Master Thread: Are measurements Everything or Nothing?

[Alice of Old Vincennes]

Perhaps it's not dismissiveness. Perhaps you are expecting people to catch you up in a couple of posts on some bit of knowledge and insight that took them decades to learn, without committing your own effort and time to the journey.

When I left the public sector the first time, I was hired as a consultant to what was then a prestigious national consulting firm in another city. When I arrived, I was asked to hold a lunch meeting with the younger consultants there to explain my area of expertise to them. I thought, "Huh? You think I can compress my 15 years of progressive experience into a one-hour brown-bag lunch? You must think this stuff is easy."

I'm not saying experts here aren't sometimes curt. But I am saying that most of the questions I have had have already been answered before, if I was willing to read through some of those 100-page threads. And in that reading, I came to realize how weary the experts must be to feel as though they have to explain the same thing over and over again to those who don't want to do that digging and reading. It makes me more forgiving of them, to be honest, when they are abrupt.

I actually learned that before joining this forum. Back in the day, I was active in the Hardcore Bicycle Science mail list, and our favorite curmudgeon was Jobst Brandt, a mechanical engineer at HP who was the first to perform a finite-elements analysis sufficient to actually explain how the structure of a bicycle wheel works, with data to back it up. Lots of technical wannabes challenged him on topics and got singed as a result. A few pouted and sulked and never learned anything. But others became experts in their own right.

On this forum, I'm still one of the fresh-faced beginners sitting at the feet of real experts, faking it until I can make it. For a person my age, though, I believe the correct term to describe me is dilettante. Guilty as charged.

Rick "who doesn't know you, and therefore has no idea whether or to what extent you fit into this concept" Denney

Asking questions on this site should be treated respectfully. You are full of yourself.

 

[rcarlbe]

Perhaps it's not dismissiveness. Perhaps you are expecting people to catch you up in a couple of posts on some bit of knowledge and insight that took them decades to learn, without committing your own effort and time to the journey.

When I left the public sector the first time, I was hired as a consultant to what was then a prestigious national consulting firm in another city. When I arrived, I was asked to hold a lunch meeting with the younger consultants there to explain my area of expertise to them. I thought, "Huh? You think I can compress my 15 years of progressive experience into a one-hour brown-bag lunch? You must think this stuff is easy."

I'm not saying experts here aren't sometimes curt. But I am saying that most of the questions I have had have already been answered before, if I was willing to read through some of those 100-page threads. And in that reading, I came to realize how weary the experts must be to feel as though they have to explain the same thing over and over again to those who don't want to do that digging and reading. It makes me more forgiving of them, to be honest, when they are abrupt.

I actually learned that before joining this forum. Back in the day, I was active in the Hardcore Bicycle Science mail list, and our favorite curmudgeon was Jobst Brandt, a mechanical engineer at HP who was the first to perform a finite-elements analysis sufficient to actually explain how the structure of a bicycle wheel works, with data to back it up. Lots of technical wannabes challenged him on topics and got singed as a result. A few pouted and sulked and never learned anything. But others became experts in their own right.

On this forum, I'm still one of the fresh-faced beginners sitting at the feet of real experts, faking it until I can make it. For a person my age, though, I believe the correct term to describe me is dilettante. Guilty as charged.

Rick "who doesn't know you, and therefore has no idea whether or to what extent you fit into this concept" Denney

Thank you for the thoughtful response. I opened my post with the fact that I'm new to audio to be humble in this community about a topic that is complex and I have no training in, but please know that being new isn't the same as me expecting people to catch me up etc. Quite the contrary. I am 46 years old, small business owner, father, and have read about audio a lot over the last year. Yet, I do consider myself new to this. I see an issue about how poorly data is communicated and felt strongly enough to write about it. I still stand by that.

 

[rdenney]

Thank you for the thoughtful response. I opened my post with the fact that I'm new to audio to be humble in this community about a topic that is complex and I have no training in, but please know that being new isn't the same as me expecting people to catch me up etc. Quite the contrary. I am 46 years old, small business owner, father, and have read about audio a lot over the last year. Yet, I do consider myself new to this. I see an issue about how poorly data is communicated and felt strongly enough to write about it. I still stand by that.

Fair enough. My post "keep reading" was really not dismissive, but rather was encouragement for you to persist, and I still stand by that.

Rick "still persisting" Denney

 

[rdenney]

Asking questions on this site should be treated respectfully. You are full of yourself.

When an expert dismisses my question with a terse response, I could decide that I've been treated disrespectfully. But it may also be true that he thought I treated him disrespectfully, by expecting him to provide beginner training to me. Most of the time, though, humility results in helpfulness, which is advice I give but admittedly do not always follow.

Rick "to your second point: Yes, sometimes" Denney

 

[Holmz]

I wonder what percentage of ASR are subjectivists, vs other forums. I feel like ASR has more subjectivists or subjectivist activity than objectives in subjective forums.

Is a poll?

It could have percentage levels.

 

[Holmz]

Can you EQ to some curve and inverse-EQ the result so that the resulting bitstream is bit-identical to the input?

I'm not talking about a trivial case here like "lift all frequency range by 1db" but a more realistic example.

Yeah one can..
Sort of like doing and FFT and then an inverse FFT can give back the same time domain signal that we started with.

I know with symmetric FIR it works, and it needs to be causal… so one cannot take say 900 samples of a 1024 FFT and do it.
We need all 1024.

So whether it is amplitude or phase, or both… The inverse of the filter would undo its effect.]

With IIR filters, or analogue solder parts, git is more non-causal and I do not think it can be undone.
Also things like notch filters which are not something like 64-bit floating point, but rather 24 bit fixed points, then there is the concept of loosing resolution/precision that is a possible thing…

But in general that bit stream would be unchanged.

 

[Doodski]

Any useful analysis of equipment surely must start with an understanding of hearing and perception. Tell us more clearly why and how certain data matter or not. What will we hear? "Subjectivists" must agree some measurements do matter, but they should challenge those who perform measurements to explain their meaning.

I think ASR does decently at coaching peeps through the tests and charts. Are you saying you believe this ^^^or you need this?

 

[KSTR]

Of course that EQ is reversing an acoustic effect, but that’s not what my question is about.

EQ is applying a function on the data, like f(s). I’m curious if there’s a way to generate its reverse, let’s call it f’, so that f’(f(s)) := s. With Roon or whatever else you want, but it has to be possible with the existing software/hardware and it must work in real time, just like EQ.

If it’s possible, then f is not lossy. If it’s not possible, we don’t know.

It's possible with FIR provided the filter fullfils some certain contraints (#1 it must be invertible to begin with) and it depends on the exact algorithm. FFT/iFFT-based convolution which is usually used (for computational efficiency) may not always make it to 24bit precisision even with 64bit floating point. Time-domain convolution in 64bit float does it (and it does exist because I wrote code for it ;-).
Dithering is not allowed for obvious reasons. Realtime operation only depends on processing power (it's either enough or it isn't) but there always is propagation delay which cannot be reduced to zero (theoretical minimum is time in the impulse kernel until the main peak occurs).

 

[Chr1]

I agree, but I feel like many people don't really want to have that conversation. Some of the people that do are pretty rude about it and that leads people to not want to continue the discussion. Or, they use a lot of very technical jargon that are hard to understand. Ofc there are things that are hard for a lay person to judge even when things are simplified. Usually there is enough consensus in measurements, but 2-3 factors I wish were covered from start to finish in a really in depth way.

1. Noise. How much noise can we hear, in relation to background noise, frequency of that noise, frequency of noise that tend to be generated by poor amps/dac. How do we calculate how much noise we expect to hear at a certain distance with dac output dialed down a bit and some digital volume control?

2. How much power do I ACTUALLY need? Different amps struggle more or less with low impedance loads and without data that hits those low frequency loads I'm not sure we can be totally confident about power. People routinely talk about dynamic range of music, and sometimes equal loudness contours and db a weighted. I know Amir did a video on how loud is loud, but that to me was more in context of understanding his THD graphs for speakers and not calculating how much power we actually need. For example we know bass frequencies require more power, Amir pointed out that music is mixed with that in mind. The raw spl of music is mostly in the bass. I think we need a nice tool/guideline/thread that neatly lays everything out.

3. Closer to my original point at the very start of this post. This class D amp FR change under complex loads business. Wish it'd just get ironed out so we're stop getting angry threads about it or no context graphs with people who don't want to explain anything.

For 1 and 2 I'm not saying experts don't know the answer, but rather the answer hasn't been communicated well enough. I don't think I'm alone in especially point 2 because I see sooooo manyyyy posts about power and how much power is enough.


I think the most popular KEF speakers here (r3, Meta) are very well designed speakers with great aesthetics. If we're talking Muon or Blade then yeah, but many brands have flagship products for those who want to spend that kind of money. Their budget stuff are just okay.

Have to say that I agree, particularly with points two and three. I am certainly not an expert in electronics and how they should be measured but I get the impression that the methodology used here for amplifier testing is not sufficient to measure the subtleties, and this is why "all amps sound the same" seems to be the prevailing mantra. (Apologies if I am being a bit simplistic, but hopefully you know what I mean.) Surely to test an amplifier properly, it needs to be done in a way that simulates it's real life use-case... Seems like a bit of a no-brainer to me tbh.
DACs, obviously are not like this. I bought my DX3 Pro within 15 minutes of reading it's review here and doubt I will be upgrading unless I end up with an amplifier with balanced inputs. I will qualify this by adding that i may be completely wrong as I have a very limited knowledge of the prerequisites required for a good amplifier, only that I believe that I can perceive subtle differences between some of them.

 

[Doodski]

I am certainly not an expert in electronics and how they should be measured but I get the impression that the methodology used here for amplifier testing is not sufficient to measure the subtleties, and this is why "all amps sound the same" seems to be the prevailing mantra. (Apologies if I am being a bit simplistic, but hopefully you know what I mean.) Surely to test an amplifier properly, it needs to be done in a way that simulates it's real life use-case... Seems like a bit of a no-brainer to me tbh.

Do you have a test in mind that will be conducive to your goal of showing the subtleties? Maybe many of the amps look passable from the test result because they are that good.

I will qualify this by adding that i may be completely wrong as I have a very limited knowledge of the prerequisites required for a good amplifier, only that I believe that I can perceive subtle differences between some of them.

There are some that sound different when run within their means. Not many but some.

 

[Chr1]

Do you have a test in mind that will be conducive to your goal of showing the subtleties? Maybe many of the amps look passable from the test result because they are that good.


There are some that sound different when run within their means. Not many but some.

Yes. I think several people, way more knowledgeable than me have mentioned that they should be measured in such a way that simulates driving a real speaker (variable impedance/resistance?). I presume that this could be why, what I perceive to be good quality amplifiers often quote figures into a 2 Ohm load? I assume that this is very relevant to the actual quality of the sound produced?

 

[Geert]

You have taken a fraction of what rdenney said and focused in on it. He makes a long, nuanced and interesting post and you don't deal with the whole of his message but a tiny fraction of it - is this arguing in good faith?

I'm OK with Rick's post so why should I need to respond to every aspect of it? I'm not arguing with Rick, as in trying to persuade him. I only suggested there might be more to what he said. Am I not allowed to do so and share my opinion? It's again you showing misunderstanding, so you're seriously out of line when questioning my intentions and intellectual abilities.

Sorry if I am wrong about your nationality, but even though the Dutch speak very good English, are you sure you understand...

I just don't understand how you can get hung up on this flawed understanding on the word subjective.

I wasn't discussing the meaning of that word. I questioned the often used argument of "variability of hearing".

 

[Doodski]

they should be measured in such a way that simulates driving a real speaker (variable impedance/resistance?).

That's fine and dandy although there is no standardized fixture/test jig for this test using capacitive and inductive reactance to simulate a speaker as the load. Because there is no standard for this test what will the test results be compared to? Nobody else has data with this sort of test in a standardized form. Therefore this reactive load test becomes kinda moot.

what I perceive to be good quality amplifiers often quote figures into a 2 Ohm load? I assume that this is very relevant to the actual quality of the sound produced?

Sometimes you'll see good power output linearity specs into 2Ω although much of the time it's not very good.

Good power output linearity specs would look something like this:
100w@8Ω
200w@4Ω
400W@2Ω

Linearity is generally considered as the output signal of the amp being the same as the input signal but with more magnitude. In this case we are looking at the power output linearity.

 

[Chr1]

That's fine and dandy although there is no standardized fixture/test jig for this test using capacitive and inductive reactance to simulate a speaker as the load. Because there is no standard for this test what will the test results be compared to? Nobody else has data with this sort of test in a standardized form. Therefore this reactive load test becomes kinda moot.


Sometimes you'll see good power output linearity specs into 2Ω although much of the time it's not very good.

Good power output linearity specs would look something like this:
100w@8Ω
200w@4Ω
400W@2Ω

Linearity is generally considered as the output signal of the amp being the same as the input signal but with more magnitude. In this case we are looking at the power output linearity.

Thank you. Does this mean that it is not possible to accurately quantify/measure how accurately the amplifier will control the woofer? (I assume the tweeter is less of an issue ie clarity/transparency of the signal are the important factors) Anyway, I appreciate your help explaining the relevancy of these things. Ta.

 

[Doodski]

Does this mean that it is not possible to accurately quantify/measure how accurately the amplifier will control the woofer?

Most if not in the high percentile of the issues of woofer operation will be because of the woofer's own issues. It is possible to create a test circuit that does present capacitive reactance and inductive reactance plus resistance but again it would be solely for internal ASR comparison purposes because as I mentioned nobody has standardized a test like this. It would possibly negate all the testing done previously because ASR would switch to a new test method and so reference to pre-tested gear would not be possible.

 

[DanielT]

I'm probably like most people here, looking at data and tests based on them in terms of everything in Hifi, but I have to test speakers at home. For my part, I think it's easier to have at least two pairs of speakers at home and test. Better two ok pairs than a pair better at the same price. I only buy used and then ones I know I can sell and buy for about the same price. Then I let the subconscious rule. The speakers I listen to the most after a few months I keep.
Right now two cheaper, but good. I have refomed the JPW. I'll see which ones I keep.
Edit:
The white speakers QLN One. No resonances in the box. Sealed, padded with moff, filling. The most difficult speaker I have had. 200 W amp is barely enough. They really eat effect. Sounds good in and of itself. You want to play at high volume with them. Indicates relatively low distortion. I suspect.
Though I have to blame myself. Currently playing old unbroken loudness destroyed ruined CDs. Lots of dynamism in them. That in itself sets its requirements. Fun not to listen to even porridge per se.

 

[Chr1]

Thanks. Yes, I can see that

Most if not in the high percentile of the issues of woofer operation will be because of the woofer's own issues. It is possible to create a test circuit that does present capacitive reactance and inductive reactance plus resistance but again it would be solely for internal ASR comparison purposes because as I mentioned nobody has standardized a test like this. It would possibly negate all the testing done previously because ASR would switch to a new test method and so reference to pre-tested gear would not be possible.

Thanks again. I guess that I would be interested to know how amplifier designers/manufacturers test and develop their designs. I think that I have figured out by reading up here that enough linear power is of great importance. I may be wrong but perhaps keeping the signal "clean" through good circuit design and components may also be ideal. I suspect that this is why I (possibly wrongly) perceive better sound from one of my amplifiers. It appears to have a very robust power supply and is rated at 50w@8, 100w@4 and 200@2 Ohms. Aparently it is good for driving difficult loads/speakers. "Scintilas"? I gather that it is capable of very high current even though only rated at 50w@8o...(not sure what that means in practice however!)

 

[cd45123]

I’m glad this thread started. I have slowly come to the realization that measurements have major significance as I’ve gone through the idiotic process of DAC comparisons at a 70db or less typical listening level…think I’m just fine with my blah Bluesound.

While there are certain amp designs that I appreciate and have an affinity for voicing that is a little more forward (if you believe in that), having dropped a ton of cash on these things has proved to be mostly a waste and my shiny new object fascination (syndrome?) has caused me to run in circles and forget about listening to great music.

This site gets a lot of hate from the reviewers and other boards, but I’d argue that it’s much more civil and productive than some of the typical haunts (Audiogon anyone?)

 

[Geert]

I suspect that this is why I (possibly wrongly) perceive better sound from one of my amplifiers. It appears to have a very robust power supply and is rated at 50w@8, 100w@4 and 200@2 Ohms. Aparently it is good for driving difficult loads/speakers. "Scintilas"? I gather that it is capable of very high current even though only rated at 50w@8o...(not sure what that means in practice however!)

Some brands specify a lower power rating at higher impedances so they can make it look the amp is capable of doubling power at lower impedance. They know audiophiles see the power doubling trick as a sign of quality. The only thing that matters is that it's capable of providing the power you need for your speakers and your preferred listening level.

 

[Chr1]

Some brands specify a lower power rating at higher impedances so they can make it look the amp is capable of doubling power at lower impedance. They know audiophiles see the power doubling trick as a sign of quality. The only thing that matters is that it's capable of providing the power you need for your speakers and your preferred listening level.

Yes, but some amplifiers are not capable of driving low impedance loads. This is my point. The fact that it can drive Scintillas, which are rated at 1 Ohm, I presume is a good indication of a very stable linear power supply. Not all amplifiers will have this.