Schiit Bifrost 2

Yup, just tried Piety in my main rig and on low gain the BF2/64 treble issues are tolerable. But I don't plan to use Piety there, and it doesn't let the benefits of the upgrade board come through.

I swapped back to OG board, and though I miss the tighter bass and spatial precision, I think it's staying that way. I haven't heard LIM so this is just speculation on my part, but as Marv said ...

Edit: I wonder if there's a FW upgrade that can fix the tonality (DSP tweak?) ... maybe I should hang onto it.

 

I think in a recent Schiit stream Dave from the digital team mentioned the quad 8812 approach requiring some specific DSP to help match the outputs of the two DACs so it would combine correctly. I can't imagine it being any different from the LIM, so could there be any analog tolerance differences between the two we're unaware of, like the OG BF2 using a lower grade of the AD5781?

 

@Ash1412 This looks potentially interesting, but could you put this into layman's terms for those of us who are DSP-speak challenged?

 

I am wondering the same thing as I am on the fence whether I should pickup a BiFrost or wait for another update. LIM is very enticing, just not sure that I have the space.

 

I'm afraid that's all I got from that stream, since it's in Schiits interest to keep this stuff proprietary. But purely armchair speculating, I would say that a single 18-bit ladder like in OG Bifrost 2 where the entirety of the ladder for one channel is on one calibrated chip has an inherent precision advantage over trying to combine 4 16-bit ladders into one 18-bit channel like in Bifrost 2/64 where any of the separate ladders could undergo manufacturing, temperature, etc... mismatches that show up in the combined signal as distortion (this is similar to the issue that plagues discrete resistor DACs). On the other hand, @purr1n did confirm that the OG Bifrost 2 also needed specific DSP to fix the code switch glitching by the Analog Devices chips (the original designer of these chips showed up on headfi and pointed to glitching as a reason for why the AD chips would be challenging to use in audio applications, but also commended the Schiit DAC implementations). So my guess is that both Bifrosts are running forms of dithering to mask the glitches and/or the ladder mismatch, and Schiit themselves have taken the stance that the hardware-balanced ladder combination method in the Bifrost2/64 is superior to the fake-balanced single ladder method in OG BF2. Off the top of my head, Chord products and the dac2541 also implement the single DAC into fake-balanced output stage, so we should bring @soekris into this discussion as well.

What's strange is how the LIM can be praised for having amazing treble while the Bifrost 2/64 is criticized for treble issues if the chips and code are the same. I don't see any information in the 8812 datasheet pointing to different grades of chips. Assuming it is a mismatch issue, more dithering could always mask these errors but then you will compromise the signal i.e. dull sounding.

 

I dunno if it's that weird. Original BF2 had complaints about the treble too, and obviously some stuff is the same between the two DACs. It's not all about the DAC chips and code, right?

 

So you're saying you suspect it's either due to different grades of chips (no possibility of a FW update), or a mismatch issue (more dithering possible in a FW update)?

Yeah, I guess some had issues with BF2 OG's treble, but I never did. But I definitely do with the BF2/64.

Maybe we should collect some data on who had/has treble issues with which version.

 

The entire Analog Devices line of Schiit DACs had a reputation for sounding like "Modern R2R". Then the LIM came out and got rave reviews especially when it comes to natural timbre and coming a bit closer to "Classic R2R", so the expectation (including for myself as an OG BF2 owner) is that the BF2/64 would be an improvement in treble performance, but impressions from SBAF and other sources have pointed to better bass and mids at the cost of worse treble, or at least a brighter overall tonality. I, being the total armchair engineer that I am, am inclined to chalk it down to the chips and code, but of course thats not 100% of the circuit and there definitely are other aspects that could be at fault.

The LIM exists with much the same topology, so there's a discrepancy there that's difficult to find the root of. Dithering could help to alleviate treble issues but a big sell of Schiit DACs is having the actual chip precision to not rely on dithering/noise-shaping and retain the original samples. Collecting more impressions is one way of understanding the discrepancy better. I don't have issues with my BF2 OG treble either, just mostly the overly warm signature. I was holding out on upgrading cause $300 is a lot if it's not unanimously better. Maybe a loaner of the modules is in order?

 

If nobody in CONUS has one to spare, I'd be happy to send mine across the pond and have it do a tour of SBAF homes.

 

There are way more qualified people on this board than I but there is a lot more to the LIM than the TI chips and the software. When I had the 2/64 it flat out seemed like it was trying too hard with what it had (aside from the TI chips) to pull off what it was trying to pull off (went back to OG Bifrost 2 and its meaty sound). I own an LIM and it is as graceful and out of the way as can be, in the best kind of way.

 

What do you mean it was trying to hard?

i just thought it was much too bright and as if the sound/soundstage had been ‘ pulled apart’.
So it wasn’t cohesive.

shame because I kinda dug the textural aspect

 

I'm pretty amused by the emerging disenchantment with the 2/64, especially in light of the original Schiit narrative of testers (who exactly?) unilaterally preferring 2/64 over OG.

This loosely reminds me of the initial introduction of the silver screen/mesh Sennheiser HD 6X0s in the late 2000s/early 2010s. Hifi communities heralded it as a long-awaited lifting of the Sennheiser veil, remarking at its clarity and treble quality. Fast forward 10 years later and the preferences of these communities seem to have reverted to the prior "black silk" variants.

While the HD 6X0 may not be the best analogy given it is a transducer rather than a DAC, it's safe to say I will be holding on to my BF2 OG and eminently curious on which variant people seek out months/years from now (tangentially curious about the same re: MMB2 vs. OG).


EDIT: I also find the BF2/64's "highest measured performance of any of our True Multibit DACs ever" intriguing. We know how folks feel about Yggdrasil MIL vs. LIM.

 

Why would you care about what people prefer in the future/past/present? best thing you can do is to triangulate multiple impressions and see if you think the 64 would be fit for your system, and still you will never know until you hear it by yourself in your system.

In my system (USB BF64>Freya S>Kinki M7>Triangle Cellos in medium treated room) I need more treble bite/etch and I prefer the treble of SMSL SU9 over the BF64 even tho it's a less refined treble.

Yet I would not bitch about the lack of treble bite of the 64 on this thread every week because I know it's probably just my amp being too smooth.
IME The 64 board is a bassy-V sounding DAC with smooth sound overall. That's my TLDR about the 64.

 

the detail was forced = trying to hard. It lacked the other parts to present the detail in a cohesive way. I only used it in a Jot 2 at the time and have since sold that amp, so possible that it was a poor match with the Jot 2 and better with other amps, although I'd guess a lot of solid state amps would not be the best fit with the way they handle treble

 

Ok that makes sense and what I heard as well.

the cohesiveness suffered

thanks

 

Oh yeah, I didn't mean to suggest surrogacy of personal evaluation by internet sentiment.

It's more of a matter of curiosity on Schiit marketing alleging a unilateral preference for the 2/64 in preproduction evaluation among their testing group, where current opinions seem more mixed (not sure whether this is par for the course, but I don't remember them being as unequivocal about a product being preferred over a version it was replacing in earlier releases).

 

I 100% agree with this. I also wonder if overall impressions of the bifrost 2/64 being "bright" overall is because a lot of the impressions so far have been from people coming from and used to the warmer bifrost 2 og tone.

While definitely relatively brighter than the 2 og, I don't think the bifrost 2/64 is bright overall-- to me it's more neutral than anything (feeding the piety and the jar800). I'm one of those people who always found the 2 og to be a little too dark and murky/slightly muddy for my tastes; so the extra bite, air, clarity, and treble presence from the 2/64 was a very welcome change for me*.

What I'm trying to say here is: like nishan99 says, when interpreting other people's impressions it's very important to consider the context of their system and what they're coming from and triangulate from there. Using myself as an example, I didn't like how dark the bifrost 2 og and the modi multibit 1 sounded, so I knew from people saying it was relatively brighter that it would probably be preferable to me. I appreciate those impressions, even though I didn't agree with the subjective preferences, because they stated why and I could easily triangulate from there.

I don't know, maybe this is irrelevant nonsense from me haha. Ultimately I would just caution about reading into when people say things are objectively "bright" or not, and try to read between the lines to triangulate and draw your own conclusions. I think this is part of the implicit "literacy expectations" for readers at sbaf, but I think that might not be super well known to newcomers.


*To play into this and provide context for myself, I tend to prefer stuff that play things straighter: I prefer the 600 over the 650, the utopia to the clear/elex, the 3f with amorphous over nickel and over the black widow, the Gungnir Multibit a2 & bifrost 2/64 to the bifrost 2 og & modi multibit og

 

Wasn't this just largely like in-house impressions (i.e. Mike/Dave/Jason/other Schiit staff)? I don't recall much language about a large testing group enjoying it subjectively more. Maybe I'm just remembering what Jason drilled down in the Schiit Happened 2/64 chapter, though, vs. website language.

Anywhoo, I do think that it's a bit too reductive to think of LIM vs. 2/64 as same chips/same software = same sound, as others have stated. For starters, there's different opamps in the 2/64 output stage, and then (as Marv has pointed out before) the BF2's transformer is different than the other Schiit DAC's with internal trafos, and has its own sound. That plus other output stage stuff in the LIM vs. 2/64 (just look at the boards), the choke/separated supplies in Yggdrasil, etc. etc.

I get it, though, I also have found it interesting that there seems to be a big gulf between subjective impressions of 2/64's sound vs. LIM, both positive/negative. You'd think 2/64 would pick up at least a bit of that LIM timbre, but nope, so obviously there's other factors in play.

 

I have Modi Multibit, Gungnir MB A2/Unison and just got Bifrost 2/64, with a Modi Multibit 2 on the way. I'm looking forward to comparing all four. Maybe I'll sell them all for a Y+ LIM lol. We'll see.

 

Personally, I think the only way to settle this is a Schiit MB DAC-off with ten silent revisions of OG Bifrost Multibit boards with discrete ZOMG opamp mods vs. sixteen silent revisions of Gungnir A1 and A2 with USB 2, 5, and Unison boards and the Freya output mod vs. a tube modded unicorn zero day Yggdrasil A1 vs. 2/64.