Setting t=0

The same thing happens when using loopback. Previously, I only took measurements with the loopback. Then I assumed that with acoustic synchronization, the measurements would become stable. But it turned out to be almost the same. It happens that each new measurement occurs with a displacement, but sometimes it happens 5-7 in a row as it should be, that is, without displacement.
 
Have you tried the suggestions about optimizing the PC for sound applications?
I think I was having problem also prior to making some PC changes years ago.
Here is one site I just quickly found. I think the advice I found was at the Focusrite website. Possibly most audio interface companies offer some advice. I think turning off wifi was some help to me, but it may have been a combination of things. Do a search and try some of the ideas offered.

Found another:
Here
 
I just remembered that I discovered that moving the mouse quickly during a measurement causes problems for me. I don't do that any more.
 
I read the information on your link. I know about everything that is said there. Thank you. The only thing I disagree with is to turn on the maximum and minimum states of the processor to 100. At this value, the fan is making noise almost constantly. Almost at maximum speed. Therefore, it is necessary to set the maximum state of the processor to 95 or 90 or 85, depending on the hardware. Then the processor does not overheat, and the fan always runs below the average speed, or at the minimum.
In my case, if no one tells me how to solve my problem, I will try another audio interface. Or another amplifier.
 
So, you actually tried not moving your mouse during the measurements as well as all the other PC setup suggestions on the internet?

You now think the problem is with the amp or the audio interface. I don't see how that could be.
 
you actually tried not moving your mouse during the measurements
Sure. And I never tried. And all the recommendations were known to me.

You now think the problem is with the amp or the audio interface. I don't see how that could be.
When you seem to have checked everything and tried it, you start to believe in anything.
 
The impulse response is never changed, all that is being done is to pick a point along the time axis and define that point as zero time.
In the latest versions of REW, the impulse response changes from a change in position t = 0. An example in the screenshot. I made a copy of the measurement and shifted t = 0 on the copy. This is one question.

Second question, REW for defaults set t = 0 at the first maximum IR peak. This is correct if all drivers in the speaker are connected with positive polarity. And if the polarity is not inverted when measured at the input.

If the input polarity is reversed due to the microphone or microphone preamp, or the wires in the connectors are not properly soldered, then REW correctly sets t = 0 to the first maximum peak, 'negative'. The user must click the 'Invert' button to see the real picture.

But, there are some speakers in which several drivers are connected with reverse polarity. I have so. In the attached measurement and screenshot there is such a speaker. In this case, it is correct to set t = 0 on the first positive peak. Question.
Can REW detect this by default? If not, is it possible to add a button, the activation of which will tell REW that the speaker is being measured, for which t = 0 should be set to the first positive peak, and the second negative and bigger to be ignored?
 

Attachments

  • okt 5.mdat
    okt 5.mdat
    7.7 MB · Views: 40
  • t=0.PNG
    t=0.PNG
    65.1 KB · Views: 25
If the option to set t=0 at the IR peak is selected when not using a timing reference t=0 will be placed at the maximum magnitude. When a timing reference is used the position of t=0 depends on the time at which the reference is detected.

You can set t=0 wherever you like using the controls provided. The 'correct' position for t=0 can depend somewhat on the DAC reconstruction filter, but in the absence of acausal filtering the response should start at t=0.

To preserve the shape of the IR when t=0 is changed an IR oversampler selection is required on the Analysis preferences, usually Windowed sinc. That only affects the appearance, however, the sample points are in the same positions with or without an oversampler selection.
 
If the option to set t=0 at the IR peak is selected when not using a timing reference t=0 will be placed at the maximum magnitude. When a timing reference is used the position of t=0 depends on the time at which the reference is detected.
My measurement used acoustic synchronization.

To preserve the shape of the IR when t=0 is changed an IR oversampler selection is required on the Analysis preferences, usually Windowed sinc. That only affects the appearance, however, the sample points are in the same positions with or without an oversampler selection.
'IR oversampler selection' should be done before measuring? I tried to turn this parameter on and off in the measurement from my last post, in both cases the IR shape changes when I move t=0.
 
The option only applies to measurements when they are generated, whether by response copy or making a measurement.
 
In the measurement of post #57, 'IR oversampler' was set to 'None'. Therefore, the IR form changed with the movement t=0.
Thank you for your help in understanding how REW works.

How do you comment on what jtalden wrote about the end of the phase response? I agree with him.
Posted phase charts for speaker system responses seen here and elsewhere tend to most often have the phase flatten out near 0° for the highest measured frequencies.
 
John Mulcahy
You once gave a link to the site of the German guys who were testing microphones for linearity. With great statistics. Can you repeat the link?
 
In the measurement of post #57, 'IR oversampler' was set to 'None'. Therefore, the IR form changed with the movement t=0.
Thank you for your help in understanding how REW works.

How do you comment on what jtalden wrote about the end of the phase response? I agree with him.
I don't agree that phase should tend to zero. The phase response should end up as a horizontal line for a measurement that goes well beyond the bandwidth of the device being measured (more than a decade above its roll-off frequency), but in general that line would not be at zero degrees since most devices have a low pass response and the final phase will depend on the roll-off rate.
 
Back
Top