four minutes of burning up the voice-coils, represents all or most of what you need to know, then absolutely purchase this and use only the XLR outputs (at least for headphones).
Buy the Magnius because it's cheap and measures very well (exemplary), at least from the balanced output. For super efficient IEMs, the SE output at low gain (x0.5) also measures very well (little distortion and sufficient S/N ratio) at typically encountered micro-voltage levels.
Having listened to the Magnius for a while today, I really wish @schiit had gone with their gut feeling and gone with the Heresy-like design.
I know people here poo-pooed the Heresy "because opamps" (no, I won't believe you if you tell me that had nothing to do with your opinion), but I too quite liked it for it's harmonic richness.
I'd have taken that for a worse pot. Or no pre-amp functionality. Or a slightly higher price. There's a lot of trade-offs I'd have been more than happy to accept to have the Magnius sound more like I'd imagine it would as an expanded Heresy.
...Maybe I should just leave it at that.
Two inconvenient truths in audio. Others being equal (craps filtered out at this stage lol),
1. The hotter the better
2. The heavier the better
Nevertheless, jot volume pot and iem-friendly gain finally got me as I ain't very happy with magni 3+'s volume control. Both themselves justify 100 usd premium (over m3p) to me. One thing I am still doubtful is whether it sounds better than mag3+ in its unbal out. Guessing not, but i'm kinda hoping a small wonder.
I poopoo'd Heresy at x1 gain, which I think is fair. At high gain, it was pretty decent, good even, exceeded expectations. This keeping in mind that I went through the effort of a level matched triple blind test here (note that Magni 3+ also measures very well, but is overshadowed by that last few SINAD): https://www.superbestaudiofriends.org/index.php?threads/magni-3-heresy-and-3-blind-listening.8663/
Magni 3+ is one of my favorite amps period when including factors as performance for price and manageable footprint. I wouldn't be honest (and would eventually be accused of protecting Schiit) if I didn't eventually say that Magni 3+ murders Magnius. However, that is just my opinion and the way I hear things. Maybe I like the sound of distortion. Maybe this microdynamics, PRaT, expressiveness, emotional connection factor* is something I am imagining.
If one thinks SBAF audiophile speak is too confusing, unreliable, and ultimately bullshit; and would rather rely on one or two objective metrics with one or two static parameters, then they should absolutely do so. I totally get it and wouldn't blame them. Schiit at least gives us a choice.
*High correlation with more open loop circuits and circuits with less parts. FWIW, opamps and monolithic devices have dozens and dozens of parts in them.
Maybe the sound of the Magnius comes from using opa1688 as input buffers. The Heresy uses opa1688, but only as output buffer, and this could be the reason why it sounds better at high gain. Opa1688 has relatively high voltage noise density, around 8nv/Hz, even worse than 5532 across the spectrum. The opa1662 which is Heresey's input stage, is around 3nv/Hz, and should dominate the noise at gain >3.
The Magnius probably uses the opa1688 to lower the impedance sees from Lme49724 to reduce current noise. 49724 actually have voltage noise around 2 nv/Hz, so the 1688 might be doing more harm than help. The opa16xxs are pretty similar in the structure, so maybe a replacement with opa1612 would have helped. It makes sense not to use 1612 for the cost, but going $1 more with 1662 could have helped. The Modius uses 1662 for se, which sounds worse than the 49724 in balanced, that maybe 1662 isn't good enough. The best option is probably discrete jfet input stage that people loves here(might measure pretty bad and needs a servo). Almost all jfet input opamp have pretty bad noise, not a great option.
@purr1n Ok, that's fair, in regards to the Heresy high vs low gain. I generally only test with stuff like the HD600/650, so high gain often makes more sense for me.
Agreed. I was more looking at it as an alternative for people looking at the Modius. There are very few XLR amps in this priceclass.
I bet if you ask nicely, @schiit will send you their 36 chip paralleled opamp prototype used to compare (internal tina turner thunderdome), gotta build it to be sure ;D -- caveat could be you need to provide cryogenic cooling or, run it less than 5 mins.
I think the Muses 8920 and 02 are jfet input and measure extremely well. Shame Schiit didn't use them as I always found the muses stuff to sound the best among op amps . They are pricey outside Japan though so I imagine that is why they didn't bother. The TI stuff makes measurement based purchasers happy for cheap and they are clearly who the Magnius is for.
Schiit uses stuff like the OPA1688 in a few products. When you're trying to keep costs down, sometimes you have to weigh subjectivity, objectivity, and overall utility in component choices.
As a DIYer, a dollar or two difference may not register on your radar, but that can be a huge difference in these cases.
Also has to be something readily available in bulk quantities, not obsolete, and with support if needed.
If ever, then, they wanted to entertain something like a Muses opamp, I think they'd be in the realm where it would be better to just design their own discrete solution instead.
I could be wrong, though. These are just armchair counterpoints.
@Vtory: Do you have the Magnius on order? Curious for a head-to-head against the Geshelli Elish. Thinking of sending you the iFi DAC Sig & Magnius ASAP.
This please, @Vtory
I bought magnius because of your excitement over the pairing with 650 with the geshelli and figured it would sound pretty close.
At this point I feel like I need counseling or an intervention.
Muses opamps aren't that great for noise performance neither. 8920 is at 8 nV/Hz and 02 at 4.5 nV/Hz, though they don't have current noise. As input stage, discrete design could achieve 1 nV/Hz with jfets easily and smash all opamps. Schiit has used jfets for a long time, if they have the budget to use Muses, they would have used jfet.
I've had Magnius on a few hours and it's cool to the touch. How come it doesn't get warm ?
Feeding it with bifrost 2 XLR, yesterday I had an amazing session with BF2 and asgard 3 using HD6xx. Compared to Asgard 3 magnius seems to have some crispy ass mids. Using low gain Pot around 12 listening level. It's much more forward sounding than Asgard 3 or compressed sounding perhaps. I'm sort of seeing where @Hands is coming from with his impressions.
Maybe not on paper but they rejected noise very well when used in the Burson Play LPF better than most of the competition including some with better specs according to their datasheets. Which op amp performs the best probably depends on the design of the amplifier. Datasheets are often misleading at least based on real world results.
Would be interesting to see Jason design the best sounding op amp amp he could without being restrained by measurements or budget. It would not be practical though as I imagine the discrete stuff would sell better and is also more entertaining.
Yeah I am getting it in (delivery estimated next week or so?) Shootout planned too.
Just sending me zen dac sig would be enough.
Having spent a few days with the Magnius, here are some general thoughts:
1. It's not very good. I might even go so far as to say it sounds bad. It might sound a little better after a few days of continuous playback, but you know it's a really bad sign when your first thought on listening to a new piece of gear is, "Oh...Oh no...Oh no, no, no...Oh god...Oh no." *double checks all their software and other gear* "...Really? Oh no..." *starts trying different settings and I/O combinations on amp* ".....Going to go eat dinner or something. Can't do this right now."
2. It's quite two-dimensional and flat in terms of staging.
3. It seems like it has decent macro-dynamic, driving capabilities at first glance. I mean, it's certainly not weak sounding. But it's incredibly dynamically compressed. It's like the loudness wars of headphone amps.
4. It's not overtly bright, but there is a noticeable glare or sheen to the sound. Even with a pretty smooth, warm, NOS DAC, it sounds unnaturally clear, like an overdone HDR filter. You know the type, where someone uses an HDR filter that more or less looks normal, except they tweak the settings to make the picture look overly detailed without the usual shit that comes with sharpening and contrast boosting in digital photography manipulation.
So, while nothing immediately stands out in terms of tonal nasties, nor is timbre particularly bad, it's just a fatiguing amp to listen to. Hard to explain. I just do not feel well when listening to it. Gives me a headache. And it gets worse with OS DACs, even with gentle digital filters.
Edit for additional context: Even when you can get a DAC, selection of music, and headphone that present no subjective sense of brightness or glare with the Magnius, it still somehow manages to be fatiguing.
5. Mids sound thin and recessed. Lower end is OK, I guess. I feel like the low end can't decide what it wants to do. It's mostly clean, mostly fast, but has a mild, mild bloom at times. It's fine, whatever.
6. Probably the weirdest thing, this amp will override the sound of whatever you have plugged into it. Got a dark, mushy DAC? Too bad, Magnius is gonna clean things up REAL good for you. Got the brightest DAC out there? Haha, too bad, Magnius is gonna clean that up for you too with extra glare on top, while still obliterating any nuances between sources for the most part.
On the low end, it's not so bad. It's a little easier to pick out source differences. But, listening as a whole, I was pretty shocked how the Magnius sounded just as bad going from OS to NOS, where normally the NOS softening helps take the edge off brighter amps and headphones.
It's like it has built-in machine learning that's trying to make everything sound homogenous on the output.
Maybe better put, it makes everything sound fake in its own way.
I mean, I guess it's an...option, albeit not a good one...if you're a living embodiment of the "bUt ThE sEnNhEiSeR vEiL" meme.
@Hands You would prefer the Heresy over this then for single ended use?
Separate names with a comma.