DAC types and their sonic signature

[SIY]

Now. . . I have a real actual question—could ultrasonic frequencies ever have an impact on perception of audible frequencies? Even at the extremes?

At the levels in this context? No.

 

[SIY]

Sorry, if a (say) 50 kHz signal at 40 dB SPL is added, it's just not going to be heard.

Want to talk 130-140dB SPL ultrasonics? Sure, that can have an effect. But those are stunningly unrealistic unless you have something purpose-built to achieve that.

 

[solderdude]

Now. . . I have a real actual question—could ultrasonic frequencies ever have an impact on perception of audible frequencies? Even at the extremes?

The question is incomplete because you did not specify whether or not the ultrasonics are related or not related.
Also you did not state what you consider extremes.

That question can easily be answered by yourself when you want to know your hearing abilities.
For this there are many tests around.
All you need to do is have a DAC capable of reaching 192/24 and have files which have been properly made and compare these files with AB test software.
A possible snag here is how well the transducers and amplifiers used deal with ultrasonics.
When this is O.K. it is perfectly fine to use your ears... as long as levels are unchanged and one does not know what file is being reproduced.

 

[StevenEleven]

Sorry, if a (say) 50 kHz signal at 40 dB SPL is added, it's just not going to be heard.

Want to talk 130-140dB SPL ultrasonics? Sure, that can have an effect. But those are stunningly unrealistic unless you have something purpose-built to achieve that.

Got it. Thanks.

 

[solderdude]

Other questions would be ...
How do they shape the sound?

Ask them and they will claim that it is done by selecting 'special' components and 'combining house sounds of components' or 'special filter algorithms' or usage of this or that material.
When you know of 2 different typical 'house sound' signatures that both appear to measure well within the audible range I would invite people to compare both DACs blind (as in not knowing what DAC is being used) while being level matched.
I am quite sure the 'house sound' cannot be proven at all unless substantial differences (that are very measurable) exist.

Funnily enough all the 'housesound' manufacturers all claim their DAC sound more 'real' than other brands or 'is closer to analog'.
How can they all be right ?

 

[StevenEleven]

The question is incomplete because you did not specify whether or not the ultrasonics are related or not related.
Also you did not state what you consider extremes.

That question can easily be answered by yourself when you want to know your hearing abilities.
For this there are many tests around.
All you need to do is have a DAC capable of reaching 192/24 and have files which have been properly made and compare these files with AB test software.
A possible snag here is how well the transducers and amplifiers used deal with ultrasonics.
When this is O.K. it is perfectly fine to use your ears... as long as levels are unchanged and one does not know what file is being reproduced.

Thanks. I know you could come at it from an a/b/x perspective but if the answer is it would take a 130-140 SPL signal that’s not something I’m going to try at home, even if I could educate myself enough to set it up and have transducers that could reliably reach those frequencies. But your reply does give me a little more insight.

 

[solderdude]

Using an ABX test at home with the correct equipment one could test this at 'comfortably loud listening levels'.
That's the realistic approach. Ultrasonic content will always be a LOT lower than the higher audible frequencies anyway.
They also would need to be recorded as well which is not trivial.
One would have to check the files to ensure actual relevant ultrasonic frequencies are present and it is not just noise.

 

[andreasmaaan]

As someone who's training was done many years ago my standard load for line outputs is 600 ohms. If an opamp output, balanced or unbalanced does better than 0.02-0.03% THD into 600 ohms, then I'm happy the device is working properly. In practice, it'll most likely to be used into a bridging impedance of around 10k, so the distortion will likely be lower, but as long as the distortion is well below audibility, that's quite good enough for Government work.

Occasionally, a device will have the minimum load specified, in which case I test at that load, otherwise 600 ohms is a good number to use.

S.

How did you arrive at 600 Ohm as the figure to test at? Is this kind of an absolutely worst-case scenario terribly-performing amp input stage figure? Or some other reason?

 

[andreasmaaan]

Thanks. I know you could come at it from an a/b/x perspective but if the answer is it would take a 130-140 SPL signal that’s not something I’m going to try at home, even if I could educate myself enough to set it up and have transducers that could reliably reach those frequencies. But your reply does give me a little more insight.

At least in theory, ultrasonic content that is inaudible could cause a transducer to create intermodulation distortion that is audible. So, for example, if a transducer is reproducing a 30kHz tone and a 32kHz tone, this may result in an intermodulation product at (32-30=) 2kHz that, if loud enough and if not masked by signal content, would be audible.

I'm not aware of any tests of common HF transducers that have investigated whether this is actually of any audible concern, however. Given that most studies into the audibility of high-res vs. redbook have returned negatives (i.e. no audible difference), it would seem safe to presume there is probably in practice nothing to worry about.

 

[StevenEleven]

At least in theory, ultrasonic content that is inaudible could cause a transducer to create intermodulation distortion that is audible. So, for example, if a transducer is reproducing a 30kHz tone and a 32kHz tone, this may result in an intermodulation product at (32-30=) 2kHz that, if loud enough and if not masked by signal content, would be audible.

I'm not aware of any tests of common HF transducers that have investigated whether this is actually of any audible concern, however. Given that most studies into the audibility of high-res vs. redbook have returned negatives (i.e. no audible difference), it would seem safe to presume there is probably in practice nothing to worry about.

Awesome, thanks. What I was trying to get my mind around is whether this was a question of degree or whether theory would just give me a flat out no. So from two perspectives now from people who know a lot more than I do I see it is a question of degree. It’s a factual matter, not one where theory alone will give us an answer.

I think as a factual matter the difference between redbook and hi-res is smoke and mirrors because the hi-res releases are based on recordings with mics that at best are just not going to record anything beyond the tiniest (virtual) whisper at inaudible frequencies, and the ranges of most instruments most often drop off to inaudible well before ultrasonic frequencies even if we are considering harmonics. So, from my perspective, and based on my limited knowledge, hi-res is pointless for music due to what is inherently in the music before it is recorded and then once again based on how the music is recorded. Just from the get-go it seems inevitable on two independent bases that hi-res is futile as a way of attaining audible benefits or generating audible changes in recorded music. If my understanding is correct for example in actual music or recorded music there is just not ever going to be a 30 kilohertz tone and a 32 kilohertz tone to clash and create the two kilohertz intermodularion distortion.

 

[777]

How did you arrive at 600 Ohm as the figure to test at? Is this kind of an absolutely worst-case scenario terribly-performing amp input stage figure? Or some other reason?

You do not know very much about op-amps..... First of all, we want to extract more current from the output stage of op-amps to force an exist from a A class and entering in a B or AB class. There are few loads standards like 2.2kohm and 600ohm.

 

[SIY]

How did you arrive at 600 Ohm as the figure to test at? Is this kind of an absolutely worst-case scenario terribly-performing amp input stage figure? Or some other reason?

600 ohms was common in studios decades ago. Nowadays, it only hangs on as an artifact, even in studios. 2k is a modern day worst-case.

I can think of one or two fashion audio amps that use 600 ohms as an actual input impedance, but they might represent (on a good day) 0.001% of situations. Otherwise, it's only used to bring out differences in circuits that aren't actually relevant to the intended use.

 

[sergeauckland]

How did you arrive at 600 Ohm as the figure to test at? Is this kind of an absolutely worst-case scenario terribly-performing amp input stage figure? Or some other reason?

It goes back to telephony in the late part of the 19th Century. (No, I don't remember it personally) Open wires spaced a certain distance apart have a characteristic impedance of 600 ohms, so terminating equipment had to present that impedance to avoid reflections. It has no real bearing on 'normal' cable lengths, but becomes relevant on cable lengths of hundreds of metres or kilometres such as telephone cables. By custom and practice, it became the standard sending and receiving impedance in studios. Now, with equipment having essentially a zero ohms output impedance, and a high >10k ohm input impedance, 600 is of historic interest, but nevertheless, any decent piece of modern equipment should drive 600ohms, so it's still used as a 'standard' load on specifications.

S

 

[777]

http://www.nanovolt.ch/resources/ic_opamps/pdf/opamp_distortion.pdf

 

[andreasmaaan]

You do not know very much about op-amps..... First of all, we want to extract more current from the output stage of op-amps to force an exist from a A class and entering in a B or AB class. There are few loads standards like 2.2kohm and 600ohm.

I think you may have misunderstood my question. I wasn't asking why an opamp would be likely to perform worse into a lower impedance load, which is clear, but rather why the figure of 600 Ohm was arrived at as a standard given that no decent real-world device presents such a load to a DAC or preamp (well I have come across one power amp that does present such a load, but I'm not sure it could be described as decent).

Anyway I think the answer is clear now.

http://www.nanovolt.ch/resources/ic_opamps/pdf/opamp_distortion.pdf

Very nice resource, thanks

EDIT: there are some more opamp measurements here FWIW, although the load impedance is not stated as far as I can tell.

 

[777]

I think you may have misunderstood my question. I wasn't asking why an opamp would be likely to perform worse into a lower impedance load, which is clear, but rather why the figure of 600 Ohm was arrived at as a standard given that no decent real-world device presents such a load to a DAC or preamp (well I have come across one power amp that does present such a load, but I'm not sure it could be described as decent).

Anyway I think the answer is clear now.



Very nice resource, thanks

EDIT: there are some more opamp measurements here FWIW, although the load impedance is not stated as far as I can tell.

Very nice resource, thanks from me too

 

[Ron Texas]

It's well known that the sonic signature of certain delta sigma DAC's causes insanity and blindness.

 

[Calexico]

The question is incomplete because you did not specify whether or not the ultrasonics are related or not related.
Also you did not state what you consider extremes.

That question can easily be answered by yourself when you want to know your hearing abilities.
For this there are many tests around.
All you need to do is have a DAC capable of reaching 192/24 and have files which have been properly made and compare these files with AB test software.
A possible snag here is how well the transducers and amplifiers used deal with ultrasonics.
When this is O.K. it is perfectly fine to use your ears... as long as levels are unchanged and one does not know what file is being reproduced.

It's a wrongz method.
You must first validate that the dac is perfect until 96khz to use it for such test.
If your hypothesis is that the dac is perfect until 96khz then your hypothesis is what you want to proove. Then you proove nothing.

 

[solderdude]

When we use 'Amir approved' DACs it performs well enough

And... it does not need to be 'perfect' to up to 90kHz to reproduce small signals of max 40kHz to 50kHz.

 

[Calexico]

Well very strange.
When logged in there are lot of technical problems to post messages and to load the forums. When loading the same page in anonymous it loads very fastly...