So we now have a speaker with >5% distortion which isn't audible subjectively. Is a SINAD of 96dB poor?

[ROOSKIE]

What do people think?
My experience tells me the SNR needs to be better than -80dB for it to be inaudible (to me) listening to music and I had accepted the old limit (around for 50 years+) that better than 0.1% distortion (-60dB) was inaudible, which is way worse than easily achieved in electronics these days.
Maybe I am wrong, I knew speakers were the limiting factor in distortion but never thought such a high level would be subjectively inaudible, and it was even in the frequency range where most of the music resides.
I have always been sensitive to colouration in speakers (I like to think ) but had mainly attributed it to poor cabinet design or drive unit resonance.
This has left me a bit confused, on an objective front.
I had always felt that the performance available from a lot of electronics was way beyond anything needed for SQ - more a measurement of good engineering.
Is there a particular aspect of the distortion of a speaker which may make it inaudible whilst in electronics it would be? If so what?

Found this online blind test
http://archimago.blogspot.com/2020/01/internet-blind-test-is-high-harmonic.html?m=1

 

[Frank Dernie]

Long and short is the speaker will always be the determining distortion level of your system unless something else in the chain is broken. The electronics are going to set the noise level.

Which is one big reason I don't like SINAD for electronics. Give me distortion, and it will almost always be inaudible levels. And give me noise levels separate. They too should be inaudible and if not knowing only SINAD confuses the issue. You can have poorly gain-staged systems that will have audible noise without any distortion audible beyond what the speaker provides.

Exactly

 

[tuga]

I've done that in a few of my published reviews.

Could you post a link to those?

 

[Blujackaal]

Long and short is the speaker will always be the determining distortion level of your system unless something else in the chain is broken. The electronics are going to set the noise level.

Which is one big reason I don't like SINAD for electronics. Give me distortion, and it will almost always be inaudible levels. And give me noise levels separate. They too should be inaudible and if not knowing only SINAD confuses the issue. You can have poorly gain-staged systems that will have audible noise without any distortion audible beyond what the speaker provides.

Yeah this is why i also never understood, Why headphones/IEMs that avg 0.6 ~ 2% to be a issue either. From 0 - 80db the distortion should be <0.1 ~ 1% instead, At 90db them being 0.6 ~ 2% means nothing at that SPL. Any Death metal, Noise, Noise rock & more. They could mask the distortion at higher SPLs.

Blame the ones who consume the charts, Because they seem to have forgotten that a Speaker/Headphone will have inaudible distortion. Making the 16bit @ 44.1KHz goal kinda moot.

 

[tuga]

Yeah this is why i also never understood, Why headphones/IEMs that avg 0.6 ~ 2% to be a issue either. From 0 - 80db the distortion should be <0.1 ~ 1% instead, At 90db them being 0.6 ~ 2% means nothing at that SPL. Any Death metal, Noise, Noise rock & more. They could mask the distortion at higher SPLs.

Blame the ones who consume the charts, Because they seem to have forgotten that a Speaker/Headphone will have inaudible distortion. Making the 16bit @ 44.1KHz goal kinda moot.

The distortion audibility thresholds of electronic equipment is tested using loudspeakers and headphones.

Looking at some of the levels shown by @andreasmaaan here, I am not convinced by the arguments that:

a) loudspeaker distortion masks electronics distortion (I will avoid mentioning my anecdotal experience) thus electronic equipment engineers/designers shouldn't bother with the pursuit of ultra-low distortion (as per Benchmark, etc.)

b) loudspeaker engineers/designers shouldn't bother with the pursuit of lowering distortion below what has been determined by available thresholds of studies of distortion audibility in loudspeakers

 

[Blujackaal]

Electronics' distortion audibility thresholds are tested using loudspeakers and headphones.

Looking at some of the levels shown by @andreasmaaan here, I am not convinced by the arguments:

a) that loudspeaker distortion masks electronics distortion (I will avoid mentioning my anecdotal experience) thus engineers shouldn't bother with the pursuit of ultra-low distortion (as per Benchmark, etc.)

b) that engineers shouldn't bother with the pursuit of lowering distortion below what has been determined by available thresholds of studies of distortion audibility in loudspeakers

DAC/amps's should push for <0.1% regardless which i ment also. I was more meaning that judging the speaker/headphone sound by it distortion dosen't mean anything at all.

The 5% mark on that speaker sounds like the driver is being stressed out, If under 80db cue the bass distortion being 0.8%(guessing). Since headphones like the Shure SRH1540 it 0.4% under 80db, But >1% at 90db.

 

[SIY]

Could you post a link to those?

Print publication. But you ought to subscribe because all the cool kids do.

There's issues with doing multitone unless you have an actual anechoic chamber, which I don't- the measurements can't be gated and tend to get dominated by room noise.

 

[vavan]

Any guides for dummies to try IMD tests with Rew+umiq @home?

 

[scott wurcer]

The distortion audibility thresholds of electronic equipment is tested using loudspeakers and headphones.

Looking at some of the levels shown by @andreasmaaan here, I am not convinced by the arguments that:

If you eliminate the ones that call out high order distortion (poor/ancient class A/B amplifiers), and the so called transient IMD (nothing but poor slew rate and/or full power response) there is little left at low levels. Some of this is long out of date.

 

[andreasmaaan]

If you eliminate the ones that call out high order distortion (poor/ancient class A/B amplifiers), and the so called transient IMD (nothing but poor slew rate and/or full power response) there is little left at low levels. Some of this is long out of date.

I agree.

I would add that these studies are demonstrative of why a single distortion figure is next to meaningless.

This is one reason why I think a more valid way of inferring distortion audibility thresholds is with reference to studies of auditory masking, rather than (or at least in addition to) studies of distortion audibility per se.

 

[scott wurcer]

This is one reason why I think a more valid way of inferring distortion audibility thresholds is with reference to studies of auditory masking, rather than (or at least in addition to) studies of distortion audibility per se.

And this requires trained listeners and a lot of work to follow something like the ITU spec so not much is getting done.

 

[SIY]

Any guides for dummies to try IMD tests with Rew+umiq @home?

On speakers? Tough job with a cheap mike and no anechoic chamber.

 

[andreasmaaan]

And this requires trained listeners and a lot of work to follow something like the ITU spec so not much is getting done.

You don't think we already have enough data to draw some fairly solid conclusions?

 

[andreasmaaan]

Any guides for dummies to try IMD tests with Rew+umiq @home?

Its never going to be perfect, but if you can go outdoors somewhere quiet and take groundplane measurements, those are probably going to be the most accurate you can do without specialised equipment.

If that's not possible, you can probably manage ok results by getting the speaker as far from room boundaries as possible and then measuring semi-nearfield, say 20-30cm.

 

[tuga]

You don't think we already have enough data to draw some fairly solid conclusions?

I am not particularly impressed with the data but I still have a lot to read and learn.

I agree with @scott wurcer in regard to the need for trained listeners, and I'd add improved methodology as a possible area that needs attending.

 

[andreasmaaan]

I am not particularly impressed with the data but I still have a lot to read and learn.

I agree with @scott wurcer in regard to the need for trained listeners, and I'd add improved methodology as a possible area that needs attending.

If you can get your hands on a copy of Psychoacoustics by Zwicker and Fastl, that contains a pretty thorough chapter on auditory masking.

 

[scott wurcer]

You don't think we already have enough data to draw some fairly solid conclusions?

Only if you apply some judgement with respect to studies that simply are not relevant with modern technology, there are old time diehards that refuse to acknowledge progress.

 

[andreasmaaan]

Only if you apply some judgement with respect to studies that simply are not relevant with modern technology, there are old time diehards that refuse to acknowledge progress.

It seems here you're referring to old distortion audibility studies, is that right?

By "data", I actually meant data on auditory masking.

I think we know enough about masking to draw some pretty solid inductuve inferences about distortion audibility (quite independent of the dated studies of distortion audibility per se).

These data on masking are not beset by the same concerns regarding equipment that the distortion audibility studies I think you're referring to are.

 

[pma]

Any guides for dummies to try IMD tests with Rew+umiq @home?

I would say - it IS possible. See my measurements on tweeters under voltage and current drive and quite significant difference in odd harmonics IMD components.

http://pmacura.cz/speaker_dist3.htm

 

[AnalogSteph]

On speakers? Tough job with a cheap mike and no anechoic chamber.

Eh, I dunno. I've had pretty decent results using a t.bone SC400 cardioid condenser (which not only happens to be one of the cheapest true condenser mics you can buy but also quite flat on axis and reasonably enough in noise) at about half a meter from a reasonably free-standing speaker in my not-very-optimized living room. You don't absolutely have to use an omni, you know? (This one came about because my calibrated Monacor measurement mic has high distortion and I wanted to measure that anyway.)

The biggest issue for most folks is going to be absolute level calibration. Is there a good, cheap SPL meter that would be suitable for this task?

Another point I wanted to mention:
Keep in mind that electronics are generally processing the full audio range, while speakers are generally two or more ways for good reason. Fullranges can't get much excursion going and are more or less messy in the treble. This also means that intermodulation is generally going to be restricted as each driver only sees part of the spectrum. Might be why tweeters still seem to sound fine even with relatively high levels of 2nd-order distortion, f1+f2 is high enough that we don't care too much and f1-f2 is more or less outside their reproduction range. If so, these waveguide-y times would be increasing the performance demands in this regard.