• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Review and Measurements of Okto DAC8 8Ch DAC & Amp

boXem

Major Contributor
Audio Company
Joined
Jun 19, 2019
Messages
2,014
Likes
4,852
Location
Europe
I found this site after Dave corrected me:
https://crookwood.com/blog/dealing-with-unbalanced-gear-in-the-studio/
Look at the section floating balanced vs hard balanced.
Furthermore Dave pointed out, that hard balanced outputs in pro audio gear are short circuit proof in some cases.
Thanks for the link. Issue is that Okto dac may be in case 5 or 6. And the solutions offered in the link are conflicting. Transformer is a solution that is working in both cases, even if a bit of an overkill.

Concerning the short circuit proof:
- Do you know if the Okto dac is protected against these short circuits?
- If the protection exists, is it a protection against permanent shorts or intermittent ones?
- Is it certain that everything will measure that well with one output in protection mode?
- Do you think that throwing 1000 eur. in a fine piece of gear to purposely put it's outputs in permanent protection mode is intellectually satisfying?

Well, anyhow, not my money.
 

BDWoody

Chief Cat Herder
Moderator
Forum Donor
Joined
Jan 9, 2019
Messages
6,948
Likes
22,625
Location
Mid-Atlantic, USA. (Maryland)
This is what I was afraid of -- that I'd pass on subtle differences and conclusions that took me several days of careful comparison to reach, and people would see the review as negative.
..

And this is why the regulars here will often seem quick to jump on these no doubt sincere and well meaning, but really subjective reviews or commentary. Someone comes along to an Audio Science site, and gets his worst fears verified when he reads some 'review' by some random guy (no offense) talking about a digital sound and how one of the worst measuring devices in the history of the forum easily bests one of the best measuring devices... describing it as more 'transparent,' etc...

It's hard for some to get their heads' around the idea that beyond a threshold competence level of fidelity, the rest is audiophoolery and placebo.
 

FrantzM

Major Contributor
Forum Donor
Joined
Mar 12, 2016
Messages
4,335
Likes
7,719
And this is why the regulars here will often seem quick to jump on these no doubt sincere and well meaning, but really subjective reviews or commentary. Someone comes along to an Audio Science site, and gets his worst fears verified when he reads some 'review' by some random guy (no offense) talking about a digital sound and how one of the worst measuring devices in the history of the forum easily bests one of the best measuring devices... describing it as more 'transparent,' etc...

It's hard for some to get their heads' around the idea that beyond a threshold competence level of fidelity, the rest is audiophoolery and placebo.
+1

And for a few minutes reading some of the back and forth posts concerning how the Yggi did this or that better, I sincerely had to verify the URL bar to see if I wasn't on some subjectivist audiophile site ...
 

graz_lag

Major Contributor
Joined
Oct 13, 2018
Messages
1,296
Likes
1,583
Location
Le Mans, France
+1

And for a few minutes reading some of the posts cncerning how the Yggi di this or that better, I sinerely had to verofy the bar address to see if I wasn't on some subjectivist audiophile site ...

... Sent from my smartphone, please excuse any mistake or typos ... ;)
 

josh358

Senior Member
Joined
Jun 13, 2017
Messages
493
Likes
387
And this is why the regulars here will often seem quick to jump on these no doubt sincere and well meaning, but really subjective reviews or commentary. Someone comes along to an Audio Science site, and gets his worst fears verified when he reads some 'review' by some random guy (no offense) talking about a digital sound and how one of the worst measuring devices in the history of the forum easily bests one of the best measuring devices... describing it as more 'transparent,' etc...

It's hard for some to get their heads' around the idea that beyond a threshold competence level of fidelity, the rest is audiophoolery and placebo.
Honestly, the only time such debates interest me is when someone has themselves done an exhaustive level-matched A/B comparison, preferably blind. Otherwise people are talking through their hats. Otherwise, they seem to be between religious believers who think that everything sounds the same or "listen with their eyes" to an incomplete suite of measurements that they don't know how to interpret -- or, at the opposite extreme, people in the audiophool camp who believe that cables are directional and that you can hear the difference between power supply fuses, and don't understand the placebo effect or how to counter it.

We have to remember that most audio gear can be and has been ABX'd. I haven't been able to find any ABX tests of converters, but certainly, things like op amps have been successfully ABX'd. And a rapidly switched A/B test with focus on a particular attribute will generally show up any differences in sound -- which is very different from a value judgment. Thus when I say that the DAC8 has brighter highs, I'm not making a value judgment any more than I would be if I adjusted a parametric equalizer. Indeed, when I hear these differences, I often don't know *which* is right, because I don't have access to the analog source. Whereas when I say that the Yggdrasil sounded more real to me, I am making such a judgment, which is why I went to pains to say that this judgment that could be peculiar to me or my system or the material I favor -- or a result of confirmation bias, which, as the Harman research shows, affects us audio engineers just like it affects everyone else.

I'd add that, as I said, Amirm didn't measure what seems to be a cause of the sonic difference between these DAC's, the overshoot in the feedback loop of the Sabres, which the folks at ESS themselves say makes a difference to the sound and which I suspect is the source of the commonly-perceived difference in the highs.

As to being a "random guy," all I can say is that I'm a retired EE who spent his life working in pro audio and video. I have heard and evaluated a lot of gear, as have my compatriots, none of whom are in the "everything sounds the same" camp any more than they are in the audiophool one. But I don't take that as evidence that my conclusions are correct; part of making successful evaluations of audio equipment is recognizing the significance of listening bias and of poor methodology. The only real test here is to try it yourself -- the DAC8 demo is on loan, and an Yggdrasil or a Gungnir can be had on trial.

My one criterion in these discussions is that the other person has done a thorough level-matched A/B test himself, by which I don't mean listening to a few pop songs, but to demanding material such as orchestra, chorus, and piano as well. I listened to more than 26 albums, often to multiple cuts within each one, often coming back a couple of days later to revisit a track, make independent observations, and see if they matched my initial one. Amirm has done such a test between the Yggdrasil and another DAC without hearing a difference, and so I'm curious as to why his results were so different than mine and will set it up and make some loops. (I just heard from Pavel and I'll have to pack the DAC back up this evening if the next person says he's ready to receive it, so whether I can do this depends on the availability of my partner in crime). Otherwise, I can only suggest that people try ABing it themselves and reach their own conclusions.

But -- since people have a tendency to think that these differences are larger than they are -- I reiterate that the Okto is a fine DAC and an incredible bargain, and that I'm going to buy one myself. I bought a superb $4000 9038 DAC and ended up returning it, so I think my decision speaks for itself.
 

josh358

Senior Member
Joined
Jun 13, 2017
Messages
493
Likes
387
+1

And for a few minutes reading some of the back and forth posts concerning how the Yggi did this or that better, I sincerely had to verify the URL bar to see if I wasn't on some subjectivist audiophile site ...
Friendly question: have you ever done a level-matched A/B switched comparison between a Yggy and a Sabre DAC? If you haven't, do that with some violins, massed and not, and with high tessitura soprano, and tell me what you hear.

A more interesting question is which is right, and without access to an analog source I don't know what the answer is. Just that it may be a consequence of the overshoot in the Sabre's feedback loop, in which case the ladder DAC probably is. (And let's not forget that ESS themselves say that you can hear it.)
 

sq225917

Major Contributor
Joined
Jun 23, 2019
Messages
1,360
Likes
1,610
The okto has an utterly flat frequency response, perfect linearity and noise far below audibility. The Schit has the measured performance of a 30 year old 14 bit cd player.

It's nice that listeners have a preference for one over the other, but one of them is state of the art hifi and one is an effects box.
 

THW

Senior Member
Forum Donor
Joined
Nov 21, 2018
Messages
412
Likes
630
^ what was the listening test methodology?

if it wasn’t done blind with attempts to remove bias, I can’t say I’m convinced by those conclusions, tbh
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
There are a few things.
First of all regardless you are a retired ee I'll still reiterate some of the basics.
The edge of the square wave indicates high frequency response. The ringing indicates the type of low pass filtering. The ramp slope indicates the slewrate hence high frequency response.
From digital system, the frequency response is firstly set by the sampling rate. They are from 44.1/48khz 88.2/96khz to 384khz.
The scope showed the time frame is 0.05ms. And it responses to 20khz. And the edge is roughly 1/10 of that, so 200khz then recalculate back to the samping rate it's 384khz. So there are a few conclusions from this. It's utterly inaudible at such frequency of solely the slewrate it self. And how many records do you have are 384khz? It's completely meaningless. Let along the digital filter could be the sole reason that it could be different.

Then to the schiit part. I personally can hear difference between dacs. But because subjective evaluation being inaccurate nor consistent, we made objective measurements. It's clearly that the difference is one of them being worse. Subjective preference is one thing, objective performance is another. There is absolutely no way that schiit dacs are better.
 

dmac6419

Major Contributor
Joined
Feb 16, 2019
Messages
1,246
Likes
770
Location
USofA
This is what I like to see a Company that's not afraid to have a product reviewed (Tested), instead of all the B.S. word play from the B.S. member only audiophile clubs (I.E. reviewers).
 

josh358

Senior Member
Joined
Jun 13, 2017
Messages
493
Likes
387
The okto has an utterly flat frequency response, perfect linearity and noise far below audibility. The Schit has the measured performance of a 30 year old 14 bit cd player.

It's nice that listeners have a preference for one over the other, but one of them is state of the art hifi and one is an effects box.
You're assuming that those measurements matter. As Amirm often points out, typically, they're at levels that are below the threshold of hearing and *way* below anyone would hear in a real room at real playback levels.

Ladder DAC's will typically have some issues that delta-sigma dacs don't -- but by the same token, delta-sigma DAC's have issues that ladder DAC's don't. I showed you one, from ESS -- overshoot in the feedback circuit. And yes, ESS says that is audible.

Don't make the mistake of taking a naive approach to the interpretation of measurements. They must be interpreted with an understanding of the psychoacoustics of hearing, and the measurement set must be complete, as well.
 

josh358

Senior Member
Joined
Jun 13, 2017
Messages
493
Likes
387
^ what was the listening test methodology?

if it wasn’t done blind with attempts to remove bias, I can’t say I’m convinced by those conclusions, tbh
Fair enough. As I said, I'll attempt a blind test today, assuming I can round up help. Otherwise, I'll suggest that someone else on the demo list try it.
 

THW

Senior Member
Forum Donor
Joined
Nov 21, 2018
Messages
412
Likes
630
Fair enough. As I said, I'll attempt a blind test today, assuming I can round up help. Otherwise, I'll suggest that someone else on the demo list try it.

voltage level should be matched at output too, tests should also be repeated if possible for statistical relevance IIRC
 

Bliman

Senior Member
Joined
Nov 9, 2018
Messages
419
Likes
150
Location
Belgium
And this is why the regulars here will often seem quick to jump on these no doubt sincere and well meaning, but really subjective reviews or commentary. Someone comes along to an Audio Science site, and gets his worst fears verified when he reads some 'review' by some random guy (no offense) talking about a digital sound and how one of the worst measuring devices in the history of the forum easily bests one of the best measuring devices... describing it as more 'transparent,' etc...

It's hard for some to get their heads' around the idea that beyond a threshold competence level of fidelity, the rest is audiophoolery and placebo.
To me this is an interesting one. First I don't like the attacks on the reviewer. I appreciate very much that he expressed his opinion here.
But now for the interesting part. With this statement, it doesn't really matter what dac that you choose. Because every dac is good enough beyond what the ear can hear. Then all we here are basically doing is useless. If the difference between these dacs cannot be heard.
So can you hear a difference or not? This is a science website and I see almost all the time the bias thing. That there can be no difference and that it is in your head.
So the big question is there a difference in the dacs if you hear them and I mean all of them? And is there measurements that are not done but can make a difference in sound?
 

josh358

Senior Member
Joined
Jun 13, 2017
Messages
493
Likes
387
There are a few things.
First of all regardless you are a retired ee I'll still reiterate some of the basics.
The edge of the square wave indicates high frequency response. The ringing indicates the type of low pass filtering. The ramp slope indicates the slewrate hence high frequency response.
From digital system, the frequency response is firstly set by the sampling rate. They are from 44.1/48khz 88.2/96khz to 384khz.
The scope showed the time frame is 0.05ms. And it responses to 20khz. And the edge is roughly 1/10 of that, so 200khz then recalculate back to the samping rate it's 384khz. So there are a few conclusions from this. It's utterly inaudible at such frequency of solely the slewrate it self. And how many records do you have are 384khz? It's completely meaningless. Let along the digital filter could be the sole reason that it could be different.

Then to the schiit part. I personally can hear difference between dacs. But because subjective evaluation being inaccurate nor consistent, we made objective measurements. It's clearly that the difference is one of them being worse. Subjective preference is one thing, objective performance is another. There is absolutely no way that schiit dacs are better.
The overshoot appears to be in the delta-sigma feedback loop. Obviously, delta sigma converters don't operate at 44.1 kHz. I'd note too that someone from ESS says that the overshoot is audible.

I note too that you've heard differences as well. Honestly, everybody and his mother has heard "Sabre glare." I heard it long before I even knew that there was such a thing.

As to objective measurements, those only work if you make the right measurements and interpret their psychoacoustic effect. To use an obvious example, speakers that measure flat in a two channel room sound tipped up. This phenomenon has been long known, hence the use of a "house curve" to normalize subjective response. It's one of many examples of how measurements must be interpreted in the light of psychoacoustics. Those who naively assume that the DAC with the measurements that "look good" will be more accurate may be disappointed.

Here, for example, is Amirm's measurement of the Yggdrasil's linearity:
1564329811430.png

Looks awful, right? The Topping is clearly more accurate. Or is it?

Note the levels. Even if you fry your speakers and ears with 120 dB peak output levels, the Schiit doesn't show half a dB dB of nonlinearity until maybe -105 dB. That's going to be at a level of 15 dB SPL -- *below the threshold of hearing in a listening room.* So if we consider audibility rather than visual impressions, the linearity of these DAC's is identical.

But, as I've pointed out several times, Amirm didn't measure impulse response. This is an area in which the DAC's differ, audibly, according to ESS.

As far as the HF emphasis (and some other things go), I don't pretend to know which is more accurate. Most of the measurements I've seen can't tell me that, because the measured levels of most of these things are inaudible in almost any good DAC. Even Amirm's nightmarish measurements of the TotalDAC don't tell us much -- he said he didn't hear most of what he measured.

In my experience, most equipment designers put a lot of effort into correlating measurements with audibility. When they find something that matters -- clock stability, say -- they use the measurements to refine their design. Someone like John Siau can interpret measurements with a practiced eye, but most of us cannot.

So when I hear a difference, I frequently don't know which device is correct. Just what I like. Even distortion could be in the original recording and simply more obvious in the better performing device. But just looking at the measurements doesn't solve that problem, unless the measurement is far worse than these are.

To really know which device is better, you have to compare it to an analog input -- and unfortunately I'm not set up to do that now. Because all the measurement above tells you is that you're measuring a ladder DAC rather than a delta-sigma one.
 

josh358

Senior Member
Joined
Jun 13, 2017
Messages
493
Likes
387
voltage level should be matched at output too, tests should also be repeated if possible for statistical relevance IIRC
That's where I matched them, at the output of the amp, actually. To within +/- 10 mV.
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
The overshoot appears to be in the delta-sigma feedback loop. Obviously, delta sigma converters don't operate at 44.1 kHz. I'd note too that someone from ESS says that the overshoot is audible.

I note too that you've heard differences as well. Honestly, everybody and his mother has heard "Sabre glare." I heard it long before I even knew that there was such a thing.

As to objective measurements, those only work if you make the right measurements and interpret their psychoacoustic effect. To use an obvious example, speakers that measure flat in a two channel room sound tipped up. This phenomenon has been long known, hence the use of a "house curve" to normalize subjective response. It's one of many examples of how measurements must be interpreted in the light of psychoacoustics. Those who naively assume that the DAC with the measurements that "look good" will be more accurate may be disappointed.

Here, for example, is Amirm's measurement of the Yggdrasil's linearity:
View attachment 30219
Looks awful, right? The Topping is clearly more accurate. Or is it?

Note the levels. Even if you fry your speakers and ears with 120 dB peak output levels, the Schiit doesn't show half a dB dB of nonlinearity until maybe -105 dB. That's going to be at a level of 15 dB SPL -- *below the threshold of hearing in a listening room.* So if we consider audibility rather than visual impressions, the linearity of these DAC's is identical.

But, as I've pointed out several times, Amirm didn't measure impulse response. This is an area in which the DAC's differ, audibly, according to ESS.

As far as the HF emphasis (and some other things go), I don't pretend to know which is more accurate. Most of the measurements I've seen can't tell me that, because the measured levels of most of these things are inaudible in almost any good DAC. Even Amirm's nightmarish measurements of the TotalDAC don't tell us much -- he said he didn't hear most of what he measured.

In my experience, most equipment designers put a lot of effort into correlating measurements with audibility. When they find something that matters -- clock stability, say -- they use the measurements to refine their design. Someone like John Siau can interpret measurements with a practiced eye, but most of us cannot.

So when I hear a difference, I frequently don't know which device is correct. Just what I like. Even distortion could be in the original recording and simply more obvious in the better performing device. But just looking at the measurements doesn't solve that problem, unless the measurement is far worse than these are.

To really know which device is better, you have to compare it to an analog input -- and unfortunately I'm not set up to do that now. Because all the measurement above tells you is that you're measuring a ladder DAC rather than a delta-sigma one.
You contradict yourself and you don't seem to know how fundamentals work. It's meaningless to argue with you.
 

1234VICE

Member
Joined
Jul 25, 2019
Messages
27
Likes
27
Location
The Netherlands
It bothers me largely because it suggests to me that we aren't measuring the right things, or doing a good enough job of correlating the measurements that we do make with what we hear in a real system.

Indeed, if one is to compare gear based on the transfer function alone, one is making the implicit assumption that there are no time-delay operators in the differential equation of the respective devices. Only then, the steady-state response fully describes the behavior of the device in time.

As an example; if one is to adopt the believe that the steady state response fully describes an audio system, this leads logically to the conclusion that all audio systems with the same in-room FR must sound the same. In reality, rooms with different acoustics/speakers sound radically different, even with EQ applied. Moreover, humans do not even prefer flat in-room FR. I am highlighting room acoustics here, for naturally they contain many time-delay operators (reflections).

The underlying mechanism here is that humans obviously do not perform FFT, and our hearing features a time dimension. Then, to fully characterize a system by measurements, this would require to fully resolve its governing differential equations as far as I am concerned. Restricting oneself to the transfer function might not be justified. Another challenge indeed is to connect the differential equations (measurement) to psychoacoustics.

I do not know to what extend feedback (time delay) can generate audible distortion in a DAC, but it is a valid and interesting question. Naturally, room acoustics are much more extreme.

I feel some members in this thread reject any discussion beyond the published measurements. However, you are proposing a scientific method to test the suggested audibility of feedback (abx), and in my opinion, any scientific attempt to improve our knowledge on psychoacoustics should be applauded, rather than be disregarded as unscientific. Your hypothesis could be wrong of course, and we should treat subjective experience with extreme skepticism, but I value your two cents.
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
Indeed, if one is to compare gear based on the transfer function alone, one is making the implicit assumption that there are no time-delay operators in the differential equation of the respective devices. Only then, the steady-state response fully describes the behavior of the device in time.

As an example; if one is to adopt the believe that the steady state response fully describes an audio system, this leads logically to the conclusion that all audio systems with the same in-room FR must sound the same. In reality, rooms with different acoustics/speakers sound radically different, even with EQ applied. Moreover, humans do not even prefer flat in-room FR. I am highlighting room acoustics here, for naturally they contain many time-delay operators (reflections).

The underlying mechanism here is that humans obviously do not perform FFT, and our hearing features a time dimension. Then, to fully characterize a system by measurements, this would require to fully resolve its governing differential equations as far as I am concerned. Restricting oneself to the transfer function might not be justified. Another challenge indeed is to connect the differential equations (measurement) to psychoacoustics.

I do not know to what extend feedback (time delay) can generate audible distortion in a DAC, but it is a valid and interesting question. Naturally, room acoustics are much more extreme.

I feel some members in this thread reject any discussion beyond the published measurements. However, you are proposing a scientific method to test the suggested audibility of feedback (abx), and in my opinion, any scientific attempt to improve our knowledge on psychoacoustics should be applauded, rather than be disregarded as unscientific. Your hypothesis could be wrong of course, and we should treat subjective experience with extreme skepticism, but I value your two cents.
Multitone is not any different than music signal. If you can't get your head around that you don't really understand much. Another test that can be done ultimately is null test. Then you'll just see how linear things can be these days. And how ridiculous this is. Also if you know a bit about auditory system, we do perform "FFT" in our concha and essentially digitizing the signal(into pulses). There are many more holes here. I don't want to waste too much time.
 

Bliman

Senior Member
Joined
Nov 9, 2018
Messages
419
Likes
150
Location
Belgium
Multitone is not any different than music signal. If you can't get your head around that you don't really understand much. Another test that can be done ultimately is null test. Then you'll just see how linear things can be these days. And how ridiculous this is. Also if you know a bit about auditory system, we do perform "FFT" in our concha and essentially digitizing the signal(into pulses). There are many more holes here. I don't want to waste too much time.
You are very aggressive here. So it seems you think you know a lot. Me I don't pretend to know a lot. I have no knowledge here.
Do all the dacs here sound the same? And if not what measurements must we look at to know if they sound different?
And are there still important measurements that are not done yet to determine the difference in sounds? Or do you think that we know enough to separate them according to how they sound?
 
  • Like
Reactions: MWC
Top Bottom