What is 'incompetent digital' ?

[March Audio]

Fair enough, looks as though I did miss your point.



I must confess to being rather skeptical myself - they look to be variations on FFTs which don't seem to suit measurements of noise, at least in my understanding. Added to that the ADC would need to be considerably better than the DAC being tested, or at least produce artifacts orthogonal to these being examined in the DAC. My own hunches lead me to consider that some kind of analog signal processing (passive filtering maybe) might be required prior to digitization for the dynamic range of the ADC not to be the bottleneck in measurement. Certainly this is already the case in very low level THD+N measurements of amplifiers.

mmm.... seen this before somewhere......

The point you are trying to make is that high level signals going into an ADC produce distortion in the ADC so you cannot be sure if the distortion / noise is in the source (DUT) or the measurement system. Hence -60 dB is usually used. So if you want to look at the noise floor in the presence of a high level signals its better to notch out the primary signal before it hits the ADC. Obviously you need to be very careful how you go about performing that filtering.

 

[March Audio]

Perceived loudness and softness, yes. In terms of listening it also includes the ability to surprise the listener with suddenness in contrasts. -'jump factor'.

OK, so what is your definition of noise floor modulation?

Also what makes you draw the leap to noise floor modulation being the cause of a reduction in perceived dynamics?

 

[Opus111]

mmm.... seen this before somewhere......

So am I channeling someone? Where have you seen it?

The point you are trying to make is that high level signals going into an ADC produce distortion in the ADC so you cannot be sure if the distortion / noise is in the source (DUT) or the measurement system. Hence -60 dB is usually used. So if you want to look at the noise floor in the presence of a high level signals its better to notch out the primary signal before it hits the ADC. Obviously you need to be very careful how you go about performing that filtering.

Not just 'high level' signals, all signals are going to distort to some degree. I said nothing about '-60dB' being used. Yes to relieve the dynamic range pressure on the measurement device there's the need for some analog pre-conditioning. That's still the case with the AP as far as I'm aware - the ADC alone is way too poor to make accurate THD measurements without the notch filter in place ahead of digitization.

 

[Opus111]

OK, so what is your definition of noise floor modulation?

Any signal correlated shifts in the system's noise.

Also what makes you draw the leap to noise floor modulation being the cause of a reduction in perceived dynamics?

As far as I'm aware there's no leap here - noise floor modulation is the sign there's non-linearity in the system. When music is playing there won't be harmonic distortion of any significance (too many very small tones) but there will be IMD whose error power will be noise-like. Hence as the level increases so will this noise-like error term. The subjective effect of having varying background noise strongly correlated with the signal is a reduction in dynamics.

 

[March Audio]

So am I channeling someone? Where have you seen it?



Not just 'high level' signals, all signals are going to distort to some degree. I said nothing about '-60dB' being used. Yes to relieve the dynamic range pressure on the measurement device there's the need for some analog pre-conditioning. That's still the case with the AP as far as I'm aware - the ADC alone is way too poor to make accurate THD measurements without the notch filter in place ahead of digitization.

Sorry thats not true, very low distortion ADCs exist but they all will have input buffers. I know you didnt say anything about -60dB, it is just normal practice and a notch filter is not required.

 

[March Audio]

Any signal correlated shifts in the system's noise.



As far as I'm aware there's no leap here - noise floor modulation is the sign there's non-linearity in the system. When music is playing there won't be harmonic distortion of any significance (too many very small tones) but there will be IMD whose error power will be noise-like. Hence as the level increases so will this noise-like error term. The subjective effect of having varying background noise strongly correlated with the signal is a reduction in dynamics.

Does it have to be correlated?

I just wanted a clearer definition of what you think it is. So harmonic distortion fits into this? Intermodulation distortion? I just wanted to find out if you are referring to something specific in nature.

What level do you consider audible?

 

[Opus111]

Sorry thats not true

Which part isn't true?

very low distortion ADCs exist but they all will have input buffers.

How low distortion are they? And with what measurements? THD+N doesn't cut it here if you recall.

I know you didnt say anything about -60dB, it is just normal practice and a notch filter is not required.

Normal practice for what?

 

[Opus111]

Does it have to be correlated?

Yes I believe so to be an issue in audio. Of course there must exist uncorrelated noise floor modulation but seems to me the correlated type is the problem for SQ.

What level do you consider audible?

I don't know yet as there aren't any measurements that give rise to a number which might be interpreted as 'a level'.

 

[March Audio]

Which part isn't true?



How low distortion are they? And with what measurements? THD+N doesn't cut it here if you recall.



Normal practice for what?

This is untrue :
the ADC alone is way too poor to make accurate THD measurements without the notch filter in place ahead of digitization.

Well I can measure distortion down to -160 dB. How low do you think you need to go?

 

[March Audio]

Yes I believe so to be an issue in audio. Of course there must exist uncorrelated noise floor modulation but seems to me the correlated type is the problem for SQ.



I don't know yet as there aren't any measurements that give rise to a number which might be interpreted as 'a level'.

So uncorrelated spuria are of no consequence?

OK, how far does the noise floor need to rise to be a problem? Is something below -130 dB as an arbitrary example, audible?

 

[Opus111]

This is untrue :
the ADC alone is way too poor to make accurate THD measurements without the notch filter in place ahead of digitization.

Do you have an AP? If so you can do the experiment to verify what I've said is indeed correct.

Well I can measure distortion down to -160 dB. How low do you think you need to go?

How accurately can you measure it and what's the method you're using? If you feed the output of your DUT directly into your ADC its fairly certain your ADC's distortion performance is going to be a LOT worse than -160dB. But by all means feel free to show me the error of my ways. Show mind, not just claim.

 

[Opus111]

So uncorrelated spuria are of no consequence?

It would depend on their level. Do you have some level in mind?

OK, how far does the noise floor need to rise to be a problem? Is something below -130 dB as an arbitrary example, audible?

Like I've already said, I don't know.

 

[March Audio]

I dont have an Audio Precision. However what I use is very low noise and distortion, better than most DACS.

Here are some measurements of the MDAC you have been listening to.

 

[Opus111]

Looks to be with a very low level stimulus signal, say -60 to -70dB. Now repeat at 0dB stimulus and re-post. Incidentally to be able to see noise floor modulation by comparing plots means knowing the effective bandwidth of each FFT bin for each plot. Care to post that up too?

 

[March Audio]

It would depend on their level. Do you have some level in mind?



Like I've already said, I don't know.

Well you have made the statement regarding noise floor modulation affecting perceived dynamic range. To take that further you need to establish what sort of levels are an issue. You postulated that its an issue, so you must be making the assertion that current DAC noise floors (modulated) arent low enough. This may be the case, I dont know, but you need to put some boundaries around your statement.

 

[Opus111]

You postulated that its an issue, so you must be making the assertion that current DAC noise floors (modulated) arent low enough.

I hear some DACs sounding more dynamic than others, but your reasoning is flawed here. A DAC (I mean a box, not a chip) is just part of a system which also may include a pre and will include an amp. Its entirely possible for DACs to be incontinent and put out signals into the ultrasonic range. Those ultrasonics aren't a modulated noise floor in the DAC but could quite likely be so after another non-linear stage like a pre and/or amp.

 

[March Audio]

Looks to be with a very low level stimulus signal, say -60 to -70dB. Now repeat at 0dB stimulus and re-post. Incidentally to be able to see noise floor modulation by comparing plots means knowing the effective bandwidth of each FFT bin for each plot. Care to post that up too?

Of course it will be higher - please refer to my initial post on this.

If you want to limit this to high level signals instead of the far more generalised statement OK, but it wont change the fact that the ADC is lower noise and distortion than the DAC.

It would have been a 131072 point FFT, but I cant remember if those particular plots were 48 or 96kHz

 

[March Audio]

I hear some DACs sounding more dynamic than others, but your reasoning is flawed here. A DAC (I mean a box, not a chip) is just part of a system which also may include a pre and will include an amp. Its entirely possible for DACs to be incontinent and put out signals into the ultrasonic range. Those ultrasonics aren't a modulated noise floor in the DAC but could quite likely be so after another non-linear stage like a pre and/or amp.

No my reasoning isnt flawed, you are now moving the goalposts to include other "distortions".

Just for info here is the track you sent me looking up to 48kHz with two different MDAC filters

 

[Opus111]

If you want to limit this to high level signals instead of the far more generalised statement OK, but it wont change the fact that the ADC is lower noise and distortion than the DAC.

The statement you've claimed is untrue is the focus of the current discussion. Seems you've misunderstood what I was claiming in your assertion of its untruth. Are you going to retract your erroneous claim?

 

[Opus111]

No my reasoning isnt flawed, you are now moving the goalposts to include other "distortions".

I don't see any goalposts they must be a figment of your imagination.