yes, I am aware of all that. You 'could' do these things. or not. We just need to normalize. other companies or other products have different 0FS. So that is the purpose.... not that a work around can be used to good advamtage.
The AD part is not spec'ed any better or worse than others. I left off the name because it was just a generic example that you cannot use the headline number when the Standard which is given below it is the number to use. again for comparing with others -- use the Standard. The number to use is the THD+N for dynamic range.
Now, if we start with 90db dynamic range, and reduce from there by adding in jitter and muti-tone distortion levels and all other artifacts which we must listen to as well as the music, the listener is looking at a much reduced 'real' dynamic range and it would be audible. All has nothing to do with your companies part.
Then there are the other issues mentioned. The -60db distortion. To balance tha scales, a couple years ago when i was trying to find -60dB thd levels.... not everyone does spec them.... I found AD to be the best at -60 and ordered some DAC Kits from Taiwan which used them.
Also recommended them to mfr in Asia because they could sound better.
THx-RNMarsh
The AD part is not spec'ed any better or worse than others. I left off the name because it was just a generic example that you cannot use the headline number when the Standard which is given below it is the number to use. again for comparing with others -- use the Standard. The number to use is the THD+N for dynamic range.
Now, if we start with 90db dynamic range, and reduce from there by adding in jitter and muti-tone distortion levels and all other artifacts which we must listen to as well as the music, the listener is looking at a much reduced 'real' dynamic range and it would be audible. All has nothing to do with your companies part.
Then there are the other issues mentioned. The -60db distortion. To balance tha scales, a couple years ago when i was trying to find -60dB thd levels.... not everyone does spec them.... I found AD to be the best at -60 and ordered some DAC Kits from Taiwan which used them.
Also recommended them to mfr in Asia because they could sound better. THx-RNMarsh
Last edited:
Work around - it's known as doing your job and knowing what is actually going on, in other words being a competent engineer. The spec you keep showing is something that anyone could design right now, it is not some target to aim for in the future. Things that fall short don't interest me.
Scott,
You are the one who keeps poking fun at my resistor measurements. The first order result was that price is not indicative of resistor distortion. The Mouser Xicon 1% 1/4W metal film resistor samples measured had excellent results at about the lowest price.
Now at the expensive end the naked Vishays did not perform as well as the best of the Dales but cost more. I sent some to a few folks and even with the measured distortion the lowest of any samples, the comments were consistent with the measured differences. Now the random chance of that happening is low but certainly not zero. (Probably around 5-10% or so.) Now you can assume that any metal film resistor with the same tempco will behave the same, but measurements show there are still differences.
In the Linear Audio vol. 4 article the proposed weighting was the harmonic order divided by eight and the result cubed. The weighting falls off for frequencies above the upper midrange to match our auditory system.
Those results are consistent with the resistor measurements, as the distortion increases as you decrease frequency.
Pavel,
I think you are still missing the point. Scott has raised the distortion issue soley based on harmonic distortion. Of course there are lots of other issues. But sticking to harmonic ones, a loudspeaker instead of masking the amplifier's harmonic distortion appears to make it more perceptible. If you have an amplifier with the 5th harmonic down by 80 dB. and a transducer with the 5th harmonic down by 40 dB. the 25th should be down by 120 dB. (5/8)**3 = .244 (25/8)**3 = 30.5 20 x log(30.5/.244) =42 dB apparent increase.
Or more simply there is a trade off point. For fifth harmonic the system response should be -122 db or greater to not have the 25th harmonic be perceived as more of a problem then the combined 5th. You can do this calculation for every combination.
The kicker is that the energy of the fundamental is probably 25 dB. greater in musical content than in test signals. So the design goal for the 5th harmonics combined should probably be -150 or so.
You are the one who keeps poking fun at my resistor measurements. The first order result was that price is not indicative of resistor distortion. The Mouser Xicon 1% 1/4W metal film resistor samples measured had excellent results at about the lowest price.
Now at the expensive end the naked Vishays did not perform as well as the best of the Dales but cost more. I sent some to a few folks and even with the measured distortion the lowest of any samples, the comments were consistent with the measured differences. Now the random chance of that happening is low but certainly not zero. (Probably around 5-10% or so.) Now you can assume that any metal film resistor with the same tempco will behave the same, but measurements show there are still differences.
In the Linear Audio vol. 4 article the proposed weighting was the harmonic order divided by eight and the result cubed. The weighting falls off for frequencies above the upper midrange to match our auditory system.
Those results are consistent with the resistor measurements, as the distortion increases as you decrease frequency.
Pavel,
I think you are still missing the point. Scott has raised the distortion issue soley based on harmonic distortion. Of course there are lots of other issues. But sticking to harmonic ones, a loudspeaker instead of masking the amplifier's harmonic distortion appears to make it more perceptible. If you have an amplifier with the 5th harmonic down by 80 dB. and a transducer with the 5th harmonic down by 40 dB. the 25th should be down by 120 dB. (5/8)**3 = .244 (25/8)**3 = 30.5 20 x log(30.5/.244) =42 dB apparent increase.
Or more simply there is a trade off point. For fifth harmonic the system response should be -122 db or greater to not have the 25th harmonic be perceived as more of a problem then the combined 5th. You can do this calculation for every combination.
The kicker is that the energy of the fundamental is probably 25 dB. greater in musical content than in test signals. So the design goal for the 5th harmonics combined should probably be -150 or so.
I place another of mr Gerhards fantanstic links: What is dither.
What is dither? | EarLevel Engineering
What is dither? | EarLevel Engineering
No, I'm saying put up a real DBT listening test (any protocol you want) that says any of the resistor differences matter at all in a BLIND comparison.
BTW have either you or Dick done the Souza Band test, I don't think so failure is not an option. I really have to resist any comment on the numerology.
Last edited:
I won't use Dale resistors anymore. By all means there's nothing wrong with them in typical measurements, but with an amplifier I've been using they destroy the operation. If they're in the signal path there's a problem. If there's more than one per channel it gets much worse. Perhaps this amp is particularly sensitive, but it seems like the closer to the amp they are the worse the affect. That affect gets multiplied greatly say if there's a volume device with them up the chain towards the source, and then some are thrown in next to the amplifier.
I have no way to explain it, but the horrific effect of the music on the ears makes it readily apparent.
Maybe the Zfoils don't measure as well, but I don't expect them to obliterate my sound quality if they're in the chain.
I have no way to explain it, but the horrific effect of the music on the ears makes it readily apparent.
Maybe the Zfoils don't measure as well, but I don't expect them to obliterate my sound quality if they're in the chain.
3 sentences in, red flag, sample bit depth has nothing at all to do with Nyquist. 24 bits is at the "atomic level" why not just read Lipshitz.
Last edited:
My attempt to do that had everyone notice switch position 3 was the outrider. So a silver plated rotary switch was the issue. So the first issue is to find the tools to isolate the test to a single variable.
I have done the Souza band test and looked at it. Doesn't violate any of the perception rules, just doesn't illustrate much that hasn't already been said.
Let me know when you try the high harmonic threshold test. It should take less time than you writing why it is unimportant.
I freely admit out of hand my result would be no different than your graph which I assume represents some kind of average over the general population. It is still unimportant.
Anyway you want to parse it, Scott. The dynamic range is the THD+N figure which is much lower than the often quoted and highlighted as the first Spec.
I would add to that jitter noise. And I would normalize the data across all mfr'ers to get a real and quick view of comparison.
I would add to the THD+N more test tones... multitone test artifacts as added noise.
If anyone ever wondered why careful listeners do not like what they hear with digital, it can be directly dirived from the the Standard def (IASA) for dynamic range and stop promoting something else. Fortunately, the mfr do put up the THD+N under the Dynamic Range heading.
The other point is that 24 bit is not enough if you care about the details represented in the THD at levels other than max output. Like -60dB.
The digital 'sound' that JC and others dont like is right before us in the specs and Standards. The artifacts heard bring the total 'noise' to well above audible thresholds.
More bits are needed for careful listeners over a clean dynamic range of only 60dB isnt being met.... 10 bits. Six more bits at -60 might do it... or a 32 bit minimum system.
Now, dont go say its already been done. We need it in all record and playback for home as well as studio. That will push the 'real' dynamic range higher and give the low level distortion an undetectable amount of THD+N.
That low level detail that we all need.... max level is rarely listened to.... is so far the problem JC and others have with digital sound. More bits and lower extraneous noise from jitter and the like will get us there. Noise level increase from multitone tests should be added via a Standard for that.
Meanwhile, get rid of the intermediate step of pressing and spining an Lp or CD and stream HiRes .... just what will we call it when HiRes isnt so high anymore and we have 32 bit systems. Ultra HiRes? Like Ultra HD.
THX-RNMarsh
Last edited:
Yawn.
I have a standard for you, any car that does not achieve 40 mpg should be banned from the roads or taxed exponentially. After all we have the technology to do it.
Last edited:
Take the Johnson noise equation for a resistor, flip it round a bit, and also look at the ratio of max output to min output of a 16 through to 24 bit D/A setup, with a 20Khz bandwidth, what do you think is the full power that has to be dissipated in the DAC feedback resistor if you want the backgound noise to be the same as the LSB...
Like I said, I don't get it, but it's anything other than subtle. I'm looking elsewhere for resistors, for any audio relation.
Scott,
Problem with those fuel mileage numbers from the EPA is they are simulated! Definitely not real world numbers but okay for comparison reasons only. Nobody gets those numbers in real driving situations unless you are rolling down hill.
I have a feeling that holds true for most published distortion numbers in audio also, so many quote the simulated distortion numbers instead of actual measured response. I see that here and in many of the amplifier threads, where guys are quoting extraordinary distortion numbers you could never reach in reality.
Problem with those fuel mileage numbers from the EPA is they are simulated! Definitely not real world numbers but okay for comparison reasons only. Nobody gets those numbers in real driving situations unless you are rolling down hill.
I have a feeling that holds true for most published distortion numbers in audio also, so many quote the simulated distortion numbers instead of actual measured response. I see that here and in many of the amplifier threads, where guys are quoting extraordinary distortion numbers you could never reach in reality.
wrinkle....I would like to know.... what?
I dont use anything special for resistors and get very good measured and audible results. Dale works fine for me..... Not that i have gone ahead and tried every other brand (some at extream costs). Just there are more pressing issues. And, there are worse R's than Dale MFilms. Just sayin.
THx-RNMarsh
I dont use anything special for resistors and get very good measured and audible results. Dale works fine for me..... Not that i have gone ahead and tried every other brand (some at extream costs). Just there are more pressing issues. And, there are worse R's than Dale MFilms. Just sayin.
THx-RNMarsh
Last edited:
I think this question has been answered in the past but couldn't some type of optical switch be used in a DBLT or at least a remote optical control so the user could not tell a switch position? It seems that you do need to remove any clues that can be used to skew the test. A test where a computer was randomly changing the test that only the program would know the switcch position until the test was complete and then the computer would do the statistical analysis of the variation in answers?
A rhetorical comment. I disagree with Dick on a basic level, that bunch of folks trying to archive recordings are not asking for some not yet achieved performance. I don't see his point.
Scott,
I agree in the respects that it seems that what is really being asked for is a perfect transfer from whatever media is storing the information. I really don't see how storing the data on a spinning disk, CD, hard drive or record could ever be perfect as long as we are going through conversion processes. Now what I think that Richard seems to be wanting is an ever higher bit rate and clock frequency. I just don't see how you ever completely remove jitter or noise ever, yes we can use dither to reduce any unwanted effects but it seems you are just using finer and finer steps but it is an asymptotic function isn't it, the closer you get to perfection the more you realize you can never quit get there.
Gladly humans really aren't that good at discrimination of all this noise and we have evolved to ignore most of the background noise anyway as long as it is low enough in level compared to what we are paying attention to in the fist place/
I agree in the respects that it seems that what is really being asked for is a perfect transfer from whatever media is storing the information. I really don't see how storing the data on a spinning disk, CD, hard drive or record could ever be perfect as long as we are going through conversion processes. Now what I think that Richard seems to be wanting is an ever higher bit rate and clock frequency. I just don't see how you ever completely remove jitter or noise ever, yes we can use dither to reduce any unwanted effects but it seems you are just using finer and finer steps but it is an asymptotic function isn't it, the closer you get to perfection the more you realize you can never quit get there.
Gladly humans really aren't that good at discrimination of all this noise and we have evolved to ignore most of the background noise anyway as long as it is low enough in level compared to what we are paying attention to in the fist place/
This is not relevant - I've mentioned many times playing a classical piece on a test CD, that is a separate track and attenuated 60dB on the disk, without dither! I wind up the volume to absolutely maximum, stick my ear hard against the cone of the speaker, and can just hear what's going on - and that is, music! While the level is about at average the sound comes across with complete validity, nothing is lost or sounds terribly damaged - when the sound level of the music drops off, yes, you can hear the digital chattering happening; but as soon as the level picks up the chattering disappears, subjectively.
I have only one CD where the mastering was messed up - at the start you can hear digital gurgling when the volume is right for the performance - an operatic aria highlights, of all things - and ripping that track and looking at the encoding you can see that someone stuffed up - digital noise is way, way too high ...
- Status
- Not open for further replies.