Can someone explain this?

[Cornelis Huizinga]

Can someone explain this for me?
The attached screenshot shows 3 control measurements of Audiolense with REW. The first two are the left and right channels measured separately
The bottom one is a measurement of the 2 channels simultaneously. What is striking is the dip in the frequency characteristic around 10 kHz.
At first I thought it was caused by the placement of the speakers, but that turned out not to be the case. I have a completely different system in the attic, but there too the same effect occurred when both channels were measured at the same time. The dip in this case was at 8kHz. Maybe someone has an idea what could be going on here and what I can do about it?

 

[kevinzoe]

I'm guessing that it's wave cancellation which occurs due to short wave lengths at higher frequencies.
It's happened to me too.
Just for fun, try and recreate the lower curve line (both speakers firing simultaneously) and when you see the dip again at 10khz, try moving the mic to the left or right a couple of inches and note how the null shifts or disappears . . . If this in fact occurs then you know it's a cancellation which won't be noticed because your two ears are in different locations than the mic is when the null is detected.

 

[Cornelis Huizinga]

Hi Kevin, thanks for the tips. As soon as I have time this week I'll give it a try and let you know the results here.
I don't expect much from it. It's not a big deal either, but I was curious if others had this experience as well. You usually only see readings from individual channels.

 

[Omid]

Can someone explain this for me?
The attached screenshot shows 3 control measurements of Audiolense with REW. The first two are the left and right channels measured separately
The bottom one is a measurement of the 2 channels simultaneously. What is striking is the dip in the frequency characteristic around 10 kHz.
At first I thought it was caused by the placement of the speakers, but that turned out not to be the case. I have a completely different system in the attic, but there too the same effect occurred when both channels were measured at the same time. The dip in this case was at 8kHz. Maybe someone has an idea what could be going on here and what I can do about it?

You can also use REW to mathematically add R+L (using vector addition). That will show you if there is a phase cancellation. At 8kHz your wavelength is 4.25 cm so as Kevin said in real life (if it’s a phase cancellation thing) it’s not relevant.

 

[Cornelis Huizinga]

Hi Omid, as you suggested, I added the left and right channels in REW. It is indeed a phase cacelation thing. You can see the result in the attachment.
Thanks for thinking along.
Now let's see if the dip is reduced by moving the microphone, as suggested by Kevin. That takes a little more time.

 

[Cornelis Huizinga]

Omid and Kevin, you were right; it was indeed a non issue.
I couldn't even reproduce the dip

 

[jjazdk]

Cornelis, please be aware of the scaling on your y-axis, as a roughly 200dB axis will hide any important information.
A good ruel of thumb is to never have more than 60dB as the total scale on the y-axis.

 

[DanDan]

This is a test I did some time ago to investigate this phenomenon. It is difficult to get the measuring mic precisely central, both distance wise and acoustically, so combining, adding them, often shows cancellations at HF. My conclusion, takeaway, is that you simply cannot. I don't remember if REW has Vector Summing I think Fuzzmeasure/Rodetest has, but try some Trace Arithmetic. More simply put, I would use dual speaker drive for ONLY LF information. Trust the individual speaker measurements for HF. Oddly the Delay Compensations which Sonar and Dirac Live offer to perfectly align the time of arrival at the microphone result in a horrible phasey sound, particularly in highly treated direct field listening. Those Delays, however small, are best bypassed, i.e. off. Some systems, minidsp's hardware Dirac Live does not have bypass, which makes life difficult in critical situations. e.g.Pretty much anechoic Control Room listening.