General debate thread about audio measurements

[March Audio]

Okay, fine. You're comparing the performance of a 24 bit stream in a 24 bit DAC (at -90dBFS) vs. a 24 bit stream in a 16 bit DAC and you're surprised the latter looks bad?

Now I really am confused

No that is not what we are doing.

The - 90dBFS time waveform and linearity test is to look at how well and how accurately a dac renders low level signals.

There are very few 16bit dacs out there these days for good reason. The squared waveform is simply an indication of how a good 16bit dac would perform, that's all, so put it out of your mind for the moment.

Why - 90dB? Well it represents the bottom end of a 16 bit waveform, which probably the majority of recordings the majority of us own. It's not going to an extreme,.

A modern 24bit dac should be able to reproduce this signal well and many, even budget designs, do. Some however don't. No names mentioned. As shown above and below some don't even come close to good 24 or 16 bit performance.

 

[Wombat]

FrivolsListener.

To avoid the lengthy to-and- froes I suggest that you do some prior research and be more accurate/specific with your questions. It is unfair on forum members who respond generously to your questions to be have be mind-readers, and then to find you repeatedly bringing-in qualifications, corrections and new information in your posts. The bar re discussion on this forum is higher than where you appear to be. Listen and learn. Know your stuff and clearly put it forward.

 

[Dismayed]

Which? About the Lyr? I didn't have a specific example in mind, but I can go and dig up a link on a measurement article of a Schiit product by atomicbob. Hold that thought.

You quoted my response to your post. The context is there.

 

[Dismayed]

Ill take the fuzzy distorted one

Because the reviewer said it produced a broader soundstage? Because of the DAC's base slam? Or was it the palpable presence that drove you to the fuzzy distorted one?

 

[March Audio]

Because the reviewer said it produced a broader soundstage? Because of the DAC's base slam? Or was it the palpable presence that drove you to the fuzzy distorted one?

I just fink it looks more purty

 

[Dismayed]

You are confusing two important things: my measurements, and the general population of the forum. You can't use the latter against the former. I walk into every review of Schiit products hoping for great performance. And I have spent my own money on tons of their gear, much of which I have yet to review. Yes, I have had plenty of negative experiences but I do not let that impact my measurements one bit. I do not in any way create a situation to make them look bad. I pick fair comparisons to their products as best as I can to see how well they do.

There is a constant attempt to dismiss measurements by talking about listening tests. But where are your controlled, scientific tests of that? You say this and that is not audible but based on what study? You are throwing arguments forward to defend less than performant measurements. That is not going to work when they are purely arguments. Show me controlled tests that says distorting -90 dB signals is not material in "music." Implying it is not proper science.

I would ask a Test & Measurement person to weigh in on your procedure. An experienced person with years under their belt. Again, the tool doesn't make the person irrelevant. See your Chinese chef example above.

Perhaps FrivolsListened should read up on Amir's background. He's actually remarkably qualified.

 

[Wombat]

Perhaps FrivolsListened should read up on Amir's background. He's actually remarkably qualified.

The subjectivists nightmare-a more qualified, experienced and rational person.

Who needs experts? E.g. people who advocate measurement, acknowledge psycho-acoustics, don't trust 'golden ears'. Fakes, all of them.

Well, it just doesn't wash here.

 

[FrivolsListener]

Okay, let us back up here a step or two.

This all started with a Bifrost I think which is 16 bits only. I couldn't find where atomicbob fed it a 16 bit signal at -90dbFS. He did feed it one at 24 bit. A good 16 bit DAC with low noise and good linearity would have resulted in this kind of graph. Yes this is a 16 bit signal fed to a 24 bit DAC, but a good working 16 bit DAC would look this way too.

Instead we get this. It would appear there are linearity issues in the lower bits of this 16 bit DAC which is one reason 16 bit DACs aren't much used anymore. If it had handled the top 16 bits, and ignored the rest you would have gotten the above graph with maybe a little more fuzziness from a higher 16 bit noise floor. My synthetically generated graph that looked similar to the above is one where I took a 32 bit version of the signal and saved the top 16 bit without dither truncating the lower bits. They were just chopped off, ignored.

Here is a digitally created version of -90 db 16 bit with dither.

View attachment 11889

That doesn't look much like the Bifrost at 90 db. Yet atomicbob said: Incedible performance for a 16 bit resolution DAC.

Now just to confuse things we mostly have talked about the Gugnir. Here is the Gugnir with 24 bit signal after 27 hrs warmup out the balanced connection via Toslink. All the other digital connections were worse than this and single ended even worse still. It barely looks better than dithered 16 bit.

So I think my post with all the bunches of graphs and back and forth over this and that begins to become mind numbing like atomicbob's. I appreciate atomicbob supplying measurements as they are valuable. His description of the results seem not to match what I see sometimes. But the measurements are there I can ignore comments and look at what he measured. The same is true of Amir's measurements. If you think he has it in for a brand or is a fan boy of a given item you can ignore that to look at measures for yourself.

That is if you understand what they are showing you. FriVolsListener you've said you don't know about some of this stuff just trying to understand. That is fine none of us were born knowing these things. And maybe looking at this is jumping in the deep end of the pool in a sense. We'll try and explain it to anyone interested who doesn't know.

I think it has been lost in a few recent threads that Amir's and atomicbobs measures are much the same. They indicated similar problems or similar good results. It is the descriptions of what that means which differ, and some owners of gear they feel was maligned because results weren't as good as expected seem to have lost that amidst the back and forth.

Thank you. That's where I was coming from.

Okay, so that 16 bit plot is the one I cared about.

Now, does the Biifrost look bad? It does. Is it significant? That far down from FS is probably (red flag: It's a guess. I don't *know*) inaudible.

Me, I happen to have the Gungnir (and D-30). But 98% of my content -- the stuff I listen to, as opposed to look at on scopes -- is 16/44. I know the Gungnir sounds extremely good. (The D-30 isn't bad, btw, I had it in the system exclusively for a week.)

Schiit, like any company, does tradeoffs on design. They have branded themselves as a cost-effective way to get into good DACs. When I started looking for an external DAC, they were the only game in town. Now they have competition. That's good.

SO, for me, it comes down to whether I worry about crap at -98 dBFS at having the stuff at -70 dBFS and above more faithful, or not. Measurements matter, but as others have pointed out above, some things sound better than others, even with worse measurements. (For me, that means that 1. listeners can self-delude, 2. judging by someone else's ears is unreliable and 3. we don't know what to measure or how to measure what we percive.)

Have to leave it there -- work calls.

 

[Purité Audio]

It does appear possible that some manufacturers can design an inexpensive and yet fine measuring piece of equipment, sadly Schitt isn’t one of them.
Keith

 

[FrivolsListener]

It does appear possible that some manufacturers can design an inexpensive and yet fine measuring piece of equipment, sadly Schitt isn’t one of them.
Keith

It still comes down to whether or not it matters at -90 dBFS.

From my (admittedly very limited) understanding, DS will always measure better than r2r, but doesn't necessarily do a better job. I know that's heresy here, but what's happening -100 dB down just doesn't bother me as much as what might be going on at -70 dB or -40 dB.

And, again, I have the Topping D-30 based on its' rave reviews here. I'm interested in cheap and good.

 

[Purité Audio]

It may we’ll be inaudible, but it is stil poor engineering, even their biggest fan cannot condone that .
Keith

 

[DonH56]

So I have two DACs, both of which do great at -70 dBFS, but one falls apart at -90 dBFS and the other looks fine. Hmmm, which shall I buy?

I know the toast is burned, but I like burned toast, and don't want your golden-brown toast even if a spectral analysis shows it to be better.

 

[amirm]

I know that's heresy here, but what's happening -100 dB down just doesn't bother me as much as what might be going on at -70 dB or -40 dB.

No, no, no. No one is more realistic than our membership here about what people can or cannot hear. So don't go there please.

Your argument which you have lost was that I complain about measurements and others do not. And that must mean that something is wrong with me or the measurements.

Once you decide to run a measurement, then it is your responsibility to characterize good or bad response on it. I run the -90 db test because it is an industry standard. Stereophile runs it and so does atomicbob. In that regard, all three of us better say when those measurement results are poor when they are so. I have shown you how atomicbob results are awful in this regard. So the measurements he made and I are in sync. What is not in sync is that he heaps praise or at best, says nothing when such measurements are atrocious. I think that is wrong and does a disservice to his audience.

If atomicbob thinks those measurements have no value at all, which has to be the case since he ignores horrific results in them, then he should not run them. It does not at all follow that something is wrong with my work as you claimed at the start of this argument.

We here care about -90 dB not because we think you can hear those artifacts. But because it is supposed to be a super easy test in this day and age to pass. Indeed DACs as low as a cheap dinner pass it. So when a device doesn't, we/I point it out. We say that it is poor engineering. We don't say, "oh, anyone listening to it is going to cry that it is distorted." Indeed it is not because so many people buy these products.

Last but not leasat, we think it is important to point out such when the manufacturer has implied otherwise. That because they device uses R2R technology or so other thing, that it is more accurate than competing solutions. We set the record straight which is a service to the community. That allows such potential or actual customers go into listening test unbiased by mistruths.

Are we clear on this before you jump again to another argument?

 

[amirm]

And one more thing: what doesn't bother you is fine. What we are dealing with is you trying to make sure no one else is bothered by creating FUD around the data. You can't have it both ways.

 

[mindbomb]

@FrivolsListener
Dac resolution (which is what that -90db test shows) is arguably the single most important aspect of a dac. And schiit themselves seem to realize this, as that is the main difference between their flagship yggdrasil and their less expensive multibit dacs.

@Thomas savage
Can we move a lot of this to the bifrost multibit thread or something

 

[amirm]

It seems that we face protests, complaints, etc. about measurements provided for audio devices in the forum. Instead of having every review have such debates in it, I am creating this thread so that the discussions can be in one place and not repeated.

You can still ask questions in review threads as long as it is brief and not argumentative. If it is, the discussion should be placed here.

I am going to start by moving a bunch of stuff in the recent Schiit Lyr here. So stand by as those posts are moved.

 

[garbulky]

This thread seems like a way to move dissenting views on reviews out of the review itself. Maybe that's not your intention and you are trying to clear up the signal to noise ratio for a more useful review thread. But that's what it comes off as.

 

[Thomas savage]

This thread seems like a way to move dissenting views on reviews out of the review itself. Maybe that's not your intention and you are trying to clear up the signal to noise ratio for a more useful review thread. But that's what it comes off as.

Or a way to defragment these measurement inquiries so there’s a coherent narrative in both the reviews and for those questioning the measurements ( here) .

Might work might not but no conspiracy.

 

[FrivolsListener]

So I have two DACs, both of which do great at -70 dBFS, but one falls apart at -90 dBFS and the other looks fine. Hmmm, which shall I buy?

I know the toast is burned, but I like burned toast, and don't want your golden-brown toast even if a spectral analysis shows it to be better.

What if the golden brown toast is made from stale bread, but it didn't look stale. You may have not measured everything significant.

Otherwise, we'd only need measurements, and every amplifier with a flat response and < 0.001% THD would sound exactly the same, right?

 

[Thomas savage]

What if the golden brown toast is made from stale bread, but it didn't look stale. You may have not measured everything significant.

Otherwise, we'd only need measurements, and every amplifier with a flat response and < 0.001% THD would sound exactly the same, right?

If it were stale I’d of made bread and butter pudding out of it , though amir insists i use french loafs for his..

Maybe you’d prefer brioche buns for yours but I have my standards so you can make your bloody own if that’s the case.