[GRUMPYOLDGUY]

  1. If they're getting 40dB error, then they need new test equipment.
 
2, The audibility of something is irrelevant when characterizing the hardware. The hardware is the hardware, period.
 
3. And I doubt if I comb through these forums I will find a single post that says "Oh boy, 10dB noise figure? Man I'm glad they used an A weighted measurement, that's so much more relatable to me now!! Praise the sun god!"
 
1. If your profession was audio engineering you'd know that every single audio performance measurement system from the multi-kilo$ AP right down to the free REW includes several standardized weighting filters.  Every SPL meter worth anything over $10 has weighting filters.  No new equipment needed.  It's a question of relevance.  And correctly applied weighting in noise measurements is very, very relevant. 
 
2. Not if it's audio hardware.  Sorry, that's completely naive.  In audio equipment the measurement "holy grail" is audible correlation of measured parameters.  In fact, it's about all that matters.
 
3. You won't.  Because there's no such thing as a 10dB noise figure.  All noise figures must reference something like full output, a specific sound pressure, even thermal noise in a stated resistor, there MUST be a reference, and the noise figure is a ratio.  
 
What you wouldn't understand unless you've done it is, you can measure two devices unweighted, they measure very close to, or exactly the same noise level, but one sounds much quieter than the other.  So, what do you do?  Publish unweighted and misleading noise specs?  That's what weighted measurements attempt to reconcile.  To understand why one device (or acoustic space) sounds noisier than another you must...absolutely MUST...take hearing response into account.  Every single audio engineer knows this, and applies it.  You can take hearing response into account with fixed weighting filters, or plotted curves and spectrum analysis, but you have to do it or you aren't engineering anything.  

 
1. Just because they include features to alter the actual measurement, doesn't mean they should. By the way, I was addressing your point that if you don't use weighting you somehow end up with a 40dB error... The implication of needing weighting to correct a 40dB error in the raw measurement is that your test equipment is completely broken. 40dB is a lot. A LOT. 
 
2. This is absolutely ridiculous... If the dynamic range measures 120dB, then the dynamic range is 120dB, full stop. This is the measurement of the hardware. If you change that value because the ear behaves differently than your test equipment, you're no longer characterizing the hardware, you're characterizing a system that includes the hardware and some industry organization's definition of the standard ear. This is dumb. When I look at hardware specs, I want the damn hardware specs.

 

[Brooko]

 
1. Just because they include features to alter the actual measurement, doesn't mean they should. By the way, I was addressing your point that if you don't use weighting you somehow end up with a 40dB error... The implication of needing weighting to correct a 40dB error in the raw measurement is that your test equipment is completely broken. 40dB is a lot. A LOT. 
 

 
Aren't you being a little disingenuous here - and arguing for the sake of arguing.  Anyone with a little common sense can see that he is referring to what you can hear at the ear (and there is the 40dB difference).  It has nothing to do with the measuring equipment's accuracy.
 
The A-weighted is compensated to adjust for what is heard at the ear.  And even though you may not like it - it is considered to be a standard.  And in many cases is a lot more useful than raw data. I use it a lot.  And normal people can understand it a lot more easily than if I gave them the raw uncompensated data.
 
For your uses it may not reflect what YOU want. But the assertion it is dumb simply because you don't find it useful for yourself is perhaps a little narcissistic - don't you think? 

 

[GRUMPYOLDGUY]

Aren't you being a little disingenuous here - and arguing for the sake of arguing.  Anyone with a little common sense can see that he is referring to what you can hear at the ear (and there is the 40dB difference).  It has nothing to do with the measuring equipment's accuracy.

The A-weighted is compensated to adjust for what is heard at the ear.  And even though you may not like it - it is considered to be a standard.  And in many cases is a lot more useful than raw data. I use it a lot.  And normal people can understand it a lot more easily than if I gave them the raw uncompensated data.

For your uses it may not reflect what YOU want. But the assertion it is dumb simply because you don't find it useful for yourself is perhaps a little narcissistic - don't you think? 

No, that's not what I got from the excerpt he posted which said C weighting can result in 40dB error compared to A weighting. And that C weighting was relatively flat.

My point is that it's an incredibly dumb standard that doesn't exist for products higher up in the spectrum. When I report SNR or dynamic range measurements to our customer, it better be the real number, and not "weighted"... otherwise, as I said, the company I work for would be blacklisted by that customer. And this makes complete sense, because they want to know the capabilities of the hardware, period.

What I think is naive is that someone would actually believe that if you told the average consumer that the dynamic range was a 126dB A weighted that it would be of anymore significance or understandable to them then if you told them it's 120dB without any weighting.

The only conclusion one can reasonably come to regarding the purpose of weighting is to give manufacturers the ability to hide shortcomings in their designs.

 

[Brooko]

But which is more useful to the average consumer - number relative to what they hear, or the raw data. Assuming that because you want a figure a certain way does not mean everyone does. And both report the same data. As long as it is stated as A-weighted, I fail to see the problem.

 

[GRUMPYOLDGUY]

But which is more useful to the average consumer - number relative to what they hear, or the raw data. Assuming that because you want a figure a certain way does not mean everyone does. And both report the same data. As long as it is stated as A-weighted, I fail to see the problem.

126dB dynamic range is no more relatable than 120dB. The problem is that I'm making the 120dB figure up... I have no idea what the real dynamic range of this hardware is because the manufacturer has opted to only publish this weighted figure. I think if you're completely honest with yourself, you'll also think that they do so not because this figure is somehow more relatable to the average consumer*, but because it allows them to publish a number that is higher than it really is.

*Because it's not more relatable to the average consumer IMO. It's not like they don't understand what a dynamic range of 120dB is, but if you tell them it's 126dB A weighted they magically understand it. I just had to explain to someone that 0dBV is not the same thing as a muted output because dB is a log scale. It blows my mind that anyone thinks a weighted measurement is somehow makes the figure understandable or useful to a layperson.

 

[pinnahertz]

   
1. Just because they include features to alter the actual measurement, doesn't mean they should. By the way, I was addressing your point that if you don't use weighting you somehow end up with a 40dB error... The implication of needing weighting to correct a 40dB error in the raw measurement is that your test equipment is completely broken. 40dB is a lot. A LOT. 
 
2. This is absolutely ridiculous... If the dynamic range measures 120dB, then the dynamic range is 120dB, full stop. This is the measurement of the hardware. If you change that value because the ear behaves differently than your test equipment, you're no longer characterizing the hardware, you're characterizing a system that includes the hardware and some industry organization's definition of the standard ear. This is dumb. When I look at hardware specs, I want the damn hardware specs.

1.  Yeah, I get it.  Everyone is wrong...but you, the guy who doesn't work in audio.  Righto. I posted that quote from an instrumentation web site, and yes, they are correct.  You can be off by 40dB measuring acoustic noise unweighted.  Subsonic noise is the key.  It's huge, and inaudible.  
 
2. Dynamic range and noise floor aren't the same thing, and that might be the start of your misunderstanding.  You still don't seem to understand the purpose of weighting.  It isn't to characterize something like dynamic range, it's purpose is to characterise the audibility of a noise floor.  Two very different things.  Both have their purpose, both are valid, neither is "dumb".  Unweighted hardware specs won't tell you how the noise floor sounds.  

 

[GRUMPYOLDGUY]

1.  Yeah, I get it.  Everyone is wrong...but you, the guy who doesn't work in audio.  Righto. I posted that quote from an instrumentation web site, and yes, they are correct.  You can be off by 40dB measuring acoustic noise unweighted.  Subsonic noise is the key.  It's huge, and inaudible.  

2. Dynamic range and noise floor aren't the same thing, and that might be the start of your misunderstanding.  You still don't seem to understand the purpose of weighting.  It isn't to characterize something like dynamic range, it's purpose is to characterise the audibility of a noise floor.  Two very different things.  Both have their purpose, both are valid, neither is "dumb".  Unweighted hardware specs won't tell you how the noise floor sounds.  

If weighting is not used for dynamic range measurements, please explain to me why manufacturers publish dynamic range measurements with A weighting.

 

[pinnahertz]

Lemme see if this works.  Here's a weighting demo FLAC file.  It has two noise signals on it, 8 seconds each.  Play this file at low volume, it should be audible, but quietly.  Listen to both signals as many times as you like.
 
Both measure within 1dB of each other with a unweighted flat meter.  Do they sound like they are the the same loudness?
 
The first signal also has more than 10dB higher peak energy than the second.  Does it sound that way?
 
With an A-weighting filter, the first signal measures more than 14dB lower than the second.  
 
Which measurement weighting represents the audible result better?
 
weighting demo flac

 

[GRUMPYOLDGUY]

Lemme see if this works.  Here's a weighting demo FLAC file.  It has two noise signals on it, 8 seconds each.  Play this file at low volume, it should be audible, but quietly.  Listen to both signals as many times as you like.

Both measure within 1dB of each other with a unweighted flat meter.  Do they sound like they are the the same loudness?

The first signal also has more than 10dB higher peak energy than the second.  Does it sound that way?

With an A-weighting filter, the first signal measures more than 14dB lower than the second.  

Which measurement weighting represents the audible result better?

weighting demo flac

I downloaded it, will try it out tomorrow evening and report back. Thanks for putting it together.

 

[pinnahertz]

If weighting is not used for dynamic range measurements, please explain to me why manufacturers publish dynamic range measurements with A weighting.

Sure.  DR and S/N are often confused, or considered to be identical.  It's a fine point, but technically, they aren't.  You can, for example, record signal below noise and consider that part of the total dynamic range.  Total DR in a digital system assumes the reference is 0dBFS.  When specific DR meters are used to evaluate recorded signals, no weighting (other than that required for a DR meter) is used.  But those readings use a dynamic reference related to the content, as the result is a differential.
 
S/N Ratio requires a reference, but may not always be a digital value, and sometimes is not stated (that's a big problem!).  For example, S/N Ratio of a power amp will reference a particular power output level or a level at which a certain amount of distortion occurs.  The total DR of that amp will be greater than S/N it the S/N reference is something other than clipping.  

 

[castleofargh]

I must be missing something here, if it's about measurement of the device, it's done under certain conditions and that gives a result.  at no point it's a matter of weighting being good or bad IMO.
the problem is that specs for consumers come too often with nomenclatures as if written by a 2 year old dog that doesn't know what it's doing(AKA marketing guys). and what's worst, some of the mistakes/omissions are made on purpose.
 
I believe Grumpy's grumpiness comes from that, because that's the origin of mine too. ^_^

 

[pinnahertz]

  I must be missing something here, if it's about measurement of the device, it's done under certain conditions and that gives a result.  at no point it's a matter of weighting being good or bad IMO.
the problem is that specs for consumers come too often with nomenclatures as if written by a 2 year old dog that doesn't know what it's doing(AKA marketing guys). and what's worst, some of the mistakes/omissions are made on purpose.
 
I believe Grumpy's grumpiness comes from that, because that's the origin of mine too. ^_^

No doubt.  But let's be clear that misleading or ambiguous specifications and well-research measurement technology are two different things.  One may be the misuse or incomplete use of the other.  That doesn't make a measurement technology like noise weighting curves nonsense, or a deliberate attempt to inflate numbers.  I get just as grumpy about the way devices are specified as anyone. Can't even count the number of times I've tried to verify manufacturers specs with real testing, only to discover just how much they've distorted reality.  But I just happen to have used weighting curves for many years for their intended purpose: to better correlate a measurement with what is perceived.  It's not the weighting curve that lies or inflates the figures, it's how specifications are written...no reference, no test conditions, etc.  Makes the spec meaningless, and impossible to verify. 
 
The problem we have with noise specifications is as soon as you state them they are used for comparison with other devices. Because hearing is sensitive to spectrum, the only way to properly do a noise measurement is to display the full spectrum, but that's a graph, and consumers don't understand graphs.  Graphs are also hard to standardize and compare, and take way too much interpretation.  So we dumb it down to a single figure, and try to make it relevant with weighting.  Actually, noise is the easy one.  Distortion specs are a total nightmare, almost none have any real meaning when it comes to audibility until you get to pretty big numbers. Again, that doesn't invalidate the measurement technique, the problem lies in how to present the data so somebody can make even a little sense out of it. 
 
Of course marketing lies.  That's true in every product, not just audio.  But the same measurement technology can be used to expose a lie as was used in creating the spec, it's just a question of application and truthfulness. 

 

[Arpiben]

  No doubt.  But let's be clear that misleading or ambiguous specifications and well-research measurement technology are two different things.  One may be the misuse or incomplete use of the other.  That doesn't make a measurement technology like noise weighting curves nonsense, or a deliberate attempt to inflate numbers.  I get just as grumpy about the way devices are specified as anyone. Can't even count the number of times I've tried to verify manufacturers specs with real testing, only to discover just how much they've distorted reality.  But I just happen to have used weighting curves for many years for their intended purpose: to better correlate a measurement with what is perceived.  It's not the weighting curve that lies or inflates the figures, it's how specifications are written...no reference, no test conditions, etc.  Makes the spec meaningless, and impossible to verify. 
 
The problem we have with noise specifications is as soon as you state them they are used for comparison with other devices. Because hearing is sensitive to spectrum, the only way to properly do a noise measurement is to display the full spectrum, but that's a graph, and consumers don't understand graphs.  Graphs are also hard to standardize and compare, and take way too much interpretation.  So we dumb it down to a single figure, and try to make it relevant with weighting.  Actually, noise is the easy one.  Distortion specs are a total nightmare, almost none have any real meaning when it comes to audibility until you get to pretty big numbers. Again, that doesn't invalidate the measurement technique, the problem lies in how to present the data so somebody can make even a little sense out of it. 
 
Of course marketing lies.  That's true in every product, not just audio.  But the same measurement technology can be used to expose a lie as was used in creating the spec, it's just a question of application and truthfulness. 

 
Marketing lies and therefore technical specifications lie by omission, meaning the results are true but only under specific conditions.
Manufacturers technical advertise their products having in mind what competitors do advertise even if nonsense. As you rightly stressed it, this is true for every product (non audio included), consumer or prosumer ones.
Manufacturers may select a golden item (more restrictive specs) for publishing results. Production items will have larger specs.
But manufacturers do not use weighting curves or FFT windows for tailoring specs.Fully agreeing with your above comments.

 

[GRUMPYOLDGUY]

Sure.  DR and S/N are often confused, or considered to be identical.  It's a fine point, but technically, they aren't.  You can, for example, record signal below noise and consider that part of the total dynamic range.  Total DR in a digital system assumes the reference is 0dBFS.  When specific DR meters are used to evaluate recorded signals, no weighting (other than that required for a DR meter) is used.  But those readings use a dynamic reference related to the content, as the result is a differential.

S/N Ratio requires a reference, but may not always be a digital value, and sometimes is not stated (that's a big problem!).  For example, S/N Ratio of a power amp will reference a particular power output level or a level at which a certain amount of distortion occurs.  The total DR of that amp will be greater than S/N it the S/N reference is something other than clipping.  

Dynamic range is an SNR measurement with a full scale input (in the context of a DAC, which is the example I was going for). It sounds like we agree on this.

I don't think I saw an explanation in your post about why manufacturers report DR specs with A weighting if you say weighting is not used for DR measurements. I might have missed it though.

Still haven't had a chance to play with your test file, it's still on my todo list though.

 

[pinnahertz]

Dynamic range is an SNR measurement with a full scale input (in the context of a DAC, which is the example I was going for). It sounds like we agree on this.

 

True, but only when applied to a digital system.  There are dynamic range measurements of analog system, which have no 0dBFS (full scale) reference, or even a well defined clipping/overload point (think analog tape, as an example), hence, much of the confusion.