Ohm's acoustic law and stuff

[Graham C]

There are two issues when it comes to phase:

firstly, eletronic components are minimum phase devices which means that the frequency response tells all. [The phase response can be calculated from the frequency response using the hilbert transform..]

Secondly, there's Ohm's law, not V=IR, the other one, which which states only the power spectrum and not the relative phases of the components determines a sounds perceived character. After 160 years of reasearch, there are only a handful of test signals (not real music) which when heard in an anechoic chamber (not a real room), where is it possible to hear phase.

If it wasn't for this second law, a large number of loudspeakers simply would not exist, because they would not work, as only the first order Butterworth crossover filter is minimum phase...

I reckon you are trying to say we shouldnt lose sleep over phase, which I would agree with.

Reactive and resistive electrical components are minimum phase. Electronic systems may well not be, over the audible range.

Is this really another 'law' of Ohms?? The guy went to great effort to find a proof for V=IR..I can't see why he should consider something as subjective as hearing in a 'law' - or am I missing something??

 

[I-S]

A conceivable "null test" for a DAC would be to have another source generating an analog output (the reference) which is then put through an 16/44.1 ADC, the digital output of which is fed to the DAC. Then you can compare the output of the DAC with the analog input going into the ADC (after adding the appropriate delay) but then you are of course including the ADC and the delay line in the null test. Assuming you can ignore the effects of the ADC and delay line...

And now you're beginning to assemble an audio test suite. The most widely used world-wide is the Audio Precision System Two:

That's the model we have at work, the SYS-2722A. Doing THD tests and so on on a digital-input amplifier is not the easiest thing in the world...

 

[michaelab]

The fact is you seem to consider THD of 2% in a Dac inconsequential but an unmitigated disaster in an amplifier. Doesn't make sense to me.

Once again, you are using your habitual tactic of putting words into the mouths of people you disagree with . I never said anything remotely like the above, so I see no need to defend something I didn't say

Michael.

 

[oedipus]

Can you shed some knowledge on why the dCS Delius sounds distincly more natural than the inbuilt Dac when EQ'ed using the RCS2.2X to offer the same frequency response? Those results don't seem to tally with your assertion.

What did you do? EQ the room once using the TACT DAC and again using the dCS DAC? [Rather than EQ once, and then swap DAC's]. The frequency response your talking about is the measured room response right? [You didn't try using the Tact to EQ the DAC's to sound the same???]

The explanations on the table are:

1) amplitude/volume differences in the test

2) The DAC's measure differently.

3) If you re-eq'd the room in the process, then errors in that process (the impulse "click" has poor noise immunity, which I've mentioned before), placement of the mic. not shutting doors etc.. could led to significantly different room measurements.

1 & 3 are the most likely culprits.

 

[wadia-miester]

Fair Play Andy, you love to get every one going, still I find you amusing and your 0.1db dips, besides if after 160 years all you humans have found is that all dac's sound the same, what a wasted existance, and there was me thinking you were on audio precisions pay roll?

 

[oedipus]

Reactive and resistive electrical components are minimum phase. Electronic systems may well not be, over the audible range.

Amplifiers and CD players are minimum phase, you'd have to be grossly incompetant to inadvertantly incorporate an all pass filter in those devices.

Is this really another 'law' of Ohms??

Yes.

am I missing something??

Yes, he (G.S. Ohm) worked on acoustics too

 

[I-S]

Depends on the amplifier... A class D amplifier will have a low-pass filter on the output that passes the entire audio range but rolls off soon after, which can easily (and does) have phase implications in the audio band.

 

[wadia-miester]

Ah phase coherence, a yummy subject, so Andy do enlighten us as to 'stateside master of blind test terror's' rules for calculating such time anomalies when appiled to say a dac?, or better still an amplifiers output stage?
Iaasc, you really must get out more, your'll grow probes instead of fingers

 

[merlin]

Oedipus, I did it all! First EQ using the Tact Dac. Then insert the dCS. Feed both signals into a Restek Challenger. Level matching using an SPL meter and adjusting the dCS to unity gain.Switch between the two. Big difference.

Then re - eq after new measurement with dCS.Again a difference but possibly less pronounced. Overlay plots from both MSR's. Mic location identical for both readings. No visible difference.

This would tend to rule out 1 & 3 IMO, but one can always fall back on the imagination theory I suppose

One thing that was enlightening was the affect of mains cables being added to the system. One new measurement and correction filter applied and the magical effects of the cables more or less disappeared. I remember mentioning it at the time but it was greeted with derision of course.

Michael, why not get that new Dac measured? You don't want something that's clearly faulty surely, even if your ears are deceiving you

 

[stickman]

You could have kept the Benchmark and used a simple tone control to get the effect of the wadia for a lot less money..

Nah, I wanted remote control.

 

[oedipus]

Depends on the amplifier... A class D amplifier will have a low-pass filter on the output that passes the entire audio range but rolls off soon after, which can easily (and does) have phase implications in the audio band.

Well then, I'd worry more about the effect of the filter on the flatness of the frequency response for the reason below. [Maybe your amp should do over-sampling to simplify the filter design - sorry, I couldn't resist..]

Ah phase coherence, a yummy subject, so Andy do enlighten us as to 'stateside master of blind test terror's' rules for calculating such time anomalies when appiled to say a dac?, or better still an amplifiers output stage?

You're merely asking me to repeat myself

In a minimum phase system, if you have a flat frequency reponse, there won't be any time anomalies...

On the other hand, if I were dealing with a non-minimum phase system (and loudspeakers with multiple drivers fall into this category), I would aim for a flatter frequency response at the expense of the group delay rather than vice versa. This is because the ear is sensitive to ampitude response, but not phase relationships, so it makes sense to optimize for frequency response.

 

[oedipus]

Level matching using an SPL meter

You won't get the required accuracy going that route. When I did my comparisons, I used only 'phones - it was much easier with that arrangement to ensure that the DAC's were precisely matched..

 

[MartinC]

Sorry, missed this earlier.

firstly, eletronic components are minimum phase devices which means that the frequency response tells all. [The phase response can be calculated from the frequency response using the hilbert transform..]

Is this an approximation for systems which have a roughly linear phase response?

Secondly, there's Ohm's law, not V=IR, the other one, which which states only the power spectrum and not the relative phases of the components determines a sounds perceived character. After 160 years of reasearch, there are only a handful of test signals (not real music) which when heard in an anechoic chamber (not a real room), where is it possible to hear phase.

If it wasn't for this second law, a large number of loudspeakers simply would not exist, because they would not work, as only the first order Butterworth crossover filter is minimum phase...

This is very interesting - but just leads to lots of questions. Quite what do you mean by "determines a sounds perceived character"? Are you trying to seperate out spatial from tonal information possibly?

Does this mean for instance that you think time alignment of drivers is a complete waste of time?

 

[oedipus]

Is this an approximation for systems which have a roughly linear phase response?

No. Minimum phase and linear phase are different concepts.

This is very interesting - but just leads to lots of questions. Quite what do you mean by "determines a sounds perceived character"?

That the ear cannot determine if you have fiddled with the relative phases of the components of a signal.

Are you trying to seperate out spatial from tonal information possibly?

No. This observation is only on the sensation of tone and has nothing to do with spatial information.

Does this mean for instance that you think time alignment of drivers is a complete waste of time?

The baffle on my (ATC) loudspeakers is flat

 

[MartinC]

No. Minimum phase and linear phase are different concepts.

Cheers. Had a feeling that might have been the case, I'll have to look into this sometime.

Are you trying to seperate out spatial from tonal information possibly?

No. This observation is only on the sensation of tone and has nothing to do with spatial information.

That's what I meant . Since low frequency spatial information comes from differences in arrival time of signals (ITD), I'd have thought differences in relative phasing of frequencies could have an effect on spatial info?

Edited to add: My point regarding ITD being that spatial info depends on arrival times, and these will be different (to those recorded) at different frequencies due to phase errors. Not that if you had a different phase response for the two channels that this would cause problems, as I susect it would. I'm still just thinking if there's any effect of a non-linear phase response, the same on both channels.

Does this mean for instance that you think time alignment of drivers is a complete waste of time?

The baffle on my (ATC) loudspeakers is flat

That doesn't answer my question now does it . My speakers have a flat front baffle, but I hadn't assumed time alignment was pointless.

 

[oedipus]

Please do explain linear and minimum phase variants Andy, they are a few keen guys here wanting to indulge in acoustical engineering ...
Instead of one sentance/word retorts Andy, how about some theroum/fumulii?/graphs/plots, l

Minimum phase is a concept from electrical engineering. You will find all the theory, graphs and formulii in Sedra and Brackett: "Filter Theory and Design: Active and Passive" (and other good filter/signal processing books).

Since low frequency spatial information comes from differences in arrival time of signals

The ear/brain does localization based on higher frequencies where sound diffracts around the head in different ways causing a different amplitude response in either ear, this difference in reponse is the key. At low frequencies this mechanism doesn't work, which is why subwoofers do [The answer to your next question is harmonics..]

Does this mean for instance that you think time alignment of drivers is a complete waste of time?

Yes, if you are doing it only to optimize the phase response.

 

[MartinC]

The ear/brain does localization based on higher frequencies where sound diffracts around the head in different ways causing a different amplitude response in either ear, this difference in reponse is the key. At low frequencies this mechanism doesn't work, which is why subwoofers do [The answer to your next question is harmonics..]

I know localising by IID doesn't work at low frequencies but I didn't mention that! My point was over the use of timing differences to localise low frequency sounds (ITD), where from a couple of quick calculations it appears that the time differences in question are shorter than the duration of a single cycle at the relevant frequencies.

(It may be you're going to suggest you don't believe ITD is used but rather that you localise on the basis of higher harmonics of such notes, but it would be nice to be clear where you're coming from. I haven't looked into this myself, but I've certainly assumed the evidence for the ITD localistion method was pretty strong.)

 

[MartinC]

Does this mean for instance that you think time alignment of drivers is a complete waste of time?

Yes, if you are doing it only to optimize the phase response.

Why else would you do it?

 

[oedipus]

<Discussion of Interaural Time Delay deleted>

What your talking about is the phase difference of a single frequency. If you think about it, you can shift that signal in time (ie phase shift it) and still have the same phase difference at the ear. So, if my CD player/amp/speaker phase shifts that frequency, it doesn't upset the phase difference. Wait, were not done...

Let me quote what I wrote a few pages back: "there's Ohm's (acoustic) law, which which states only the power spectrum and not the relative phases of the components determines a sounds perceived character."

So, what I'm talking about is the phase relationships of a collection of frequencies. Suppose we have 2 frequencies, and we mess with the phase of one (perhaps drastically changing the waveform), the ear doesn't correlate the phases across frequencies, and can't detect that change. The reason that this is an issue is because when you filter a signal, not only do you change the frequency reponse, you also change the phase relationships.

Why else would you do it (time alignment)?

To improve the amplitude response at "the listening position".

 

[PeteH]

Suppose we have 2 frequencies, and we mess with the phase of one (perhaps drastically changing the waveform), the ear doesn't correlate the phases across frequencies, and can't detect that change.

Well now, I must say that's very interesting - and it rather makes a nonsense of certain speaker manufacturers' claims of the great lengths they go to to ensure phase coherency.