Center Channel DIY build

[Rudy81]

20ms.

[Rudy81]

30ms.

[mikebse2a3]

Thank's Rudy for taking the time to post these plots.

Can you change the Time axis of your plot from (0-300ms) to something like (0-50ms) for higher resolution data display in the time domain when you use shorter gate time measurements?

I really enjoy the details you post of your projects Rudy...!

mike tn

[Rudy81]

Miketn, I have shortened the window for higher res. I'm curious as to what you are looking for. I will say that the 5.0 version of REW is just phenomenal in what it can do....even for us amateurs.

[Rudy81]

10ms waterfall with a 50ms window

[Rudy81]

15ms, 50ms window

[Rudy81]

20ms waterfall, 50ms window

[Rudy81]

30ms, 50mswindow. This one seems to show how much more effective my HF absorbtion is relative to my LF bass traps in the room. Of course, this makes sense since it is much more difficult to tame LF energy in an enclosed space. Interesting view.

[mikebse2a3]

Rudy, I thought any one interested in measurements might find this interesting in showing that when it comes to Waterfall Plots choosing the test parameters and display parameters needs to vary depending on the data information your looking for otherwise wrong conclusions can be formed due to the test parameters choosen and displayed.

This is a good example of the same test data and how it can look dramatically different depending on the graph display parameters. Below is the same test gate of 30ms compared but the view is much different for each plot.

mike tn

[mikebse2a3]

Here is a comparison of the 50ms display at 5ms, 10ms, 20ms, 30ms test gate times: (bottom left is the 5ms and going clockwise from there).

What you can see from this series of plots is how increasing the gate sampling time shows an increase in levels and peak/valley ridges especially visable at the lower frequencies. By increasing the test gate times more energy from more reflections is being allowed in the measurement and displayed with the resulting constructive and destructive effects at the microphone are displayed. One thing this demonstrates is how especially at lower frequencies in typical home listening rooms one needs to choose gate times, frequency bandwidth and display time for usefull information.

Taking the same measurement and varying one parameter(test parameters or display parameters) at a time is a very good way to get a better picture of how to use waterfall plots to help determine what is happening with a loudspeaker or loudspeaker/room/listener interface.

mike tn

[Rudy81]

Good points, and you are correct that different parameters can help to see a clearer picture. Another instance when parameters are useful is when looking at a basic frequency response curve. I originally used the smoothing curves to make the plots more readable. I then learned that in effect, I am modifying the data by using the smoothing curves and in some cases defeating the purpose of the test. I have since learned to work with the impulse response window to keep the data as captured, and yet be able to see the response more clearly.

[mikebse2a3]

I'm curious as to what you are looking for. I will say that the 5.0 version of REW is just phenomenal in what it can do....even for us amateurs.

I was interested in the capabilities of your test program and it is very impressive.

I was also interested in maybe seeing the ITD gap (initial time delay) of your system/room setup and when your first room reflections begin to come into the picture (at least in respect to the center channel).

If you change your graph time display to approximately 0-10ms or 0-20ms and use a test gate time of about 5ms (this will display usefull info down to about your chosen 300Hz point) we should see the direct sound from the loudspeaker and when the room reflections begin to enter the picture. This is assuming no early reflections around the microphone or loudspeaker returning in less than 5ms of course.

mike tn

[mikebse2a3]

Good points, and you are correct that different parameters can help to see a clearer picture. Another instance when parameters are useful is when looking at a basic frequency response curve. I originally used the smoothing curves to make the plots more readable. I then learned that in effect, I am modifying the data by using the smoothing curves and in some cases defeating the purpose of the test. I have since learned to work with the impulse response window to keep the data as captured, and yet be able to see the response more clearly.

Rudy, as you know I believe in measuring all that is possible and using that as a guide when I'm listening and evaluating what ever I'm experimenting with.

One of the largest challenges is learning the corelation between measurements and my perceptions. The ear/brain has a great capability to hear through and adapt to conditions that when measured can look so bad to the eye.

This is why I like the following quote: Einstein: "Not everything that can be counted counts, and not everything that counts can be counted" this isn't an excuse not to measure but a caution to use our measurement properly and in the context that they are valid.

mike tn

[Rudy81]

I primarily use measurements to ensure I'm in the ballpark with whatever I am trying to do. In this case, I wanted to measure the actual capability of the center channel vs. the theoretical performance based on the T-S parameters. I did the same thing when I built JC's DBBs. It is nice to see the results relatively close to the predicted performance. It is also a good way to ensure there is not some problem that otherwise may go undetected, as was the case with the wiring issue I mentioned earlier.

I also use measurements to help me since I don't consider my ear to be very 'educated' and critical. The measuring helps me validate what I am hearing or at least know if I'm in the ballpark.

I also want to add that to 'study' the effect of the room, a better plot is the IR, impulse response, graph.