God of SINAD vs. reality we get from most available music files

[dlaloum]

SINAD ... and cognitive dissonance (!)

A really good top notch speaker gets potentially down to 0.5% THD... aka SINAD no better than 46

Classic (and really good sounding) amps get THD to circa 0.05% aka SINAD no better than 66

Based on the speakers, the utility of SINAD as a measure is questionable!

So what, in fact, are we measuring?

Good engineering Hygiene?, Careful selection of components, and the use of SOTA components?
Manufacturing methods and choice... price based?

So you have an amp with a SINAD of 112db ... fantastic, wonderful design and engineering!
Then you run your preamp into it ... which achieves a SOTA SINAD of 120...
And your DAC source - again SOTA SINAD - 122

and we feed it into our audiophile speaker - say the highly regarded Revel F208.... SINAD equivalent based on THD @ 95db (roughly 5W as per the amp) - around 47 above 100Hz, and around 36 from 40Hz.

Above 200Hz the THD for the speakers are below measurable threshold (what is the measurement threshold in SINAD for speakers?)

So using SINAD on its own without some detail understanding of the underlying measurements that it is summarising, becomes a problematic issue, on a system wide perspective!

Aside from looking at the components from an academic sense of how good are they, relative to the SOTA, it is debatable how useful that raw figure is in terms of selecting components and setting up a cohesive synergistic system.

The THD and Noise data from the individual components are more useful in terms of determining what the end result would be - and shortlisting based on SINAD alone, is likely to exclude quite a large number of excellent components, which might in fact be superior in other areas/specifications, which might be critical to the ultimate outcome.

A device with a higher SINAD is definitely objectively "better" - but it also obfuscates the old adage that "Perfect is the enemy of Good"

(OK I will get off my soapbox....)

 

[restorer-john]

A really good top notch speaker gets potentially down to 0.5% THD... aka SINAD no better than 46

Classic (and really good sounding) amps get THD to circa 0.05% aka SINAD no better than 66

A passive speaker has no noise component, so it's just distortion.

0.05% is pretty poor for a 'classic' amplifier. At least another leading zero and they can do that full bandwidth (20-20k)

 

[amirm]

A really good top notch speaker gets potentially down to 0.5% THD... aka SINAD no better than 46

What? Where do you get that from? Here is Revel 328Be:

Measurements at 86 dBSPL hug near 0.0% in critical listening range. I heave measured actual SINAD of headphones at 80+. This is the limit of acoustic measurements btw. So don't compare them to electronic ones that are not polluted by noise and microphone non-linearities.

 

[amirm]

A device with a higher SINAD is definitely objectively "better" - but it also obfuscates the old adage that "Perfect is the enemy of Good"

Wrong adage. You can do a sloppy job of washing a dish and leaving soap and dirt on it. Or wash it properly. It doesn't cost more to do it right so there is no enemy.

 

[dlaloum]

A passive speaker has no noise component, so it's just distortion.

0.05% is pretty poor for a 'classic' amplifier. At least another leading zero and they can do that full bandwidth (20-20k)

To be fair it's a bit hard to tell with Classic amp specs, as their THD is at rated output... whereas Amir's measurements are at a much more reasonable 5W...

But having said that:
Nakamichi PA7 stasis... 0.1% THD @ rated output (SINAD 60 - given noise was at over 100db)
Quad 909 : 0.01% @ rated output (that's SINAD 80... given that noise is at 108db)
Quad 606: 0.03%@ rated output (that's SINAD 70... given that noise is at 105db)
Krell KSA250 : 0.1% @ rated output (SINAD 60 - spec SN=120db)

Still... I remain a fan of the current dumping designs... (which makes the AHB2 particularly interesting... as a close relative!)

So which classic designs are you referring to? - getting the THD below 0.01% is pretty rare!
(I picked a couple of my favourites, and the legendary Krell as a "Class A" example)

P.S. having said that I looked up another classic... Sansui B2302 - THD 0.003% SN120db - SINAD circa 90 so yes there are classic examples! (to be checked whether they tested as good as they claimed!)

 

[dlaloum]

What? Where do you get that from? Here is Revel 328Be:

Measurements at 86 dBSPL hug near 0.0% in critical listening range. I heave measured actual SINAD of headphones at 80+. This is the limit of acoustic measurements btw. So don't compare them to electronic ones that are not polluted by noise and microphone non-linearities.

Here is the F208 I referred to:

And Amir, are you trying to make my point for me? the charts you posted of the 328Be show THD peaking at around 0.6% above 50Hz...

Your own comment on one graph is how well it is performing at only 3% THD @ 30Hz... which unless I am much mistaken, calculates to a SINAD of around 30?... and yes if we ignore the bass the rest of the frequency range manages a very decent 0.2% THD.... (SINAD 53...)

 

[solderdude]

A really good top notch speaker gets potentially down to 0.5% THD... aka SINAD no better than 46

Even if that were true you have to consider harmonic spread and the fact that, in speakers, the FR is split in more than one drive unit.
With speakers most distortion is 2nd or max. 3rd harmonics and even 0.5% is not really problematic at higher levels in that case.
With electronics having 0.5% THD (at a specific level) there may well be harmonics that are much higher up than 3rd harmonics and those might well be audible in music.

So what, in fact, are we measuring?

Performance of a device at specific levels and circumstances that can give some clues about technical performance. Nothing more, nothing less.
SINAD is just a number that is easily ranked and just one generated number amongst many other generated numbers and plots.
A high value is indicative of a low noise/hum/distortion (opposite a specified level). Low noise, low hum and low distortion is desirable.

It says nothing about how the same device will perform in other conditions such as ground loops, interacting with other gear, in specific circumstances. It also says nothing about tonality (FR only in case of high SINAD) and longevity nor about functionality.

It is too bad a lot of people can not see 'past' the ranking of SINAD and give that table much more importance than it deserves and use it as an excuse to make fun of measurements and their importance. Usually by lack of understanding.

Yes, more measurements etc. would be welcome but then we would not be checking ASR for new reviews every day but at most once every week.

 

[solderdude]

Your own comment on one graph is how well it is performing at only 3% THD @ 30Hz... which unless I am much mistaken, calculates to a SINAD of around 30?.

SINAD is usually measured with a single tone at 1kHz. The F208 is below 0.1% = 60 (at 96dB SPL reference) and thus may well be 'external' noise dominated number.

 

[dlaloum]

Wrong adage. You can do a sloppy job of washing a dish and leaving soap and dirt on it. Or wash it properly. It doesn't cost more to do it right so there is no enemy.

But is it?

A Quad 606 is a great sounding amp - the circuits are relatively simple, and use easily sourced components, and can be repaired without much trouble (or expense)...

The AHB2 achieves legendary SOTA measurably "Clean" results...
But to achieve it you have a high complexity design... which reminds me of many Class D amps...

When you look at audibility thresholds - you have to ask have designs such as the AHB2 gone far beyond what is needed or reasonable for domestic audio?
And to achieve it, has simplicity and easy/simple/economical reparability been sacrificed? (I have the same concern with most Class D amps, although I use them myself!)

AVR's are now highly complex beasts, and a circuit board level fault, outside of warranty, almost invariably results in the unit being junk - there is no reasonable, economical repair option. This applies to AVR's to Class D amps, and (at a guess) to the AHB2.

To my way of thinking, these devices are like Formula 1 racing cars, capable of astounding results, but you probably don't want to drive one of these every day, the cost of an engine rebuild is prohibitive!

One of the things that I always found admirable in Peter Walkers industrial designs, was that he did take into account manufacturing, parts costs, long term maintainability - the designs were not the best that could be achieved at the time... (the Japanese did better... eg: Sansui power amps) - but they were more than good enough - with all the advantages that additionally acrue to a simpler design.

My point is, the SINAD charts go a long way beyond "good enough"... and a lot of people end up focusing on the pointy end of well beyond "good enough".

It draws us into that old "spec warfare" that bedevilled the late 70's through into the 80's - and keeps coming back periodically

 

[MattHooper]

Precisely. BUT it's interesting. If I make 6 versions of a track, each 1dB apart (intensity, not loudness, note), any comparison of two adjacent (in level) tracks will almost always favor the higher level track.

***BUT*** Comparing the highest level to the lowest level may very well (not always, of course) have the listener pick the LOWER level track.

It's not a transitive property.

I've done something like this with compression once. Recorded a small band. Did some modest compression. And a little more and a little more and a little more. The final copies were heavily compressed (though not brick wall limited). I had the musicians listen and pick a favorite. Each step of the way they picked the more compressed one. We took a few minutes to discuss some other things, then I had them listen to the first moderately compressed version against the most compressed. They all made a funny face, and asked what happened to that 2nd version because it sounded all messed up, bad, not really like what they remembered and a few other comments.

Both fascinating posts! Thanks! Reminds me of the frog boiling in a pot....

 

[Blumlein 88]

Something of a wild conjecture. Here are NRC measurements of Revel F12's I used prior to F208s.

THD+N @ 90dB, 50Hz - 10kHz (measured @ 2m)

THD+N @ 95dB, 50Hz - 10kHz (measured @ 2m)

Both are tonally pretty well balanced. The F208 is much cleaner sounding however. Yes, I know ill defined subjective description. And the F208 gets louder with less sense of anything being wrong. Maybe because of the F12 having that peak of distortion in a most bad location due to the crossover?

 

[restorer-john]

It draws us into that old "spec warfare" that bedevilled the late 70's through into the 80's - and keeps coming back periodically

It does return every now and then, but each time with a different twist and people focusing on one parameter to the exclusion of all others. With ASR it's SINAD (THD+N). We've had sheer power, load invariance, wide bandwidth, ruler flat response, S/N, residual noise, crossover distortion elimination, Class A, dynamic power, current capability etc.

Everything goes full circle eventually.

 

[solderdude]

One of the things that I always found admirable in Peter Walkers industrial designs, was that he did take into account manufacturing, parts costs, long term maintainability

One day at Transtec, some representatives of Quad and Nakamichi (in the Netherlands) were at a show hosted at Transtec.
The Japanese made fun of the 'old fashioned' design of Quad gear.
At one point Peter W took a flathead screwdriver out of his pocket. Opened up one of the power amps (really quick) and all boards were easily accessible from both size.
He turned to the Japanese guys and merely said: 'Old fashioned design ehhh.. I'd like to see you take apart one of your devices' and handed him the flathead screwdriver.

One would have to consider modern times though. As much as I like longevity it is a fact that most gear these days (due to SW and limited support/usability over the years) does not usually have a lifespan for more than a few years anyway. Today's customers are kind of used to buying new gear (even if very costly) every few years as long as they gain in functionality, usability and speed over whatever they owned before.

 

[Blumlein 88]

One of the old amps that were always very clean, and transparent sounding with anything I'd heard them powering. Yet the THD and noise specs are good not spectacular. The noticeable thing is the DC to 1.2 mhz bandwidth. The designer said the bandwidth wasn't a design goal. Very fast recovery from overload was a design goal and the wide bandwidth was a side effect of reaching that goal.

I owned one, and it is one of the few amps I wish I had back. I owned Quad ESL-63s when I had it.

SPECTRAL DMA-50 POWER AMPLIFIER

Specifications

POWER OUTPUT :

@ 8 ohms = 80 WRMS @ 4 Ohms = 120 WRMS

BRIDGED 160 WRMS 220 WRMS

OUTPUT POWER P.K. @ 10% Duty Cycle

STEREO

@ 8 ohms = 140 Watts @ 4 ohms = 200 Watts

BRIDGED 450 Watts 260 Watts

OUPUT CURRENT: Limited to 18A Peak

DISTORTION: Less than .1% from D.C. to 100 KHz typically .006% @ 80 WRMS / 8 ohms

BANDWIDTH: D.C. to 1.2 MHz - 3db

RISE TIME: Less than 300 nanoseconds

SETTLING TIME: 500 nanoseconds

SLEW RATE: 1,000 volts/microsecond

INPUT IMPEDANCE: 10 K ohms

GAIN FACTOR: 25 db stereo, 32 db bridged

INPUT SENSITIVITY: 1.2 VRMS for 80 WRMS output into 8 ohms

SIGNAL TO NOISE: 90db unweighted

INPUT / OUTPUT ISOLATION: 80db

CROSSTALK: 70db

DIMENSIONS: 2.5 H x 19 W x 14 D inches

 

[pma]

It does return every now and then, but each time with a different twist and people focusing on one parameter to the exclusion of all others. With ASR it's SINAD (THD+N).

It's a Golden calf here. Praying to the (Sinad) Golden Calf is to me no different from adoration of super cables or golden fuses.

 

[amirm]

It's a Golden calf here. Praying to the (Sinad) Golden Calf is to me no different from adoration of super cables or golden fuses.

There is little help for you then....

 

[Axo1989]

And Amir, are you trying to make my point for me? the charts you posted of the 328Be show THD peaking at around 0.6% above 50Hz...

I hadn't reckoned Amir thought the world began and ended at 1 kHz., but maybe it's true?

 

[amirm]

I hadn't reckoned Amir thought the world began and ended at 1 kHz., but maybe it's true?

Maybe I am an alien put on earth to annoy some of you:

 

[Thomas_A]

In space nobody will hear you. On earth yes.

 

[restorer-john]

One day at Transtec, some representatives of Quad and Nakamichi (in the Netherlands) were at a show hosted at Transtec.
The Japanese made fun of the 'old fashioned' design of Quad gear.
At one point Peter W took a flathead screwdriver out of his pocket. Opened up one of the power amps (really quick) and all boards were easily accessible from both size.
He turned to the Japanese guys and merely said: 'Old fashioned design ehhh.. I'd like to see you take apart one of your devices' and handed him the flathead screwdriver.

To be fair, the Japanese had moved on to (the precursor to) JIS, which was a vast improvement on the ancient slotted screw...

A slightly rigged and stacked challenge wouldn't you say? Especially as everything from 1968/9 onwards out of Japan was metric/JIS...