Master Thread: Are Measurements Everything or Nothing?

[Sal1950]

With all the blind testing that supports the notion that people can't tell the difference between 16/44 Redbook and higher resolution sound files, I'm sorta thinking that 24/96 provides all the margin for golden ears needed and therefore represents the "best possible".

Agreed, that was also the finding of Mark Waldrep at AIX records, both his personal opinion and the conclusion from 2 separate listening tests he ran from his site a few years back.

The people listening to 192K, 384KHz, 768KHz sampling hear music. So do I. But as a technical person it bothers me greatly because I see in my mind that Amplitude vs frequency plot with almost all of the information channel capacity being wasted. The consumer ends up with huge files for no good reason. As an added bonus, you need a lot more memory, with an added bonus of slower up/down load.

I see much of this in the market place, both on the professional-marketing side and consumer side, just more snake-oil. A shame more pushback isn't being heard but then neither is it heard against $500,000 monoblock amps.

 

[danadam]

Re. 384 kHz, AFAICT, at this sampling rate the hardware usually switches to slow or early roll-off filters, although I only have RME Adi-2 Pro and some dongles to verify:

I know that apparently one of the features of Adi-2/4 are filters selectable even at 768 kHz.

 

[Dan Lavry]

Archiving with good converters:

I know that a lot of people get into clipping digital peaks (and I try to help it). There is a lot of focus for “driving” the AD converter hard, it is known as the “loudness war”. But putting those cases aside, given large dynamic range and good linearity, lets drive the meter peaks to say -20dB (instead of full scale), not using the 3 most significant bits.

Say the analog input to the AD is 96dB dynamic range. The signal's noise floor is -(96+20) = -116dB. Now you need a converter with lower noise floor. A converter noise of say -126dB (20-20KHz, no weighting) is far below that level.

Of course the 24-bit result is missing some most significant beats, and a normalize operation is in order (digital multiply by constant).

What is the point? It eliminates an amplifier. You can drive the AD directly with small signal. 20dB gain is X10, so a professional peak level 12.88V rms is only 1.288V, and one can go lower.

This may not be suitable to most setups, but is very useful for archiving music, when the task is to transfer analog from different signal sources, with minimal alterations.

The conversion integral linearity is even better than driving signals to full scale. It may not be needed, but here is why (for non-mathematicians):

A flat mirror shows the image (flipped, I know), think of it s linear. A curved mirror (not flat) changes the image, it is not linear, it causes distortions. Take the mirror (the curve) and break it to small segments, each individual segment is closer to a straight line than the whole curve, thus better linearity. Lossing the top 3 bits means using only 1/8 of the range. The linearity improves, the noise stays the same.

 

[Dan Lavry]

One can, barely, construct a situation in THEORY where 50khz sampling with a 20khz bandwidth might just be indetectable, relying on fantastically overestimated ability of the listener in a dead quiet anechoic room. Note "theory", not "evidence" there is NO EVIDENCE.

Based on that, I have advocated 64khz for years. It's got "margin" in all even potentially fantastical situations, including little kids and such.

So 96 is fine, I do think, even for the rare person over 5 years old with an undamaged first millimeter in the basilar membrane. That would be true at 64khz too.

I don't have your deep understanding of the ear.
I thought that I can do a good job at around 60-70KHz, but 88KHz and 96KHz was exactly X2, so easier conversion (synchronus) to 44.1 and 48KHz (?). It really did not matter what I thought...

 

[kandamrgam]

Not sure if this point was beaten to death here, but going by measurements, is a $25 JCALLY JM20 all I need as a DAC for ultimate transparency? So no sonic advantage if I am to buy a Topping E30 ($130) over a JM20?

 

[Newman]

Not sure if this point was beaten to death here, but going by measurements, is a $25 JCALLY JM20 all I need as a DAC for ultimate transparency? So no sonic advantage if I am to buy a Topping E30 ($130) over a JM20?

If it has the right type and number of inputs and outputs for you, sure! IMO definitely no sonic advantage vis-a-vis Topping.

Just be aware that the consistency of production and reliability of ultra-cheap products are generally unknown, but the safer assumption would be to assume they are not high.

Reliability is the lesser concern, for home listening. Who cares if the first one is DOA, or you need to replace a dead unit a few times over the years, at $25 a pop!

Production consistency is the greater concern in my mind. The unit that Amir tested measured great, but who knows how many units per 1000 appear to be quite functional but actually don't perform anywhere near what Amir measured. 250? 2? 50? If you are a $25 DAC maker, you will be ordering the absolute cheapest passive and active electronics you can source, which will have the lowest quality and consistency in the market. If you are buying it to get a certain performance level, your confidence in getting that should be tempered.

Those are the factors to consider as I see them.

cheers

 

[kandamrgam]

If it has the right type and number of inputs and outputs for you, sure! IMO definitely no sonic advantage vis-a-vis Topping.

Just be aware that the consistency of production and reliability of ultra-cheap products are generally unknown, but the safer assumption would be to assume they are not high.

Reliability is the lesser concern, for home listening. Who cares if the first one is DOA, or you need to replace a dead unit a few times over the years, at $25 a pop!

Production consistency is the greater concern in my mind. The unit that Amir tested measured great, but who knows how many units per 1000 appear to be quite functional but actually don't perform anywhere near what Amir measured. 250? 2? 50? If you are a $25 DAC maker, you will be ordering the absolute cheapest passive and active electronics you can source, which will have the lowest quality and consistency in the market. If you are buying it to get a certain performance level, your confidence in getting that should be tempered.

Those are the factors to consider as I see them.

cheers

Thank you. It's amazing a tiny chip is all you need for perfect analog conversion that cannot be bested with bigger and bulkier devices!

 

[solderdude]

Technically it is easily bested but both are much better than human hearing capabilities.
This is the important part .... hearing limits.

 

[Romario]

On YouTube I watched a video explaining that the polarity of the wall wart ( + and - ) and equipment plug can be measured to make sure it follows that polarity through the power lines from the wall untill the power input. Does that make sense and has anyone applied this in their home system?

I usually plug in the stuff and if it turns on I'm ok with it. Just wondering what te benefits could be of measuring the polarity of the power chain.

 

[kyuu]

On YouTube I watched a video explaining that the polarity of the wall wart ( + and - ) and equipment plug can be measured to make sure it follows that polarity through the power lines from the wall untill the power input. Does that make sense and has anyone applied this in their home system?

I usually plug in the stuff and if it turns on I'm ok with it. Just wondering what te benefits could be of measuring the polarity of the power chain.

This is 100% not something to be concerned about. I don't know what the video could even mean by "measuring" the polarity of a wall wart or the equipment.

 

[Romario]

This is 100% not something to be concerned about. I don't know what the video could even mean by "measuring" the polarity of a wall wart or the equipment.

Yeah, sorry, by "measuring" I mean determine the polartity of the pins/plug (+ and -).

 

[JRR]

I don’t se alot of talk about the possibility that designers intentionally add distortion profiles the the sound that lower sinad but creates euphonium. I’d imagine that most companies would then attribute the flavor to some pseudo scientific engineering choice often implemented with the sole purpose of selling an idea of high end and exclusivity since directly tampering with the sound profile would be frowned upon by the general consumer.

 

[j_j]

I don’t se alot of talk about the possibility that designers intentionally add distortion profiles the the sound that lower sinad but creates euphonium. I’d imagine that most companies would then attribute the flavor to some pseudo scientific engineering choice often implemented with the sole purpose of selling an idea of high end and exclusivity since directly tampering with the sound profile would be frowned upon by the general consumer.

Oh, yes, that is a thing. Even people like me who use clean systems have experimented with things like simulators that pretend to add some LP-like distortions, which can widen the soundstage and create false sensation of exaggerated dynamic range.

 

[Sokel]

I don’t se alot of talk about the possibility that designers intentionally add distortion profiles the the sound that lower sinad but creates euphonium. I’d imagine that most companies would then attribute the flavor to some pseudo scientific engineering choice often implemented with the sole purpose of selling an idea of high end and exclusivity since directly tampering with the sound profile would be frowned upon by the general consumer.

Like that?

That comes from a THD+N champion (they call it "Sound Color" or something silly like that) which brings down its 120dB THD+N to 60's or 50's.

The funny thing about it is that, even changing them on the fly, the consensus around forums is that there's no difference with music (maybe tones too) .

 

[Nopiano]

Yeah, sorry, by "measuring" I mean determine the polartity of the pins/plug (+ and -).

It is a thing in Europe to check that the mains polarity (or phase as it is also known) is correct. Primare is one manufacturer I know that even includes a polarity ‘pen’ to check it. Here’s an example of their instructions on this.

 

[JRR]

Like that?

View attachment 520538

That comes from a THD+N champion (they call it "Sound Color" or something silly like that) which brings down its 120dB THD+N to 60's or 50's.

The funny thing about it is that, even changing them on the fly, the consensus around forums is that there's no difference with music (maybe tones too) .

I’m thinking more along the lines of something like a uber high end DAC with all sorts of extreme solutions to a non existent problem that for some reason measures worse than a $400 topping. Maybe that distortion profile is curated. The, from our standpoint, unnecessary Rube Goldberg type engineering is the facade, the classic audiophile pseudo science is the marketing, and the anti measurement rhetoric the defense.

 

[Dan Lavry]

I don’t se alot of talk about the possibility that designers intentionally add distortion profiles the the sound that lower sinad but creates euphonium. I’d imagine that most companies would then attribute the flavor to some pseudo scientific engineering choice often implemented with the sole purpose of selling an idea of high end and exclusivity since directly tampering with the sound profile would be frowned upon by the general consumer.

Years ago, I was designing a DA11 DAC with a built-in headphone amplifier. I realized that there is a conflict between the DA output connectors (signals for speakers) and the signals for the headphones.

Listening to speakers, the left ear hears some of the right speaker sound, and visa versa. But with headphones, the left ear hears only the left channel signal and vice versa…

So I had an adjustment, user settings for the image width. But I did not muck with adding distortions, that magic built-in distortions that make Rap, Mozart, country, Tango and all sound better.

Like the magic salt that adds to eggs, potatoes, salad, ice-cream?

My opinion: Converters should convert accurately. AD’s offer a file representing the input accurately. There are plenty of tools (analog and digital) to alter the sound for each performance. Conversion to digital is done for UTILITY not for sonic alteration. Digital format offers a lot (memory, internet connectivity, robust noise immunity…)

 

[JRR]

Years ago, I was designing a DA11 DAC with a built-in headphone amplifier. I realized that there is a conflict between the DA output connectors (signals for speakers) and the signals for the headphones.

Listening to speakers, the left ear hears some of the right speaker sound, and visa versa. But with headphones, the left ear hears only the left channel signal and vice versa…

So I had an adjustment, user settings for the image width. But I did not muck with adding distortions, that magic built-in distortions that make Rap, Mozart, country, Tango and all sound better.

Like the magic salt that adds to eggs, potatoes, salad, ice-cream?

My opinion: Converters should convert accurately. AD’s offer a file representing the input accurately. There are plenty of tools (analog and digital) to alter the sound for each performance. Conversion to digital is done for UTILITY not for sonic alteration. Digital format offers a lot (memory, internet connectivity, robust noise immunity…)

Salt on tomatoes… it might not measure like a tomato but good god!

 

[Sal1950]

I don’t se alot of talk about the possibility that designers intentionally add distortion profiles the the sound that lower sinad but creates euphonium.

There's a ton of talk like that around here. Just earlier this week I made this post here,

Do all amplifiers sound the same? Level matched listening test

Every manufacturer, even the most expensive or famous, has "baduns" - a model with an inherent defect not spotted during prototyping. So every year thousands of Class AB amplifiers are junked because they failed and were not "economical to repair". Do more Class D amplifiers fail? Or do Class D...

www.audiosciencereview.com

 

[bmc0]

[...] LP-like distortions, which can widen the soundstage [...]

Is that effect due to antiphase rumble?