Life after Yggdrasil: Watering the Ash

I wouldn't argue with that.

"A the top of the curve" didn't necessarily mean the absolute best possible value. Just a unit that's right up there. As opposed to something like DAVE which is 2-3% better, maybe, but >10x the price vs. less than double.

In short, the more ">" between each DAC, the bigger the difference, and I'd rank them as:

Yggdrasil >>> Gungnir MB >>>>>> Bifrost MB >> Modi MB​

Which clearly puts Gungnir MB in the "sweet spot".

-

So, why not use Gungnir MB as my point of comparison? Because I started this process with Yggdrasil and it's been the consistent factor in this thread/process. Switching confuses things. And I don't have a need, or desire, to take a step down in the systems I run at home - no matter how small that might be.

 

Not that I wouldn't miss your reviews and pithy comments (or maybe you'll keep the latter coming?), but I wouldn't blame you a bit if you blew it off completely. Your boat is getting lonely.

 

I'm a Torq stalker so I would really miss his reviews if he stops doing them or they get hidden from noobs and randos.

I'm still hoping for a Aqua hifi compare.

For the comparison above, with the > is there a numeric value associated with a >? 5%? 1.3%? etc...

 

Where is my dislike button when I need it. When I joined SBAF it was probably the first thread I read page by page.

 

Just random thoughts here ( and sorry if this is regarded as off topic ):

Would there be any sense of creating an open reviewing platform where you could, for example, score individual attributes ( like tonal balance ) with a graph. Other attributes, like micro-detail, you would score with percentages ( +/- ), but always against devices you have demoed. There wouldn't be any need for absolute values, ever. If you find a new reference device (in terms of micro-detail, for example), you just score it +5% compared to the second best, and so on.

When you have a comprehensive list of reviews, you could score all the devices with a some pre-existing algorithm and put them into an order that way. Alternatively, you could order all the devices by yourself and let the software figure out how to weight individual attributes to get that kind of result. This way the reviews would be more useful for everybody, especially for those who also have a comprehensive list of reviews with different preferences. With a press of a button you could organize devices differently, depending are your glasses rose-tinted or not...

You could also cherry-pick a device and use it as a comparison basis, for example, if you like its tonal balance. The software could suggest you cheaper or (averagely) better performing devices with a more hefty price tag - but with similar tonal balance.

Just my late night two cents.

 

That's pretty much where it sits now; see the last paragraph of post #716 - which really wasn't meant to be sarcastic or humorous.

Won't happen - at least not as a review, by me, at this point.

Someone else (or some "someone elses"), I guess, can take up the baton and drive that forward.

Nope, nothing so scientific.

Interesting ideas.

Not sure if they can really be done that way, and almost certainly would have to be by one person if they could, but it'd be cool if the math and slice/and dice worked out so it could.

I'd say "go for it" though!

--

Pending posts ... Schiit multi-bit quartet comparison, Pontus review, which will be "as and when I can be bothered". Then maybe the ADI-2 Pro review (if you need all of it's functionality, I'd summarize and say "just buy it now"), and maybe one final write-up if one "mystery" DAC I have in my rack actually sees the light of day.

And then I'm going to do what I said I'd do at the start of the year, and that's focus on a massive turntable upgrade (well, as far as my audio-energy/time goes at least). Any future DAC changes/upgrades will be dealt with at the time, and almost certainly not in this format.

 

So instead of a like on your thread, I'll post to say that reading all of your reviews have definitely gotten me to a better place musically. Thank you for taking the time.

For the last 8 years I've been based in China so I've got lots of access to chi-fi (and many do sound like shit) but not Schiit products, so to be honest if not for your write-ups (and Marvey too), I would probably have skipped Schiit until I return back to US. Now, I took a big risk to get a Mjolnir 2 sight unseen (or unheard for audio) via Taobao local importers (I could have bought direct, but didn't want to deal with China Customs) and I'm very glad I did. I had a Cayin iha-6 that was actually not a bad amp especially considering it's price (here in China, it's about 25% cheaper than in US). Now, I've decided to get either a Gungnir Multibit or Yggdrasil (and in this case, I'll order it to a friend's place in US, and just carry it back to China).

 

Hi Torq,

I have two questions regarding your Denafrips Ares tests. Looking at atomicbob's published measurements it seems that the 1 kHz sine wave @ -90 dB looks better when measured at the balanced output than the single ended output. SE seems more ragged and amplitude and phase are different for the different channels. Does that tie in with your listening experience? The other thing of note to me is the whooping 12 dB difference in DR, not sure whether it matters if it is 110 or 122 dB, but maybe the large difference is not healthy?
As far as I understand it, with these discrete R2R DAC the FPGA corrects the errors of the ladder in real time. Every individual ladder needs to be measured and the correction calculated. If this is well done, the difference between two ladders should not be too big. What I am wondering is, when do they make the measurement to calculate the neccessary correction of the ladder? Most people say that the ladders need a certain time to get to steady state. More importantly Denafrips say the DAC needs burn in, so what if the ladders change during burn in, would thy need a different correction? What if the resistors change over lifetime?
If I understood you correctly and the imbalance of the two channels might be due to variations in the manufaturing process, that would mean that every Ares may have its individual sound as the imbalance might be more or less pronounced?
Finally if I am not mistaken you listening is done with headphones, being a speakers only guy, I ask myself whether the channel imbalance might cause less fatigue in a speaker set up, because you perceive left and right speaker with left and right ear?

Last but not least thanks for this thread which allows to get a notion about the difference between different gear

 

My listening notes, and write-up, have me preferring listening via SE on the Ares. So, whatever is going on there is either less fatiguing or has other results that, in aggregate, I wound up finding less subconsciously distracting than via the outputs from the BAL connections.

It's not a desirable state of affairs.

It might be indicative of why one channel sounded more compressed/less open/less dynamic than the other, but I don't have a sure-fire way to correlate what I heard there to specific deviations in the per-channel measurements.

By no means do all discrete R2R DACs do any correction or compensation for linearity of the ladders at all, with or without the FPGA. In most cases, the FPGA will be used to control the switches/shift-registers in response to the incoming sample-word data, and to implement any filtering (etc.). Ladder correction/compensation is an entirely different ball of wax.

I have no idea if the Ares specifically, or any of the other DENAFRIPs units, do any kind of ladder correction. That's a question for @alvin1118. Unfortunately I'd say that, if it does, at least in the case of the Ares, I'd suggest that however it is being done - it isn't working very well. Depending on what the root cause is for the channel measurement differences are, of course (nothing says it has to down to the performance of the ladders).

The Spring DAC uses a separate ladder to compensate the primary one. Metrum do forward correction via FPGA. As far as I can tell the S19 did nothing at all in that vein and wasn't under the impression that the Ares does either.

If, indeed, there is some kind of correction or compensation for ladder performance in the FPGA with the DENAFRIPS DACs (or the Ares specifically) then if there is a real effect from burn-in this would need to be done after the burn-in period or the correction would be off.

This is not a very tenable way to do things, however. As you'd need units powered on for upwards of a week each after final assembly, and then you'd have to do high-precision measurements on every individual unit* (four ladders for a simple balanced unit), make sure those measurements were good (read some of @atomicbob's notes on how easy it is to have a measurement not be quite right for myriad subtle and easy-to-miss reasons). And then you'd have to individually update each unit with a custom correction profile (which you would probably not store on the FPGA itself).

That's a lot of work and it's hard to imagine that being what's actually happening ... at least not at this end of the market.

Then the correction would be off as they changed.

IF, and it's just speculation on my part, the differences between the two channels are down to variances in component tolerances then, yes, that's quite possible. That would remain true whether those variances were in the ladder implementation or in the analog electronics. Or both, I suppose, though I suspect the ladder is the place where this is most likely to be a real effect.

I used to audition all DACs in my speaker rig as well. As that system would now put the output of the DAC through an AD and then output it via an array of DA converters, and a whole bunch of different filtering and so on, it's not really a useful test here anymore.

But yes, it's possible that on speakers it would not result in the early-onset fatigue I got with headphones. Though significant phase differences between channels would be much more audible on a speaker setup than they will via headphones.

--

*Actually, it's worse than this even, since you'd really need to measure for every resistor/resistor-string, or at least the per-bit output of them in aggregate, in the ladder to know how to compensate for each set-bit in the input sample word. So for a 24-bit ladder, that's 24 discrete measurements, per ladder, per phase (for BAL).

 

My speculation is that channel imbalances in measurements suggest PCB layout differences. Note especially the difference in mains noise at 60, 120, 180, 240 Hz between channels in the spectrum graphs.

 

If that is the main/sole reason for the channel imbalance then all Ares will show this behaviour

 

Related to the previous few posts:

Do audio DACs have a self calibration process at boot time that matches (optimizes) the output of the DAC with the analog output stage of the device? Things like calibrating voltage regulators, offsets, reference levels, ladders etc... Do they have facilities to account (adapt) for variations over time such as thermal variance?

I recently started working on a SERDES chip that has a very (very) high sampling rate ADC inside. It is vital to match the output of the analog stage to the input of the ADC else you greatly reduce the overall performance (BERT/ENOB) of the system. I've never heard any discussion of these sorts of things in terms of audio DAC/ADC chips yet I'm sure they must exist.

 

@Torq Unless I'm mistaken I don't think the Theta Gen Va is included in the list. Would you be interested in doing another take on that one? I believe Marv put it just under the Yggdrasil in his DAC thread but I'm *cough, cough* not so sure about that. Or do you feel that you've covered the Gen Va enough in the past and you're not interested in that one personally?

 

Given some of the recent nonsense, I've pretty much given up on doing more with this at all. If I was going to continue it, it'd be an interesting unit to include. I just don't see that happening at this point. I'll wind up posting the notes for the stuff I've completed auditioning so far*, which will amount to:

And then that'll pretty much be that.

--
*Not sure when, exactly, those will post ... wasn't expecting to be online until October at this point - but that interruption is just today, and not all of it, so it won't help much there and then I don't think I'll be around for the rest of the month.

 

@Torq

That would be a shame, I've been reading your reviews with some interest! If the comments and discussions get on your nerves maybe have the mods clean up this thread and lock it so that only you can post in it? That way you'll be free to continue your reviews, at your own pace, and when it's done (if ever) it'll be a nice reference to have. From your point of view of course and people who don't like that kind of thing... well.

If people want to discuss back and forth, or offer suggestions, we could potentially have another thread for that.

 

@Torq It is obvious to me the great amount of time and effort that goes into each of your reviews. The list of DACs to audition is awfully long and always growing. It would be an impossible task to review them all and to the same high standard of your previous reviews, especially if you want to have a life / job / etc at the same time!

Maybe after a break you will feel a bit more enthusiastic again. If you decided to resume your DAC reviews, then perhaps you could be more selective and only review the most promising of DACs or those that catch your interest / meet your criteria. After all, it is you that is doing all the work!

It has been a privilege to have your expertise, both here and over at Head-Fi. Don't listen to the trolls or those that accuse you of bias, there will always be some people like that - they are a minority - don't let them get to you - they should be more appreciative.

@TwoEars Good suggestion. Could be an effective way of reducing the aggravation suffered by Torq.

Hmmmmm.... Audtion DACs or go sailing in the sunshine? An easy choice - I know which I would prefer

 

I don't think it's the commens here that are the problem, I think he's started to attact a lot of attention elsewhere on the web.

 

Appreciate the comments, but I'm not really looking for a "solution" to anything here. The, escalating, nonsense, particularly recently, both direct and indirect, including snarky references in other threads, and all the behind-the-scenes fuckwittery (most of which does not originate here, but is fueled by comments in this thread and a couple of others), has simply sucked the fun out of things).

Past tense.

Life is too short to spend time on things that aren't fun, unless they're actually necessary.

This is no longer the former; it was never the latter.

-

Combine that with the increasing number of half-assed me-too supposed-giant-killer products that show up with breathless hype (often not based on any actual experience with the f'ing things), loaded/stacked/controlled review tours, and that exhibit weak value, poor execution, poor attention to detail, and/or that cannot be readily addressed without hammering them hard (and upsetting things further), and I'd just rather not be involved at all.

-

That is, indeed, at least half, maybe more, of it.

 

As others here, I've really appreciated and benefitted from your detailed, thoughtful reviews, and I'll miss both their information and their unique style. However, unlike some, I am not optimistic that some process change will make a difference to the negativity that follows online anyone who calls it like it is. This failure mode has been manifesting itself since the beginning of online discussion. I withdrew from Usenet groups in the 1990s for similar reasons (for the most extreme case where I was a minor target, see https://www.wired.com/1999/11/usenet-ban-a-slippery-slope/), and since then I've practiced careful online self-censorship; the few times I slipped taught my inner censor to get back on the job even more sternly. Oh well, more time to listen to music or enjoy the outdoors (the sea for you, the mountains for me). Thank you!