• 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!

Could we all be wrong about SINAD?

MakeMineVinyl

Major Contributor
Joined
Jun 5, 2020
Messages
3,558
Likes
5,871
Location
Santa Fe, NM
Marketing people and some enthusiasts love simple " more is better" metrics, or "less is better". It's not just an audio thing, I am a cycling enthusiast and know people who spend serious money shaving a few grams from their road bikes despite the fact all they're doing in adding expense and that in most cases they could lose kilos by eating a few less chips. But lightness is bestest in road bikes. Cameras and the pixel count obsession have been noted. I know cooks who obsess over hardness numbers and metallurgy of kitchen knives despite the fact they know nothing about metallurgy beyond what they read in marketing material and magazines, hardest is bestest.

In almost all cases there seems to be a fundamental difference in attitude between people who see gear as a tool to facilitate an activity (listening to music, photography, cycling, cooking etc) and those for whom the gear is a hobby in itself. Those in the first group may spend a lot on equipment and have a keen interest in gear but generally don't obsess about it and maintain a sense of perspective. The second group often seem to become increasingly detached from what the gear actually does and become fixated on metrics. And of course for some it's about status and spending. I have no issues with the idea of the gear becoming an interest in itself, but in most of these fields the gear has long past a point where these metrics really matter and the problems have been solved. But, that is not the same as saying people shouldn't think about what they buy

Take cycling, bike fit is way more important than weight for almost all use cases. How durable is it? What are you going to use it for? Does it have eyes for mudguards and racks if you want them? Are the gear ratios right for you (is single speed a good choice)? Tyre choice is hugely important. The thing is most of these things are determined by use case and preference, not weight, frame stiffness etc yet cycling magazines and marketing are obsessed with frame stiffness and weight.

I see a lot of similarities with audio. It's harder to get an amplifier or DAC that is audibly perfectly fine than examples with audibly degraded performance. Speakers and headphones are different but FR tuning is a personal preference and speaker/room set up is a bit like bike fit in being hugely important. For most people feature set, user interface, build quality, industrial design and whether or not they just like it are what matter as audible performance is fine. However there are still some things to consider. Amplifier power needs to be appropriate for the load and desired listening level, it's like bike gearing in not being a right or wrong issue but one in which the specification has to be suitable for the application/use case.

I think we are subjective people. Chasing metrics beyond the point of relevance is a subjective choice, choosing gear for its industrial design, UI or feature set is subjective. Ultimately in audio the gear really doesn't matter. I can listen to the music I love on anything and enjoy it. I think in any hobby there is a certain minimum standard that makes things easier or more enjoyable but that level is invariably way way below what most hobbyists would look at. Most of the musicians I know have very little interest in audio gear as they listen through the equipment.
Very well said (as I just got back from a long mountain bike ride)

As an electronics designer, I keep my gear fetishes in check by the fact that if I want some new gear, I have to design and build it myself. I usually end up saying I really don't want it that badly and go have a beer.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,384
Location
Seattle Area
I admit I am being a little obtuse here, but the point of this forum is good measurements. But it would be good to really show why that matters to people. How does it affect them.

How is it sloppy if it meets the use purpose of great sound for the end user?
The purpose of this forum is to find devices that are excellently engineered which in most cases don't cost more or even cost a lot less those that measure poorly. When someone produces a DAC with 10X the distortion and charges you the same or more money, their aim is not to deliver "great sound" to the end user. Their purpose was to build a me-too product but either don't have the skills to build a great DAC or is chasing some other nonsense in audio that screws up performance. No extra goodness comes from such products.

Think of two chain restaurants with identical menu. One doesn't wash its dishes as well as the other. Just because you don't get sick from less washed dishes doesn't mean there was good purpose behind delivering that to you. Same is true of electronics I test. As I noted, poor performance is produced for no good reason so why would you want to purchase such products and reward said company?
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,384
Location
Seattle Area
Don't get me wrong, I think it's good that the the overall state has advanced beyond 80db SINAD. But, how does it really affect anyone?
Easy: if said SINAD is made up of decent amount of noise, then it raises noise floor of your system and can become audible. There is no way you can remotely say "80 dB SINAD" is transparent. Statement makes no sense and can be plainly shown to be wrong.
 

GXAlan

Major Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
3,866
Likes
5,953
I apologize for the abruptness, but the theory here is absurd. If there was any sort of desirable distortion, it would be preferentially added to the recording itself and played on a system with as little alteration as possible. I can’t see a cohesive argument for why the equipment should modify the signal instead of doing so in mastering.

You would think, but let’s compare Yamaha, Kawai, and Steinway grand pianos. They are designed differently and sound different. This isn’t a Stradivarius type thing which is a mystery, but is instead represents design choices that survive ABX testing. There are piano competitions where you can choose which piano to use to suit your style.

This doesn’t have a Kawai in the mix but the first thing to decide is that there IS a difference.

If you agree there is a difference, then you have agree that there is no single “best” piano, just that there may be several very popular options, but others may prefer something else. This is the Harman Curve vs. slight V curve. Klipsch doesn’t follow the Harman curve and they are one of the few speaker companies that continues to be a viable mainstream product.

There may NOT be a single desirable formula for adding distortion in the recording and you are better off letting individual listeners add the distortion they prefer.

One goal could be to reproduce the piece as the artist intended, but that doesn’t change the point of you still having a preference. George Lucas, the artist, prefers the original Star Wars trilogy in the Special Edition form to such a degree that he does not want to release any of the classics as they were in the theater. The artists are not right or wrong, it’s just that you may have a different opinion than the artist and that’s OK. That’s the second point. We add condiments to our food, and some people choose vinyl over digital.

Is the truest experience for experiencing music obtained with the very best gear money can buy, or by pairing 70’s music with 70’s gear and 90’s music with 90’s gear? Is it OK to disagree with the artist on the “best” experience?

Look at Nelson Pass. He clearly has done well in life. Pass Labs is one of those companies that has remained stable (as least to 2021) and he even gives back to the hobbyist community by doing open source projects. There is an interview where Nelson Pass explicitly says that the only ear that matters is the customer and it turns out the right combination of distortion is beneficial to selling gear because it seems to make things sound better. Look at McIntosh. Those autoformers go against the grain of “minimal parts in the signal chain” yet it is one of the reasons they continue to sell well…

The big difference is that some solid state McIntosh measures well and Nelson Pass is upfront about adding certain types of distortion because it’s what his customers want.

—-
As much as I defend subjective listening, I also defend the value of SINAD. For the most part, you can use speakers to tune your subjective preference and even EQ/target curves. Going for a high SINAD means that you can stay transparent for as long as possible before you get to the stage where you want to manipulate things. If you think something sounds better with a lower SINAD, you just have to accept that you are preferring some sort of coloration that must be in that setup, which is perfectly OK.

Last, I will close with this article
https://www.sciencedirect.com/science/article/pii/S0385814621001346

Which talks about how hearing aids apply an EQ to boost “speech”. The IEC standard was built around Arabic, English, French, German, Mandarin, and Spanish speakers. This study assessed if Japanese would benefit from different EQ and it did. In the later part of the article, the authors explain than intelligibility is OK with standard IEC target since the Mandarin language had overlap, but that the Japanese specific EQ was perceived as being more natural and better sound quality.

That paper is actually interesting because people like to claim that as you age, your brain adapts so you don’t hear a difference. In this paper, they show that a) people lose their hearing generally along a similar curve even though SD is high b) long term speech patterns are pretty consistent over 1000 words. c) in people with age related hearing loss, there were preferences for different types of EQ based upon language (Of which included the 315Hz range) which reflected perceived improvement in sound quality and naturalness.

Based upon this paper, as your high frequency hearing gets worse, you probably would, in fact, benefit from getting brighter speakers over time. (Or adjusting your EQ)
 

garbulky

Major Contributor
Joined
Feb 14, 2018
Messages
1,510
Likes
827
Easy: if said SINAD is made up of decent amount of noise, then it raises noise floor of your system and can become audible. There is no way you can remotely say "80 dB SINAD" is transparent. Statement makes no sense and can be plainly shown to be wrong.
I feel like I can. In any use case that I've used it for, I will not hear the distortion present in an 80db SINAD. This includes for movies, music etc. etc. I am aware that you've talked about the 110db SINAD for provable transparency. But I struggle to figure out how that applies to anybody's music and movie watching.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
But I struggle to figure out how that applies to anybody's music and movie watching.

What Amir means is that when 80 SINAD is noise dominated AND you are using such a DAC directly into sensitive active speakers or a power amp with sensitive speakers AND you use the volume control on the source (PC for instance) an 80 (noise dominated) SINAD DAC will have audible noise.
That will be audible and annoying in softer passages and silences between tracks.
The 110 SINAD cannot do that as the noise level is at least 30dB lower in this particular case.

When you use the 80 SINAD DAC followed by a volume control and never listen very loud the 80 noise dominated DAC will most likely be fine to use.

That's why Amir states a higher SINAD is kind of an insurance for signal fidelity.
 

garbulky

Major Contributor
Joined
Feb 14, 2018
Messages
1,510
Likes
827
Same is true of electronics I test. As I noted, poor performance is produced for no good reason so why would you want to purchase such products and reward said company?
Hey I appreciate the detailed answer. I can understand your reasoning to a point though I remain skeptical of the importance of having low distortion under a certain amount.

That's an easy answer for me. I have plenty of reasons.
For me what I thought of the sound in real world subjective use matters the most. And yes, this is a big reason why I use my DC-1 DAC. I also thought the Gungnir Multibit dac (much poorer performance) sounded great, and so did an old multibit CD player I used. But I don't think many people are concerned about my subjective audio decisions due to its subjectivity, so I'll leave that out of the discussion.

Outside of that, the overwhelming reasons would be features, looks, build, brand, and price.
Features:
A great super useful remote! Torroid power supply. Fully balanced. TWO reasonably powerful headphone amps. Two analog resistor volume controls (one for the dac outs and one for the headphone amps). Specific inputs I wanted (AES, BNC). ASRC that was sample rate independent.
Price: It was in my reach. (Well actually it was free, but that's a different story.)
Brand and Warranty: 5 years transferrable. I've dealt with Emotiva before and their CS and warranty service has been exactly what I want. Couple that to the fact that they produce most of the other electronics I own and are satisfied with.


Compared to those things, a SINAD of top numbers versus a SINAD that doesn't affect my listening - at least not in a way I can notice, pales in comparison.

Let me take it a step further and tell you why I didn't purchase the Schiit Gungnir Multibit which I thought sounded great, for the most part. And the answer was - features. There was NO remote. The input selection was very hard to see. The other was that though it did sound different in a nice way (subjectively) I didn't think it bought enough of an improvement to warrant a buy. The Yggdrassil is significantly worse performing than the DC-1 and more expensive, but I would strongly consider its purchase when Schiit includes a remote.
 
Last edited:

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
if said SINAD is made up of decent amount of noise, then it raises noise floor of your system

Good reason to avoid SINAD (in fact THD+N, you use SINAD only for the reason to become different from mid-seventies THD+N wars, which you have re-incarnated) and to distinguish clearly between noise and distortion. Second, your 1kHz "SINAD" tells nothing about possible issues of low slew rate and low GBW. I will gladly re-post the 1kHz THD plot and CCIF IMD plot of the same circuit that demonstrates that using 1kHz THD or THD+N makes no engineering sense and indicates to lack of understanding of electronic circuits.

THD 741.PNG


CCIF 741.PNG
 

garbulky

Major Contributor
Joined
Feb 14, 2018
Messages
1,510
Likes
827
What Amir means is that when 80 SINAD is noise dominated AND you are using such a DAC directly into sensitive active speakers or a power amp with sensitive speakers AND you use the volume control on the source (PC for instance) an 80 (noise dominated) SINAD DAC will have audible noise.
That will be audible and annoying in softer passages and silences between tracks.
The 110 SINAD cannot do that as the noise level is at least 30dB lower in this particular case.

When you use the 80 SINAD DAC followed by a volume control and never listen very loud the 80 noise dominated DAC will most likely be fine to use.

That's why Amir states a higher SINAD is kind of an insurance for signal fidelity.
Thanks! I doubt that use case applies to me. The most issue I've had is a mild hiss on my headphone amp which has an audible noise floor due to the nature of its circuitry (it's actually a speaker amp - the bas-x a-100). Other than that, I haven't actually had any audible noise issues in my listening. For me, things like a mild hiss are trivial at best compared to other considerations like how much I enjoy using the product.
 

garbulky

Major Contributor
Joined
Feb 14, 2018
Messages
1,510
Likes
827
Good reason to avoid SINAD (in fact THD+N, you use SINAD only for the reason to become different from mid-seventies THD+N wars, which you have re-incarnated) and to distinguish clearly between noise and distortion. Second, your 1kHz "SINAD" tells nothing about possible issues of low slew rate and low GBW. I will gladly re-post the 1kHz THD plot and CCIF IMD plot of the same circuit that demonstrates that using 1kHz THD or THD+N makes no engineering sense and indicates to lack of understanding of electronic circuits.

View attachment 152821

View attachment 152822
Can you break this down for a layperson like me what the graphs are supposed to be showing? Thank you
 

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
Can you break this down for a layperson like me what the graphs are supposed to be showing? Thank you

First plot is a spectrum of 1kHz sine tone as used here in the reviews. Harmonic distortion is 0.0044% (or -87dB) with only second harmonic (H2) in the spectrum. This is not excellent, however good enough. No chance to be audible if it was true for the whole range of test frequencies between 20Hz and 20kHz. Second plot is a test with 2 sine tones, 19kHz and 20kHz feeding the same preamp as in case of 1kHz test. It is an intermodulation distortion test. If the DUT (device under test) had sufficient slew rate and distortion independent of frequency, we would see only 19kHz and 20kHz spectral lines and a difference tone 1kHz about -80dB under the 19kHz and 20kHz peak levels. Instead, we can see all sort of intermodulation products up to -25dB below 19 and 20kHz peak levels. This is the result of too low slew rate of the DUT which converts 19 and 20 kHz sine shape into triangle shape because it is unable to follow the original shape. In spectrum, this is seen as the intermodulation products. With your ear, you hear these intermodulation products very well, though the original 19+20kHz signal would be inaudible for most of us. As a conclusion, this is a technical disaster. It is unlikely to see uA741 opamp nowadays, however TL061 e.g. is not much better. It is a necessary condition for the test person to use test procedures that are able to reveal even the less frequent issues and not to avoid and overlook these tests with some lame excuse.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Can you break this down for a layperson like me what the graphs are supposed to be showing? Thank you

1: The frequency scale is linear, something to keep in mind.
2: The example is of a very cheap and early and not used today in audio opamp (long succeeded by something much better) uA741
This is a horrible opamp that in the old days sometimes was found in (even professional) audio gear.

It shows that a SINAD 80 (or thereabout ) signal (the plot with the 1kHz + stimulus + 2kHz (thus 2nd harm only) distortion signal. The rest is buried in the noise.
That exact same SINAD 80 circuit would show up in a SINAD test at 75 to 80 SINAD and get a 'fair' rating.
However, and that's the discussion here, it will ONLY be so at 1kHz.
The moment you use 2 signals, in this case 19kHz + 20kHz, at a similar absolute level (19 + 20kHz signals combined can momentarily reach the same peak level as the single 1kHz tone) the distortion products are not at an 80dB difference but at 25dB distance and this is potentially audible.

Note in reality this effect can still happen but will be at much lower levels. The plots were simply to illustrate that SINAD can be misleading.
That's why the multitone and IM distortion tests as well as other tests are more important than SINAD.

Another case would be an amp that is rolled off in the lows and or treble (crappy transformer or too small output caps or other special design features). It could have a great SINAD but sound dull and bass shy. So here again SINAD is no indication of good sound or design excellence.

SINAD is merely an indication of the distance between a 1kHz sinewave to unwanted signal at a specific level into a specific load in dB.
A higher SINAD could be (and mostly is) an indicator for higher signal fidelity, in specific cases the number could be a bit misleading.
Signal fidelity is not always the same as perceived sound quality but for many people it feels like an insurance policy for good sound.
 
Last edited:

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,384
Location
Seattle Area
Good reason to avoid SINAD (in fact THD+N, you use SINAD only for the reason to become different from mid-seventies THD+N wars, which you have re-incarnated) and to distinguish clearly between noise and distortion.
That distinction happens naturally since noise dominates as you climb up the highest tiers of SINAD. Here is the latest example:

1631431824814.png


See how distortion is vanishingly low at -140? For SINAD to go down to 120 dB requires noise to be added to it and a lot of it (relatively speaking).
Here is another:

1631431991298.png


Once again, noise dominates. That is the beautify of SINAD: it encapsulates both distortion and noise and the worst one comes up to the top.
 

TimF

Senior Member
Forum Donor
Joined
Dec 15, 2019
Messages
491
Likes
874
And we've grown to think skinny women with HUGE boobies are ideal and doubly so if they are much, much tooooo young. We can even learn to like whiskey, and for some to listen to Garrison Keiller for forty years telling the very same story about Mayberry Minnesota where nothing ever changes which is pretty much all of Minnesota. You are right, of course, the truth of your audio gear is in the advertising and marketing. you can never explicate good advertising, and doubly so for bad advertising. Price proves its point! the equipment manufacturers have learned to add a little more to the boobies and hips to 'tune' their gear....they've got their secrets.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,384
Location
Seattle Area
I will gladly re-post the 1kHz THD plot and CCIF IMD plot of the same circuit that demonstrates that using 1kHz THD or THD+N makes no engineering sense and indicates to lack of understanding of electronic circuits.
I suggest to not do that moving forward as no way can you justify anything in music having dual tones at highest frequencies at full amplitude. There is not a tweeter that can produce anything remotely close to those levels (and you would be long deaf before that). This is why I don't run 19+20 kHz test. I run the SMPTE version which is 60 Hz + 7 kHz with the latter at 1/4 the level. That is much more realistic and at the same time shows the distortion at low and high frequencies.
 

Sharur

Senior Member
Joined
Apr 10, 2021
Messages
476
Likes
213
Is it true that using a slow filter gives better SINAD than a sharp filter?
 

MRC01

Major Contributor
Joined
Feb 5, 2019
Messages
3,422
Likes
4,029
Location
Pacific Northwest
Is it true that using a slow filter gives better SINAD than a sharp filter?
Define "slow" filter:
Sometimes it means the filter does not fully attenuate until above Nyquist.
Sometimes it means the filter starts attenuating earlier than necessary, even if it does fully attenuate by Nyquist.
 

Sharur

Senior Member
Joined
Apr 10, 2021
Messages
476
Likes
213
Define "slow" filter:
Sometimes it means the filter does not fully attenuate until above Nyquist.
Sometimes it means the filter starts attenuating earlier than necessary, even if it does fully attenuate by Nyquist.
24 kHz vs 22.05 kHz attenuation

EDIT: Nvm I forgot good SINAD doesn't equate to good THD+N
 
Last edited:

Sharur

Senior Member
Joined
Apr 10, 2021
Messages
476
Likes
213
Also, this is a bit of a tangent, but are there audible differences between a 22.05 kHz "sharp" filter and a 24 kHz "slow" filter if they are both flat from 20 Hz to 20 kHz?
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,384
Location
Seattle Area
Is it true that using a slow filter gives better SINAD than a sharp filter?
In general it is the other way around. Look at my THD+N tests versus frequency that go up to 20 kHz:

index.php


The red diagonal line is the result of a very slow filter. The measured THD+N shoots so high it is off the charts. The high sample rate in green/brown lines eliminates the effect of the filter and shows much lower THD+N.

The reason for the rise is not that the DAC is more distorted but that the spurious components getting through due to slow filter get added into what makes up THD+N and inflates that.
 
Top Bottom