My Ultra tests

[Musician]

That's the conclusion which makes me a bit nervous, as I do not know what to think about it. You would use an LPF by default when measuring some class D amps...
But on the other hand I think that this is the impact of the PSU, which is placed on the same PCB as analog outputs, and it could be reduced.

If the Ultra uses ne5532 in the analog output ( as they do in the pro+ ) the CMMR are 100 dB for that circuit. Using opa1612 will give 120 dB , and those smps spikes would probably be down further 20 dB .

Has someone opened up the Ultra to take a look at the analog parts ?

 

[steadyshot]

The limits of human hearing are well documented (if you don’t know what they are, I’m not doing the legwork for you).

If you believe that perhaps they will differ in certain circumstances, the onus is on you to provide evidence.

I am more than happy to accept that the science is wrong, or flawed, or incomplete, but the starting point would have to be properly controlled testing.

Put another way, the scientists have provided evidence about what can and can’t be heard. If you believe otherwise, the least you can do is point to an independent test showing that a statistically significant number of people can hear the difference, even if you can’t explain what it is or how it works.

As I’ve said, this is the same scientific method used to test your medicines and aircraft.

You are transferring an opinion on a forum . The only known fact for the human hearing is that we can hear frequencies in 20hz-20khz and feel sub bass even beneath 20hz if it is produced at sufficient SPL levels . The same is said for frequencies above 20khz . Because you simply transfer an opinion and you do not have a deep understanding of these claims you cannot hold a conversation . Go educate yourself first and then come back with structured and to the point arguments

 

[onlyoneme]

You are transferring an opinion on a forum . The only known fact for the human hearing is that we can hear frequencies in 20hz-20khz and feel sub bass even beneath 20hz if it is produced at sufficient SPL levels . The same is said for frequencies above 20khz . Because you simply transfer an opinion and you do not have a deep understanding of these claims you cannot hold a conversation . Go educate your self first and then come back with structured and to the point arguments

I kindly ask you to avoid any ad personam comments if you would like to prevent me of asking @Brantome for a moderation.

 

[Brantome]

Go educate yourself first and then come back with structured and to the point arguments

Please keep comments courteous, thanks.

 

[Steve Woodhouse]

You need to specify the exact thresholds for every conceivable measurement , the exact conditions , the rest of equipment characteristics and how they interact with the equipment in test . What is a transparent system - do not look equipment in isolation. Eg what is the final snr of a system at what spl level ? What is the thd +n of a system including speakers at all frequencies at all levels . How about transient speed ? How about intermodulation distortion of a system as a whole . What are the exact thresholds? Who tested them properly to find out that there is no discernible audible difference at any condition . We want exact numbers and tests in acoustically treated rooms that space cannot distort heavily the output of systems in comparison.

In the end we all want equipment as technically superior as possible in every field. The goal is to have full frequency response, with very low distortion and noise at any volume level, with any imaginable characteristic that can possibly affect sound output to be SOTA. Therefore when you have measurements for everything you have to take them all to consideration, because the better they are the higher the chance you will achieve this result. Don't mess audibility with measurements because then your are the subjective one. Stay with what you know for certain.

Hi, some good points.

I have at least some sympathy with some points. Yes, we will hear the whole system, not just one unit in it. But we cannot measure every penetration of every hi-fi product in existence. And if, for example, you are deciding between two DACs, the only thing it’s possible to know is how they measure. You almost certainly won’t be replacing your whole system, you’ll just be replacing a current DAC, or introducing one in place of an internal one.

As an aside, and as a general rule, distortion and frequency issues will almost certainly be far higher in your speakers than any issues in your source, DAC, pre amp and amp added together.

As such, let’s turn this on its head. As even superb speakers are that ‘bad’, surely their deficiencies would massively and completely mask any issues with even a modest DAC-amp combination. So how come some subjectivist golden ears can hear them?

Just thought.

 

[Steve Woodhouse]

You are transferring an opinion on a forum . The only known fact for the human hearing is that we can hear frequencies in 20hz-20khz and feel sub bass even beneath 20hz if it is produced at sufficient SPL levels . The same is said for frequencies above 20khz . Because you simply transfer an opinion and you do not have a deep understanding of these claims you cannot hold a conversation . Go educate yourself first and then come back with structured and to the point arguments

It’s not an opinion. It’s been proven, as we’ve discussed.

I think the discussion this time round has been polite and good-natured. Please don’t spoil it with language like ‘go educate yourself’.

 

[steadyshot]

The speakers will reproduce the amplified signal that came from the dac that was fed by a source all through cabling signal and power . Even if the thd of speakers is generally higher than the electronics they will play more clear if they receive a cleaner signal .

 

[Steve Woodhouse]

So I have a solution, I think.

When someone calls a DAC (or similar digital device) ‘transparent’ we all know what that means. It means that the measurable, audible aspects of the device’s output are below the threshold of human hearing, as has been demonstrated in controlled testing.

I think we all understand and largely agree with that in as far as it goes.

Now the counter argument appears to be that different devices may interact to give different results.

Okay. Nothing we can do about that, as we can’t measure every possible combination.

So let’s just accept the term as described. If someone says a unit ‘measures as transparent’, we all know what that means.

It may matter to some more than others. If it matters to you, threat, that’s the information you wanted. If it doesn’t matter to you, you can feel free to ignore it.

No need to re-open the debate every time the phrase is used.

 

[steadyshot]

Nothing has been proven regarding the specific thresholds for electronics . They are just suggestions grouped as relaxed and strict . No proofs

 

[Musician]

Again , is there ne5532 in the analog part of the ultra - it might explain this :

CMMR for ne5532 ( 100 dB ) is not great .

 

[Jls]

Nothing has been proven regarding the specific thresholds for electronics . They are just suggestions grouped as relaxed and strict . No proofs

Science does not include the concept of proof. Just numerically defined degrees of confidence.

I’m confident that if two devices reveal no difference in well controlled double blind tests, I will not hear any difference in the casual listening I will do. And I will not pay extra for things that make no difference to me.

 

[onlyoneme]

The limits of human hearing are well documented (if you don’t know what they are, I’m not doing the legwork for you).

If you believe that perhaps they will differ in certain circumstances, the onus is on you to provide evidence.

I am more than happy to accept that the science is wrong, or flawed, or incomplete, but the starting point would have to be properly controlled testing.

Put another way, the scientists have provided evidence about what can and can’t be heard. If you believe otherwise, the least you can do is point to an independent test showing that a statistically significant number of people can hear the difference, even if you can’t explain what it is or how it works.

As I’ve said, this is the same scientific method used to test your medicines and aircraft.

This is basically a copy of your previous statements. There is no explanation here why a single (or subset) test which proves a "transparency" under the test conditions might work as a proof of "transparency" of the device in another conditions.

A quick example is here https://forum.wiimhome.com/threads/advantages-of-using-fixed-resolution.3991/post-63522

Two devices may behave the same in the test when sine 1 kHz 0 dBFS signal is used with fixed resolution which forces a resampling, providing "transparent" results. So we call them transparent.

Then we use slightly different test signal with intersample clipping at +3 dBFS and 90% volume limit to provide a headroom which could prevent the clipping. The result may look as shown on the last graph of the example above, if the headroom is applied after the conversion. The result is a massive, potentially audible clipping. Or it may look similar to the one below when the headroom is applied before the conversion:




Two "transparent" devices but audibly different due to the behavior which was omitted during the "transparency" test.

 

[steadyshot]

Science does not include the concept of proof. Just numerically defined degrees of confidence.

I’m confident that if two devices reveal no difference in well controlled double blind tests, I will not hear any difference in the casual listening I will do. And I will not pay extra for things that make no difference to me.

Because you don’t care . You do not want to build a SOTA system . Science is based on mathematics and for something to be concrete there has to be a formal proof. The next safety is probabilities. So I am not making any claims on audibility I want to check how technically sound a device is and then I will have more probabilities that I get the SOTA performance . Where will you find well controlled blind tests that are remotely verifiable ? What you say is neither scientific nor practical.

I don’t care that you are a casual listener .

 

[canard]

0

Again , is there ne5532 in the analog part of the ultra - it might explain this :

CMMR for ne5532 ( 100 dB ) is not great .

You do a little false all I'm afraid .. a "plus" and its AKM will be around '113-114db' in SINAD 1K with its 5534 output (for -115db it seems ultra). it is first due to the limits of these chip akm and ess ...;-)

in the case of the "plus" these 5534 (which has equipped most of the pro machines from the 1980s until very very recamation..) we can imagine that for a cost story....
(will become sensitive by cascading these output AOPs....otherwise on just a 5534 facing a 1611...you risk being disappointed by the differences observed on just a sinad 1k )
;-)

 

[Steve Woodhouse]

Nothing has been proven regarding the specific thresholds for electronics . They are just suggestions grouped as relaxed and strict . No proofs

That’s a broad statement.

Can you be specific?

We don’t listen to electronics, we listen to the signal that comes out at the end of a chain.

For example, we know the limit of human audibility to dynamic range is c.115dB. If you have a DAC with 115dB or better, then there you have it.

It really doesn’t matter the method the electronic signal has taken to get there. That’s like saying that potatoes boiled in water in a saucepan heated by an electric hub will taste different to those where the pan is heated by gas.

It’s the end result that’s important.

 

[onlyoneme]

When someone calls a DAC (or similar digital device) ‘transparent’ we all know what that means. It means that the measurable, audible aspects of the device’s output are below the threshold of human hearing, as has been demonstrated in controlled testing.

Which ones? Which "measurable, audible aspects of the device’s output"? Can you provide a list of tests and measurements which are necessary to be made to work as a proof of the device transparency?

 

[Jls]

Why didn't explain that anyone too my maths teacher?

Math has well defined concepts of proof.

I differentiate science from math, because science deals in imperfect observations.

 

[Steve Woodhouse]

This is basically a copy of your previous statements. There is no explanation here why a single (or subset) test which proves a "transparency" under the test conditions might work as a proof of "transparency" of the device in another conditions.

A quick example is here https://forum.wiimhome.com/threads/advantages-of-using-fixed-resolution.3991/post-63522

Two devices may behave the same in the test when sine 1 kHz 0 dBFS signal is used with fixed resolution which forces a resampling, providing "transparent" results. So we call them transparent.

Then we use slightly different test signal with intersample clipping at +3 dBFS and 90% volume limit to provide a headroom which could prevent the clipping. The result may look as shown on the last graph of the example above, if the headroom is applied after the conversion. The result is a massive, potentially audible clipping. Or it may look similar to the one below when the headroom is applied before the conversion:

View attachment 9090


Two "transparent" devices but audibly different due to the behavior which was omitted during the "transparency" test.

So that’s how it’s used, not the device itself.

Most measurements are usually taken with multi tone as well as a single sine wave.

But the bottom line is this. If something does sound different, that’ll show up in a properly controlled test.

Once again, you’re confusing “if it can be heard it can be measured” with “if a difference can be measured it can be heard”, which is a clear non sequitur.

 

[canard]

and so these problems with the ultra....?

 

[onlyoneme]

Math has well defined concepts of proof.

I differentiate science from math, because science deals in imperfect observations.

Math is a science but I guess you mean a kind of science called "natural science" here.