MQN testing/experimentation thread

[sbgk]

I disagree, using a fixed position microphone it will capture any differences if they are there to be measured. Im talking about taking measurements off axis. This might explain the wider or narrower sound stage that we experience

Probably the earlier suggestion to take the output directly from the analogue stage of the DAC is better. Otherwise you are measuring the interaction with the room which is complex.

yes, but as long as you are in a geological stable part of the world the room will stay the same between versions.

[DaveF]

I disagree, using a fixed position microphone it will capture any differences if they are there to be measured. Im talking about taking measurements off axis. This might explain the wider or narrower sound stage that we experience

Probably the earlier suggestion to take the output directly from the analogue stage of the DAC is better. Otherwise you are measuring the interaction with the room which is complex.

I'd agree that measuring the analogue output of the DAC is a good starting point. Is there any other circuitry though between there and the output stage? Some of that circuity could be somewhat bandlimited and have a filtering affect. Of course opening the bandwidth more will allow more noise through from my experience.

Do you propose to run it through an ADC to redigitize and then do an FFT of the result and look for differences between differnt MQN versions?

[LowOrbit]

I disagree, using a fixed position microphone it will capture any differences if they are there to be measured. Im talking about taking measurements off axis. This might explain the wider or narrower sound stage that we experience

Probably the earlier suggestion to take the output directly from the analogue stage of the DAC is better. Otherwise you are measuring the interaction with the room which is complex.

yes, but as long as you are in a geological stable part of the world the room will stay the same between versions.

:-)

However, if the tonal balance is different, the room interaction will be different. Room modes and reflections are very frequency/amplitude dependent.

[DaveF]

However, if the tonal balance is different, the room interaction will be different. Room modes and reflections are very frequency/amplitude dependent.

Any differences could possibly be absorbed or dampened to a large to degree by room furnishings before they reach the microphone. So they might not show up in the measurement. I think measuring the output of the DAC is a better way to go.
However, in my experience, directly measuring the output of a DAC with a probe can sometimes inject more noise or distortion onto the signal thus polluting it. (all very circuit dependent though)
The noise was so small anyway that the only way to see it was to run it through an amp stage or using the DACs to drive a laser, from there through a photodiode and the noise was much more visible then. But this involves specialist equipment and is perhaps outside the scope of this.

It depends on what we want to measure though:
1. Is it the belief that the noise floor or content changes?
2. Is the freq content of the DAC output different?

[LowOrbit]

Hi Dave

Based on recent posts in this thread I think I probably differ in my view of what changes we are trying to measure here. My feeling is that somehow (not got it entirely clear in my head) the sample alignment is changed between versions of MQn and between MQn and other bitperfect players. In versions of MQn which show improvements in resolution, I suspect the average alignment of samples is better. So frequency content is more accurately portrayed and distortion/smearing of low level information is reduced.

I am not convinced a software player would have significant impact on generated electrical noise within the PC. So in answer to your two questions above, a bit of both.

Mark

[jkeny]

We have to be careful in our ruminations here
Digital delays in samples does not necessarily translate into timing delays in analogue - it's not a direct relationship.
Another aspect is that better, more accurate rendition of the audio signal often results in a perceptually louder bass - I have noticed this with reducing jitter. My point being that louder bass, does not necessarily mean distortion - it may do but we have to be careful, does the bass have texture or is it more mono-note - I believe this allows us to differentiate distortion from better reproduction.

DaveF is probably correct, it may make more sense to measure at the amp output rather than DAC output but this implies that the amp doesn't add/remove or change the signal characteristics that we are interested in measuring. As difficult ask & one that may be difficult to ascertain. However, I guess if we hear differences with that amp then we are measuring the signal that is going to the speakers.

BTW, doing in-room measurements is really the wrong approach - too many variables - just two, speaker distortions, room interactions are enough to confound any measurement analysis

[LowOrbit]

We have to be careful in our ruminations here
Digital delays in samples does not necessarily translate into timing delays in analogue - it's not a direct relationship.
Another aspect is that better, more accurate rendition of the audio signal often results in a perceptually louder bass - I have noticed this with reducing jitter. My point being that louder bass, does not necessarily mean distortion - it may do but we have to be careful, does the bass have texture or is it more mono-note - I believe this allows us to differentiate distortion from better reproduction.

DaveF is probably correct, it may make more sense to measure at the amp output rather than DAC output but this implies that the amp doesn't add/remove or change the signal characteristics that we are interested in measuring. As difficult ask & one that may be difficult to ascertain. However, I guess if we hear differences with that amp then we are measuring the signal that is going to the speakers.

BTW, doing in-room measurements is really the wrong approach - too many variables - just two, speaker distortions, room interactions are enough to confound any measurement analysis

Agree, none of this is straightforward. Interestingly my perception of bass is generally it doesn't sound as loud the "better" it gets (times better, more tuneful, tonally more accurate). But it's hard to make firm rules as bass interacts in rooms much more than other bands.

[DaveF]

Digital delays in samples does not necessarily translate into timing delays in analogue - it's not a direct relationship.

yip I'd agree here that the samples can wobble around in time but wont necessarily translate to the analogue domain. But it would really depend on the phase relationship between when the sample data changes and when the next sampling edge of the clock occurs. It they were very close then variations in the sample timing could caouse the clock to miss one or resample the previous one. Ideally you want the clock to sample in the middle of the data period. Path delays through hardware can add skew etc but we're talking about digital running in the KHz range so its not really an issue. If it is, then the designer of that circuit ought to be taken outside and beaten over the head with a rake. Repeatedly.
I really cant see how MQN could be causing varations in the data sample timing down at the hardware level just before it gets sampled by the clock.

[jkeny]

yip I'd agree here that the samples can wobble around in time but wont necessarily translate to the analogue domain. But it would really depend on the phase relationship between when the sample data changes and when the next sampling edge of the clock occurs. It they were very close then variations in the sample timing could caouse the clock to miss one or resample the previous one. Ideally you want the clock to sample in the middle of the data period. Path delays through hardware can add skew etc but we're talking about digital running in the KHz range so its not really an issue. If it is, then the designer of that circuit ought to be taken outside and beaten over the head with a rake. Repeatedly.
I really cant see how MQN could be causing varations in the data sample timing down at the hardware level just before it gets sampled by the clock.

DaveF, I know it's difficult to correlate any mechanism of operation that results in timing delays in the analogue output from some different players but I have preliminary data that seems to show this, just not ready to release it yet as it needs lots of cross-correlation, cross-checking & further accumulation of data. It also is tenuous at this time in that I can't put my finger on the timing differences between different versions of MQN that I can hear are different or between MQN & some other players.

It's a test with more parameters than I would like & needs careful consideration & analysis.

[sbgk]

I had a theory that the noise is worse when there is loud, dynamic music with treble, does something get overloaded ? I have a version where this is audible and the interference gets worse during dynamic pieces. There is no distortion during quite periods and yet it is transferring the same no of bytes, so that means the value of the byte affects the sound 8 0 bits takes less to transfer than 8 non zero bits. can upload the distortion version if people are interested.