Master Thread: Are measurements Everything or Nothing?

[Vintagear73]

I would say that any good measuring DAC can be considered “bad sounding”. Remember, “bad sounding” admits a subjective meaning: If your benchmark is a vacuum tube output stage DAC and you like it, then any perfectly measuring DAC sounds bad in comparison.

Surely.
Each of us has some equipment that he prefers more than another and there is certainly a lot of subjectivity. But I have never heard anyone buy a piece of equipment because it has a THD of 0.00000000000000% instead of 0.0000000000002%, or a 112db dinar instead of 111db.

 

[Labjr]

Depends how the specifications are measured. If it measures virtually perfect with Audio Precision, it's gonna output reference quality signal. IMO, it's best to start with a reference quality source that outputs exactly what it's supposed to. I think competent engineering is most important with digital source because bad digital can have all kinds of terrible artifacts. Color the sound to you liking further down the line.

 

[tvrgeek]

Yes, modern measurements tell everything.

I suggest they CAN. But what to measure?

I think you are saying here is…that in order to maximize the quality from a high end tweeter, you need to increase the speed of the amplifier feedback loop by moving it to higher order to achieve superior control of the tweeter voice coil.

I’ve not seen what you did…am I interpreting correctly?

Not what I found. It is the spread of the upper harmonics. DP distortion decreases with frequency as it is basically just a LP filter. Miller is more even, so above 20K, it will have more harmonics getting through that can excite the breakup. This causes IM that we hear down in the critical range. This is one of the reasons I advocate the tweeter crossover be a band pass. Roll it off steeper above 17 or 18 so those harmonics never reach it. The advantage of Miller, or two pole, is a lower level all the way through the normal pass band. So if you can avoid the top end problem, you gain across the board.

Appropriate modifications for speed and stability need to be made. Cost me two sets of MOSFETS to get it right. The critical speed here is that they blow up faster than you can spring to the power switch.

I did not invent this, I was prodded along the path of confirmation because a preeminent amplifier designer was upset my wife hated his pride and joys preferring Rotel. The question was why? After I understood it, I went on and built better speakers with less sensitivity to the problem and then, the "better" amp was in fact picked by my wife as much better. I still run my modified ( extensive) amp and it is clean enough we can hear the differences in DACs.

My assumption, The designer of the Rotel ( Dave I believe) and the other entry amp ( Erno) expected them to be used on more affordable speakers. So they chose the topology appropriately. John expected his amps to be used on serious upscale speakers, so chose alternatively. Smart folks all around. I just followed along soldering iron in hand to learn what they knew. There is no difference is production cost, so it is an engineering decision, not market pressure. It is possible, and I have modeled it in Spice, that more complex IPS and VAS will have lower inherent distortion than DP compensation on a better amplifier can beat Miller on the simple one. I do not use a Darlington differential VAS. More can me done to my CM and CCS. There seems to be magical sweet spots for the ratio of IPS gain and feedback you can adjust by the LTP degen resistors. Cordell and Self did not discuss this.

A perfect amp with a superior source, or some point where it becomes irreverent, would not have this limitation. I know amps like the Benchmark, PS5 ,and C298 are so clean, the subjectivists have nothing to say, and the objective measures say they are fantastic. They are at probably 100 times less distortion than my amp or the prestige amp I had. Good enough? I wish I had the cash to find out. What I do have is enough to do another speaker build as (some) tweeters have gotten better. Can I build a speaker that is good enough that DACs sound the same? Don't know. But now I have a system with no glaring deficiencies in the electronics, I can concentrate on my speakers. I have hauled them around a couple of times and they hold up pretty well to the 3 to 5K competition. A tiny less detail you think is the lower mids, but I suspect more of a miss in phase across the crossover. Correct at crossover, but divergence is too fast. My guess. I think I can do better.

Distortion is additive. It's not the weakest link, it is the sum total of the chain.

 

[DWI]

You mentioned all the answers in the post already, at least in my point view.

It’s basically all about money and bragging rights, but at least on this forum we brag about real feats of engineering instead of bragging about who has the product with the best marketing material.

This pretty much sums up the objectivist critique of the so-called "subjectivists".

Because objectivists either exclude or fail to understand other people's value judgements, they accuse them of being naive, vain, egotistical and materialistic, with a bit of money-envy thrown in.

When I bought a music server/streamer, I did so purely on specifications. It's hidden away, all I care about is what it does. I didn't ask or tell the wife, she doesn't care. She actually uses it, because she uses Roon, without knowing she does.

When we bought speakers, unless she liked the look of them, we weren't going to buy them. That hurdle had to be crossed before sound quality or measurements ever became an issue. It really does not matter if Genelec measure best, they fail what is the most fundamental test for most consumers of consumer audio - they are damned ugly and most people, and certainly their partners, would not have them in the living space of their homes.

When this decision was made it had nothing to do with brands, money or bragging rights. My wife knows nothing about audio brands, has never looked at audio marketing material, and the various products don't have price tags in the audio store. In fact her choice was made before asking the price.

As for bragging, strange as it may seem, we purchase audio to listen to music, not to brag. In any event, I would first need friends who've heard of Wilson Audio, and I don't have any.

It may be that some people buy audio products based on brand name and bragging rights, if they have audiophile friends. If that is how they perceive value from consumer audio and it makes them happy, then I say good luck to them.

 

[Geert]

Because objectivists either exclude or fail to understand other people's value judgements, they accuse them of being naive, vain, egotistical and materialistic,

It's you demonstrating serious lack of understanding. Do you really believe the more objective orientated people on this forum spend their money on the best measuring equipment without considering it has the features they need and it fits in their interior?

 

[Plcamp]

I suggest they CAN. But what to measure?

Not what I found. It is the spread of the upper harmonics. DP distortion decreases with frequency as it is basically just a LP filter. Miller is more even, so above 20K, it will have more harmonics getting through that can excite the breakup. This causes IM that we hear down in the critical range. This is one of the reasons I advocate the tweeter crossover be a band pass. Roll it off steeper above 17 or 18 so those harmonics never reach it. The advantage of Miller, or two pole, is a lower level all the way through the normal pass band. So if you can avoid the top end problem, you gain across the board.

Appropriate modifications for speed and stability need to be made. Cost me two sets of MOSFETS to get it right. The critical speed here is that they blow up faster than you can spring to the power switch.

I did not invent this, I was prodded along the path of confirmation because a preeminent amplifier designer was upset my wife hated his pride and joys preferring Rotel. The question was why? After I understood it, I went on and built better speakers with less sensitivity to the problem and then, the "better" amp was in fact picked by my wife as much better. I still run my modified ( extensive) amp and it is clean enough we can hear the differences in DACs.

My assumption, The designer of the Rotel ( Dave I believe) and the other entry amp ( Erno) expected them to be used on more affordable speakers. So they chose the topology appropriately. John expected his amps to be used on serious upscale speakers, so chose alternatively. Smart folks all around. I just followed along soldering iron in hand to learn what they knew. There is no difference is production cost, so it is an engineering decision, not market pressure. It is possible, and I have modeled it in Spice, that more complex IPS and VAS will have lower inherent distortion than DP compensation on a better amplifier can beat Miller on the simple one. I do not use a Darlington differential VAS. More can me done to my CM and CCS. There seems to be magical sweet spots for the ratio of IPS gain and feedback you can adjust by the LTP degen resistors. Cordell and Self did not discuss this.

A perfect amp with a superior source, or some point where it becomes irreverent, would not have this limitation. I know amps like the Benchmark, PS5 ,and C298 are so clean, the subjectivists have nothing to say, and the objective measures say they are fantastic. They are at probably 100 times less distortion than my amp or the prestige amp I had. Good enough? I wish I had the cash to find out. What I do have is enough to do another speaker build as (some) tweeters have gotten better. Can I build a speaker that is good enough that DACs sound the same? Don't know. But now I have a system with no glaring deficiencies in the electronics, I can concentrate on my speakers. I have hauled them around a couple of times and they hold up pretty well to the 3 to 5K competition. A tiny less detail you think is the lower mids, but I suspect more of a miss in phase across the crossover. Correct at crossover, but divergence is too fast. My guess. I think I can do better.

Distortion is additive. It's not the weakest link, it is the sum total of the chain.

Thanks for this…I appreciate the new (for me) info. I am learning muchly this week!

 

[audio2design]

So, this thread has degenerated into the total non-believers of any difference and those who DO wonder what we are not measuring or interpreting because we do hear differences. So much for trying to move forward in measurements. If I can hear a difference, it is measurable. SNAID is a great wheat from chaff first pass as products that measure terrible in that regard usually sound terrible. The question is in the top half. Decent measuring to superb. They don't exactly track.

I would suspect, there is a threshold of parameters we are not testing or interpreting that beyond no one can hear a difference. That would be the holy grail line in the sand. It would also probably put a lot of EGO brands out of business. From a Benchmark up to kazillion-dollar monoblocs sitting on platinum spikes, is there a difference? I am not sure. I can't afford the Benchmark, so I am not there yet.

Maybe because the believers are yet to offer any evidence of a controlled blind test complete with a test description to validate the method and yes it matters.

Trainer listeners who know what artifacts may be there and how to identify them can out listen the most golden eared audiophile in blind tests by the way.

Do you actually even read reviews? They don't look at just SINAD. You are showing your colors. Multitone IMD, linearity, SINAD versus frequency, extended frequency response beyond 20KHz, jitter susceptibility, etc. are all there. Then there are potential system level interface issues if you have ground loops / power noise coupling.

 

[audio2design]

https://josephcrowe.com/blogs/news/distortion-and-sound-quality-what-is-reasonable might be a paper that should change how distortion of speakers is measured?

This is all pretty much known, hence why distortion measurements at multiple increasing SPLs swept by frequency but I like to see more IMD stimulus to induce Doppler.

 

[audio2design]

I suggest they CAN. But what to measure?

Not what I found. It is the spread of the upper harmonics. DP distortion decreases with frequency as it is basically just a LP filter. Miller is more even, so above 20K, it will have more harmonics getting through that can excite the breakup. This causes IM that we hear down in the critical range. This is one of the reasons I advocate the tweeter crossover be a band pass. Roll it off steeper above 17 or 18 so those harmonics never reach it. The advantage of Miller, or two pole, is a lower level all the way through the normal pass band. So if you can avoid the top end problem, you gain across the board.

May I suggest trying less to impress with poorly at times used acronyms and short forms not to mention imprecise descriptions? I.e. cone breakup can excite already existing resonances in the audible band, and new resonances above the audible band that modulate with frequencies in the audible and inaudible band to create new audible frequencies. Rolling off the tweeter above the audible band avoids this.

 

[voodooless]

If all you can do is claim I am wrong, then you are not helping the further of objective testing.

No, I’m not making any claims, I ask you to back up yours. All you do is demand whatever measurements back up those claims. That is not how it works.

It’s like you claiming Leprechaun exists, and then asking us to disprove it.

So once more: you make the claim, you provide the evidence. This does not start out with measurements.

 

[audio2design]

And what about that? Total disappointment. Harsh, one-dimensionality, in short - bad. Worst then 1st or 2nd generation of Marantz/Philips CD players.

I'm sorry, but the specifications on paper are not crucial to me. To theorists for sure, but in practice they can only be a guide that can interest me in this product in the sense that I go to hear it in the best possible conditions.

30 years later people are still calling digital "harsh" and lifeless. Is it? No. The harshness is from their existing vinyle system having an actual warm frequency response with poorly matched cartridge / amp emphasizing bass, maybe rolling off highs and hence anything introduced that is flat comes across harsh. Then you have issues with the underlying CD mix not limited by the need to suppress dynamics stressing speakers more. Add in lack of euphonic harmonics perceived as "air or space" and some crosstalk to help center an image in a room that is an acoustic mess ... and the end result is the CD is called harsh, flat, etc. when the real issue is not the CD/digital at all.

30+ years of acclimatization, coupled with 30 years of speaker development and perhaps most critical 30 years of hearing loss and now digital is accepted by all but the most harsh former critics. There have been some advances in digital too of course.

 

[Spkrdctr]

Right.

So where are the 50? Where is the one?

In heaven? Sure not here on Earth.

 

[Deleted member 23982]

and the end result is the CD is called harsh, flat, etc. when the real issue is not the CD/digital at all.

i also noticed "digital" sq without ever listening to tubeamps/vinyls x) you are talking about a "whole" other group, "retro-listeners" i would call them


well i was googling a little, because i got curious how multitones are actually produced with just "one" sinewave, kinda interesting topic and it got me thinking
"noise"... how is it embedded in the sinewave we see/hear? or other phrase, how will actual noise influence the "main" sinewave? could it be that noise has some influence of the perception of the maintone, even at points where we cant set noise apart from the main signal? (and measurements just working fine)
curious what you think, this would actually explain a lot, even on a scientific way
it would also explain powercables/usb filters etc effects, because even a tiny bit noisereduction could make a difference
it would also explain why more "precise" speaker (studio monitors for example) show this way more obvious, because they are "so" precise that they also influenced more by it

also, the thing we cant measure is: the actual sound-sinewave in the air (like the actual sinewave "midair") we just can measure it through a mic again and i really think our ears/brain are to specific things way more sensitive then mics that essentially just measure "volumes" (because "altered" soundwaves could hit the mic and the mic is essetially generating a sinewave by its own and maybe "filtering" things that way?

 

[audio2design]

If noise pushes a signal above a threshold it could make a previously inaudible signal audible. That works in the lab. In music not so much.

 

[Digby]

This ^^^
Those who understand the scientific method (or methodology) trust the scientific method.
Those who do not understand the scientific method (or methodology) do not trust the scientific method.
Ergo: if you don't trust the scientific method, it's because you don't understand the scientific method.

It's not a matter of belief, opinion, or preference. It's a matter of understanding. Jim Taylor

Isn't that something of an oversimplification though, in that all science will start with some anecdotal insight, that then leads to a hypothesis, then experimentation and so on.

Every scientific fact starts its life as a baseless assertion, until proven to stand up to overwhelming experimentation designed to prove it false, at which point it becomes a fact; maybe only a temporary fact, as new information comes to light or new insights are made, that renders it unsustainable. Then it might not be a fact any more.

I don't think it is a case of trusting, or not, the scientific method, but whether the science is complete. There may well be a number of things an individual can notice anecdotally, that if tested may turn up to be scientifically factual.

Those who make claims about speakers sounding different, for instance, and that the measurements may not fully account for this difference, are not by necessity the same brand of people saying that different cables have different 'sounds'.

If the science isn't complete, then anecdotes/assertions in an area where there could be something as yet unknown are not without value (but yes, they may have more or less value depending on a number of factors).

 

[tvrgeek]

May I suggest trying less to impress with poorly at times used acronyms and short forms not to mention imprecise descriptions? I.e. cone breakup can excite already existing resonances in the audible band, and new resonances above the audible band that modulate with frequencies in the audible and inaudible band to create new audible frequencies.

I was answering a question with the facts as I learned them. I am not writing a book on how to design an amp. Self and Cordell already did that very well.

 

[audio2design]

I was answering a question with the facts as I learned them. I am not writing a book on how to design an amp. Self and Cordell already did that very well.

I will go out on a limb and say the person you responded to didn't understand even half the acronyms. I have designed electronics, amps, etc formally for decades and I was unsure of what acronyms you were using at times and some details because of that seemed wrong. Like I said it came across as bafflegab.

 

[tvrgeek]

When class D amps came out, I saw the specs and rushed to hear one. It was terrible. (B&O) I heard several others. ( Rotel, Peach) All terrible. Some years have passed and by both objective measurements and by subjective descriptions, or significantly the failure to provide a description, it could be they have arrived. I am actually considering one where a year ago I would have told anyone to stuff it. I toss out to the measuring experience, from the last to the current generation, what objective measures have changed that might let us know the dividing line from "getting there" but still with a sonic signature, to "arrived" with no discernable signature. What can we put a number on, or can we?

This does risk the wrath of EGO PRESTIGE market if a $350 Topping can sound as nice as a $7000 prestige name amp of equivalent market? (Apartment relaxed listening, what most people actually need) Can a sub $2000 Purify or Hypex implementation sound as clean as Privetta? ( $2.2 million. yes dollars) I would wager a pint it does. What differs in the better module implementations that some may have no sonic signature, and some still do? Can we identify it and measure it? That should be the focus here. How to reach "measurement is everything" not trying to prove anyone who dares to listen to music is wrong.

 

[tvrgeek]

I will go out on a limb and say the person you responded to didn't understand even half the acronyms. I have designed electronics, amps, etc formally for decades and I was unsure of what acronyms you were using at times and some details because of that seemed wrong. Like I said it came across as bafflegab.

What would you like to know? I will gladly define them.

 

[audio2design]

What would you like to know? I will gladly define them.

I would say you should define them for the person you responded to who looking at their other posts had no clue what you were talking about. I guarantee there are many of us here who probably are more knowledgeable about amplifier design who were reading that and going WTF? When you add in missing words and wrong words (IE irreverent where irrelevant should have been used) , it becomes unreadable especially when you are using acronyms for which there are no real standards or discussing in the context where multiple things are in play (speakers and amps per se) and an acronym could have multiple meanings.

It is possible, and I have modeled it in Spice, that more complex IPS and VAS will have lower inherent distortion than DP compensation on a better amplifier can beat Miller on the simple one. I do not use a Darlington differential VAS. More can me done to my CM and CCS. There seems to be magical sweet spots for the ratio of IPS gain and feedback you can adjust by the LTP degen resistors.

I am going to save this in case I ever teach a course again. This tends towards the baffle them with BS category especially with the miswording and grammar errors that destroy readability.