Home Theater Forum and Systems banner

Depression starting 2khz, not stereo related

1183 Views 7 Replies 3 Participants Last post by  jtalden
Here are the 1/6-octave smoothed measurements using a calibrated UMIK-1 of some very good speakers whose anechoic results are flat to 20khz as measured by Stereophile. Each measurement was made 2m from the speaker, on-axis. I'm puzzled as to why there's a depression of 2-3db that starts around 2khz.

The second pic shows the result of setting the IR Right window to 10ms. It stills depresses around 2khz, though a little less severe.

I'm not sure if this is due to my room acoustics, speakers, or measurement technique. Is this a common issue?


See less See more
1 - 8 of 8 Posts
mdat attached


> The Room RT60 is very high. Not the best situation for this type of measurement.
> The distortion of the measurement is very high.
> The mic distance is too far unless the speaker is elevated very high in a very high ceiling room.
> The window is too long for the mic position assuming a typically room size and normal speaker height if you used the default Turkey 0.25 filter.

More to your question:
The minor sag above 2k may be due to the elevation of the mic vs the speaker. Some measure on the TW axis and some on the bisector of the TW and MW. You will find significant SPL response differences around the XO for different elevations. It is unusual to measure on the MW axis, but I suspect a manufacturer picks the most advantageous situation.

Assuming your TW is about 36" above floor, I would suggest you move the mic to about 40" from the TW and position it on axis. Window with the left window set to 0.4 ms Turkey 0.25 and the right window set to 10 ms blackman-Harris 4. With these setting the values above about 500 Hz should be reasonably accurate.

Better yet measure outside and try different mic elevations until you get the expected result. Note that from my memory Stereophile may have taken an average of 5 measurements (on axis and 15° up, down, right and left). You will have to investigate and do the same to expect to get the same result.

In the end, the minor anechoic SPL variation is insignificant compared to the room impact on the response.

Before you do any of the above first investigate and correct the reasons for the high distortion. I don't see a soundcard cal file loaded so you will want to start there to assure that the measurement system is working properly.
See less See more
The spikes in the distortion plot are due to 1 kHz harmonics of the USB frame rate, a problem affecting a batch of mics including the one you have. It can be fixed by adding a decoupling capacitor, or MiniDSP will replace the mic, see MiniDSP UMIK-1 Update thread. This thread covers the mod to the mic.
Oh, I overlooked that this was a USB mic. That is why there is no soundcard cal. :duh: Please ignore my final sentence regarding calibration.
Before you do any of the above first investigate and correct the reasons for the high distortion.
How might I go about doing this? Is it due to the mic, or room?

Will try re-measuring on TW axis. Unfortunately there's not much I can do about the RT60. It's an irregularly shaped room.

Thanks for the very helpful info.

Edit: Just saw JohnM's post regarding the mic problem.
Repeating the measurement on tweeter axis didn't change the graph. The tweeters are a relatively high 48" off the floor for this speaker.

I've tried 2 other mics with the same results, so I'm guessing it isn't due to the mic.
That's correct.
JohnM was just pointing out that the unusually high distortion that I noticed in your measurement was caused by the mic and the options that were available if you wanted to correct this. My observation and his response was not intended to explain your SPL response question.

To recreate an SPL response that someone else created very closely, you will have to recreate the same test conditions very closely. This is not easy to do as there is a lot more than just mic distance and height to consider and a full and accurate description of a test condition is rarely available in enough detail.
1 - 8 of 8 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.