<rant>We can usually consistently duplicate a measurement if we know the conditions, so they can be called fact, and they are very useful, particularily when developing & debugging, but until there is a battery of scientifically valid blind tests that correlate objective measurements to what we hear, they are just measurements. Until such time as these correlations exist, objective measurements give no objective valuation of how something sounds.
So just as listening in a non-blind or poorly designed blind test, has to be suspect because of external influences, so do today's objective measurements carry just as little weight until such time as they can be correlated to what we hear.
Very little of any of that essential correlation exists, scientifically and statistically valid blind listening tests are not easy to set up, are time consuming (and therefore tend towards expensive) so we see few of them.
Leaves one in a real quandry, when it comes to evaluating the end result, what reaches our ears.
dave
</rant>
"- Worse sound, and better measurements? - No Go Actually this is "check your brain". It is compensating for the bad sound that you are listening every day and the good sound sounds bad now"
Absolute rubbish! If 'perfect' measured performance equated with 'perfect' performance please explain why must Japaese amps from the late 70s/80s which had exemplarary measured performance often sounded cold, hard and clinical sounding, compared with hearing a live performance, or even just strumming a guitar in your own house so you have a 'real world' musical reference point?
There are still musical parameters yet to be understood fully by science - the ear should be the final arbiter of good performance, as that is where the final part of the hifi chain terminates afterall!
Absolute rubbish! If 'perfect' measured performance equated with 'perfect' performance please explain why must Japaese amps from the late 70s/80s which had exemplarary measured performance often sounded cold, hard and clinical sounding, compared with hearing a live performance, or even just strumming a guitar in your own house so you have a 'real world' musical reference point?
There are still musical parameters yet to be understood fully by science - the ear should be the final arbiter of good performance, as that is where the final part of the hifi chain terminates afterall!
Reading this discussion, I was thinking that the measurement vs. subjective debate will go on forever. Why? because our ear/brain is not a measurement device. A frustration for the engineering approach.
However science continues to advance our understanding of the brain and maybe soon someone with deep pockets, like a Bose, will use brain imaging techniques that monitor neural activity in real time and figure out what really does sound good. They will try out a tube amp and the whole brain will light up.
Dave
However science continues to advance our understanding of the brain and maybe soon someone with deep pockets, like a Bose, will use brain imaging techniques that monitor neural activity in real time and figure out what really does sound good. They will try out a tube amp and the whole brain will light up.
Dave
Your logic is "rubbish". What you quoted from my post doesnt equate with what are you ranting about.
You imply that I wanted to say "perfect measurements" equal "perfect sound".
Actually, I have said "worse sound" cannot result from "better measurements". It is a difference - logically you cannot permutate words (like we do in math) and keep the affirmation true...
If a device measures worse than other, will sound worse.
If measurements are the same or better than our perception level for that parameter, then they still might sound different in a small way (that maybe was not measured).
There is no "magic".
Last edited:
It certainly can... take a well designed good sounding amp. Wrap alot more feedback around it to get better THD. Invariably it will measure better & sound worse.
Until such time as the science has been done to correlate measurements to what we hear they are just measurements, a useful tool, but that is all.
dave
IMO "measurements" doesn't mean only TDH! You oversimplify the measurements by either lack of kwnoledge or plain missinformation due to personal agenda.
I can't belive you didn't hear about slew-rate, full-power bandwidth, noise, PSSR, CMRR...
All those will give a good "feeling" to how the circuit will sound. The argument that "the negative feedback invariably sounds worse" is from the '60s-'70s when they where trying to hide the bad design with feedback. Is not the case today.
I can't belive you didn't hear about slew-rate, full-power bandwidth, noise, PSSR, CMRR...
All those will give a good "feeling" to how the circuit will sound. The argument that "the negative feedback invariably sounds worse" is from the '60s-'70s when they where trying to hide the bad design with feedback. Is not the case today.
I know all about those... but no one has yet done scientifically valid correlations between those and what something sounds like ... they are just tools the designer can use to help him understand his circuit.
A feeling... yes, that may be as close as it gets.
And i did say well designed amplifier.
dave
A feeling... yes, that may be as close as it gets.
And i did say well designed amplifier.
dave
"If a device measures worse than other, will sound worse."
Frankly Sir I believe it's your logic that is "rubbish" in this instance. And it also depends on exactly WHAT you are measuring, and even then one needs to be CERTAIN you are aware of what REAL WORLD effect that has on the sounds that you HEAR. Hearing is NOT looking at a set of perfect measured results on a scope. I wish that it was as those Japanese amps I was talking about earlier with near enough perfect measured performance would be all we ever needed! It was for a while in the 80s until people realised they were not getting emotionally involved with their music, and the valve revival (for example) in the 90s bought millions back in touch with that important but often elusive factor in listening to music. Just one example of how measured performance in amps didn't equate with what we were hearing.
We need a real world reference when making an annoucement as to whether an amp or DAC (etc) has achieved its goal or not - and that reference is the real thing; an instrument or voice, heard via our own measuring devices - ears! Unless you are designing audio devices to satisy measuring equipment only? :s
Frankly Sir I believe it's your logic that is "rubbish" in this instance. And it also depends on exactly WHAT you are measuring, and even then one needs to be CERTAIN you are aware of what REAL WORLD effect that has on the sounds that you HEAR. Hearing is NOT looking at a set of perfect measured results on a scope. I wish that it was as those Japanese amps I was talking about earlier with near enough perfect measured performance would be all we ever needed! It was for a while in the 80s until people realised they were not getting emotionally involved with their music, and the valve revival (for example) in the 90s bought millions back in touch with that important but often elusive factor in listening to music. Just one example of how measured performance in amps didn't equate with what we were hearing.
We need a real world reference when making an annoucement as to whether an amp or DAC (etc) has achieved its goal or not - and that reference is the real thing; an instrument or voice, heard via our own measuring devices - ears! Unless you are designing audio devices to satisy measuring equipment only? :s
Last edited:
I never said that. Just that measurement's (as they exist today) ability to help tell us how something sounds is VERY limited. They are on the other hand quite useful when actually in the design process.
There are some notable cases where there is correlation.
dave
Dave,
Maybe you want to tell that to the owners who paid $ 40.000 for a Halcro?
(And yes, there are other examples).
It would say that you can built good and bad amps with feedback, as well as good and bad amps without it or with very little feedback. Just as THD doesn't correspond to sound quality, neither does the amount of feedback db's.
You were mostly right before when saying that measurements do not (or at least not always) correspond with sound quality. I'd say that measurements do not or not always correspond with listener preference. A small difference but important. A listener's preference does not necessarily correspond to sound quality except for that listener.
jd
Last edited:
Dave,
Maybe you want to tell that to the owners who paid $ 40.000 for a Halcro?
(And yes, there are other examples).[/I].
jd
Hm
I thought that Bruce was using the Hawksford / Cordell EC circuit that you are using in the PAX.
People USE measurements to anticipate the outcome of a certain event.
We do measure the time, even if that is not the time that WE feel it is passing. We measure the weight when we buy 1kg of sugar even if that doesn't say nothing about how we will feel when we eat that. We measure the volume of the water even if that abstract number doesn't quench the thirst. We measure the Universe laws and use those to predict how a spaceship will reach to Mars.
So on...
The capacity of understanding abstract concepts like measurements diferentiate us from animals.
And now, some individuals want to convince us that mesurements don't mean nothing, that you cannot predict the outcome based on those measurements.
Agree, if you are a monkey. Or a dog. Or a horse.
If you are a human beeing, you surely can make out something from those measurements.
We do measure the time, even if that is not the time that WE feel it is passing. We measure the weight when we buy 1kg of sugar even if that doesn't say nothing about how we will feel when we eat that. We measure the volume of the water even if that abstract number doesn't quench the thirst. We measure the Universe laws and use those to predict how a spaceship will reach to Mars.
So on...
The capacity of understanding abstract concepts like measurements diferentiate us from animals.
And now, some individuals want to convince us that mesurements don't mean nothing, that you cannot predict the outcome based on those measurements.
Agree, if you are a monkey. Or a dog. Or a horse.
If you are a human beeing, you surely can make out something from those measurements.
Dave,
Maybe you want to tell that to the owners who paid $ 40.000 for a Halcro?
(And yes, there are other examples).
It would say that you can built good and bad amps with feedback, as well as good and bad amps without it or with very little feedback. Just as THD doesn't correspond to sound quality, neither does the amount of feedback db's.
You were mostly right before when saying that measurements do not (or at least not always) correspond with sound quality. I'd say that measurements do not or not always correspond with listener preference. A small difference but important. A listener's preference does not necessarily correspond to sound quality except for that listener.
jd
Jan,
I didn't say feedback was bad... what i did say id that if you take a well executed good sounding amp, and then add more feedback (or some if it didn't have any), you'd likely get better measurements and worse sound.
It is perhaps a bit contrived, but inspired by something Nelson said, to paraphrase, "with more feedback the measured distortion went down, but it didn't sound as good".
It was purely to counter Mr SoNiC's contention that if it measured better, it would sound better, and if it didn't it was in your head.
dave
With audio that is unfortunately the case. The scientific evidence to correlate what we hear with what is measured is sadly sorely lacking, so we don't have a lot of tools to let us use measurements to predict how things will sound.
You can predict whether an amp is going to be noisy, if the noise measurements are really bad, and Toole & Olive have proposed a metric that they say correlates with what people hear in speakers (yet to be independently duplicated, and some people have bones to pick with it), and Geddes has shown that THD is meaningless in loudspeakers (and hinted the same with amplifiers), again not independently duplicated. Kunchur has shown that some people can detect time differences of 5-6 uSec (he used a small sample so we don't know how far into the population it can be extended), claimed to have been indepently duplicated althou i haven't seen that paper. Real world examples beyond those anyone?
dave
It is funny that there are so many of this kind of threads, where someone gets so overheated on the subject of measurement vs hearing. I think everyone has to use some buffer in reading this responses and not take them as black or white. No one is saying that measurements are not important, but if you reread carefully what Dave was saying, you will find the wisdom in it. Here it is again:
"I never said that. Just that measurement's (as they exist today) ability to help tell us how something sounds is VERY limited. They are on the other hand quite useful when actually in the design process.
There are some notable cases where there is correlation."
Now answer please, how do you measure depth in sound, and I do not mean width, but depth, if you know what I mean. Very often that is not just created with speaker and it's position, but changes as we substitute DAC, or amp or preamp.
To go back on the subject, AK4396 vs AK4393 is improvement and worth the effort. It is not day or night improvement, but it is definitely audible. Highs are much more detailed, more sophisticated and natural with AK4396. I haven't had a chance to hear or try 4395. When I was moding my Behringer I exchanged first one DAC only and left two others in order to compare them to each other. That way I was able to have a perfect A/B set up with same volume and same set up in the rest of the system, to learn the exact difference. Now I didn't measure A to B, but I didn't need that, since I was already enjoying the difference in sound. I guess, that was my goal?
"I never said that. Just that measurement's (as they exist today) ability to help tell us how something sounds is VERY limited. They are on the other hand quite useful when actually in the design process.
There are some notable cases where there is correlation."
Now answer please, how do you measure depth in sound, and I do not mean width, but depth, if you know what I mean. Very often that is not just created with speaker and it's position, but changes as we substitute DAC, or amp or preamp.
To go back on the subject, AK4396 vs AK4393 is improvement and worth the effort. It is not day or night improvement, but it is definitely audible. Highs are much more detailed, more sophisticated and natural with AK4396. I haven't had a chance to hear or try 4395. When I was moding my Behringer I exchanged first one DAC only and left two others in order to compare them to each other. That way I was able to have a perfect A/B set up with same volume and same set up in the rest of the system, to learn the exact difference. Now I didn't measure A to B, but I didn't need that, since I was already enjoying the difference in sound. I guess, that was my goal?
To go back on the subject, AK4396 vs AK4393 is improvement and worth the effort. It is not day or night improvement, but it is definitely audible. Highs are much more detailed, more sophisticated and natural with AK4396.
Exactly.
Only a cap or a Transformer after it will bring sonic bliss.
What is better sounding? I would think a good transformer would be more transparent than caps. Looking to do the mod.
I have a bunch of Sonicaps and Russian PIO caps that I could use also. What is the best values for caps? Need to decide between the two options. Thanks!
Direct out with caps
I like the direct out mod with caps and am lucky enough to have the Dayton foil caps available for cheap in the USA. Try 2-4uf.
Hi Scott,
did you try to parallel the Daytons with a very small silver mica like 2-5nf?
It worked for the better with any cap I tried.
Also, did you try to parallel Ak439X dacs?
Alex Peychev of APL did this in all his dacs and
they are rated best among the best.
I got a sound card with four 4396 on it and
I wounder if I should give paralleling a try.
Klaus
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.